Emerging technologies for early risk stratification and precision management of diabetic kidney disease: a multimodal framework integrating digital phenotypes and clinical biomarkers

Background: Diabetic kidney disease (DKD) is a major microvascular complication of diabetes, often progressing silently and leading to end-stage kidney disease (ESKD) and cardiovascular morbidity. Early identification and risk-adapted intervention are crucial to improving long-term outcomes, yet existing clinical workflows are limited by delayed diagnosis and underutilization of available therapies.

Methods: We propose and evaluate a multimodal, risk-driven framework for the early recognition and individualized management of DKD. The approach integrates: (1) standard renal function metrics—estimated glomerular filtration rate (eGFR) and urine albumin-to-creatinine ratio (uACR)—together with validated prediction models; (2) molecular biomarkers including metabolomics, gut microbiota, and peritoneal dialysis effluent signatures; (3) digital phenotypes derived from standardized acquisition of tongue images and pulse waveforms, rooted in Traditional Chinese Medicine (TCM) diagnostics; and (4) longitudinal data from wearable devices and remote monitoring platforms. Digital features are quantified using image processing and optical signal analysis and incorporated into multimodal prediction models. Treatment is escalated based on risk stratification using renin–angiotensin–aldosterone system (RAAS) inhibitors, sodium–glucose cotransporter 2 (SGLT2) inhibitors, non-steroidal mineralocorticoid receptor antagonists (MRAs), and glucagon-like peptide-1 (GLP-1) receptor agonists. Real-time monitoring of therapeutic efficacy and safety is conducted using process end points such as eGFR slope and uACR trends.

Results: Incorporation of quantifiable tongue and pulse features provides a novel, low-cost, and non-invasive risk enrichment layer that complements biochemical and omics-based markers. Multilayered risk stratification enables earlier identification of fast progressors and more timely treatment intensification. Evidence from landmark trials—including Dapagliflozin and Prevention of Adverse Outcomes in Chronic Kidney Disease (DAPA-CKD), Empagliflozin in Patients with Chronic Kidney Disease (EMPA-KIDNEY), Finerenone in Reducing Kidney Failure and Disease Progression in Diabetic Kidney Disease (FIDELIO-DKD), and Effects of Semaglutide on Chronic Kidney Disease (FLOW)—supports the clinical utility of this approach. A closed-loop monitoring strategy based on process metrics and safety thresholds is proposed. We also outline ethical, regulatory, and data governance considerations necessary for clinical translation.

Conclusion: The integration of traditional clinical markers, digital TCM-derived phenotypes, and multi-omics data represents a promising paradigm for early, personalized, and dynamic DKD care. Future research should focus on external validation, impact on hard end points, and equitable deployment across real-world settings. This approach may help close the current diagnostic and therapeutic gaps in DKD management.

1 Background and disease burden

Diabetes has emerged as one of the most prevalent and fastest-growing chronic non-communicable diseases worldwide. As of 2024, an estimated 589 million adults aged 20 to 79 years—approximately one in every nine individuals—are living with diabetes, with around 3.4 million deaths attributed to diabetes each year (1). For the first time, global direct healthcare expenditures on diabetes have surpassed USD 1 trillion (2). By 2050, the number of individuals affected in this age group is projected to rise to 853 million. Concurrently, estimates from the Global Burden of Disease study suggest that more than 1.3 billion people of all ages will be living with diabetes by 2050, highlighting an ongoing global escalation in disease prevalence (3).

A substantial portion of diabetes-related mortality and economic burden is driven by its complications, including cardiovascular disease, kidney disease, neuropathy, and retinopathy (4). Among these, diabetic kidney disease (DKD) is particularly insidious—often progressing without symptoms yet offering a potential therapeutic window during which timely, evidence-based interventions can significantly delay renal deterioration (5).

Improving clinicians’ awareness and understanding of chronic kidney disease, particularly DKD, is essential to ensure the prompt initiation of such interventions. Evidence-based therapies—such as SGLT2 inhibitors, non-steroidal MRAs, and GLP-1 receptor agonists—can decelerate renal function decline, mitigate metabolic complications, and reduce cardiovascular risk (6–8). Prioritizing early detection and personalized management of DKD, therefore, constitutes a strategic opportunity for global public health efforts.

1.1 Rationale and use cases

Early risk stratification for DKD should prioritize adults with type 2 diabetes in primary care, endocrinology, and nephrology settings, including those with normal or mildly reduced kidney function (G1–G3a) and normo- to micro-albuminuria. Priority subgroups include long diabetes duration, hypertension or ASCVD, history of preeclampsia, rapid estimated glomerular filtration rate(eGFR) decline, family history of CKD, and marked glycemic or blood-pressure variability on continuous glucose monitoring (CGM)/ambulatory blood pressure monitoring (ABPM). The window of intervention is pre-albuminuria through early micro-albuminuria. Screening hinges on annual eGFR plus uACR (earlier/more frequent in high-risk profiles), followed by confirmation with repeat uACR (two of three positives over ~3 months), exclusion of transient causes (UTI, fever, strenuous exercise, menstruation), and standardized first-morning sampling.

Actions are then mapped to clinical triggers. Risk stratification combines the KDIGO eGFR×uACR heat map, eGFR slope, and—where available—risk equations (e.g., KFRE), with optional augmenters such as urine proteomic panels, MRI/SWE, and genetic/epigenetic scores; longitudinal CGM/ABPM metrics and research-grade tongue/pulse features can be layered for hypothesis generation. Treatment follows a stepwise pathway: RAAS blockade and SGLT2 inhibitors as foundations (if eligible), intensification with non-steroidal MRAs for persistent albuminuria, consideration of GLP-1RAs for weight and cardiovascular risk, and systematic optimization of BP, glycemia, lipids, and adherence/lifestyle. Monitoring cadence scales with risk (e.g., 12/6/3 months), tracking uACR trend, eGFR slope, BP/CGM metrics, and safety labs (serum potassium, eGFR), with re-stratification after therapy changes or acute events. This section anchors the subsequent framework that operationalizes these steps into deployable workflows. The decision nodes and corresponding actions are summarized in a pragmatic table for deployment in routine care (Table 1). To provide an overview, Figure 1 illustrates how traditional markers (eGFR, uACR, KFRE) are being complemented by emerging multimodal strategies—including omics, TCM-derived digital phenotyping, and machine-learning–based longitudinal modeling—to achieve earlier and more personalized intervention.

Table 1

Table 1. Trigger-to-action map for early DKD risk stratification and precision management.

Figure 1

Figure 1. From guideline-based screening to multimodal, risk-driven management in diabetic kidney disease. The left panel illustrates the current KDIGO 2024–aligned screening and staging pathway for patients with type 2 diabetes: annual or risk-based measurement of eGFR and uACR, confirmation of abnormal uACR on repeat testing, and risk categorization using the eGFR×uACR heat map and KFRE. The right panel depicts how emerging multimodal technologies can be layered on top of this foundation to refine early risk stratification and guide treatment intensification. Molecular and fluid biomarkers (e.g., TNFR1/2, KIM-1, CKD273, urinary extracellular vesicles), tissue-level imaging metrics (multiparametric renal MRI, shear wave elastography), and digital phenotypes (tongue and pulse features, CGM and ABPM-derived metrics) feed into longitudinal machine learning models that estimate short- and medium-term risk of kidney failure and cardio–renal events. These estimates, in turn, inform stepped intensification of RAAS blockade, SGLT2 inhibitors, non-steroidal MRAs, and GLP-1 receptor agonists, as well as follow-up frequency and nephrology referral. eGFR, estimated glomerular filtration rate; uACR, urine albumin-to-creatinine ratio; KFRE, Kidney Failure Risk Equation; CGM, continuous glucose monitoring; ABPM, ambulatory blood pressure monitoring; TCM, Traditional Chinese Medicine; MRI, magnetic resonance imaging; SWE, shear wave elastography; RAAS, renin–angiotensin–aldosterone system; MRA, mineralocorticoid receptor antagonist; GLP-1, glucagon-like peptide-1.

1.2 Literature search and selection

This narrative review was informed by a structured but non-exhaustive literature search focusing on emerging tools for early risk stratification and precision management of DKD. We conducted electronic searches in PubMed/MEDLINE, Embase, and Web of Science for English-language articles published between January 2014 and October 2025.

In PubMed, an initial broad query combined DKD terms with biomarker and digital-health concepts, for example:”diabetic kidney disease” OR “diabetic nephropathy” AND (“biomarker*” OR “proteomic*” OR “metabolomic*” OR “genetic*” OR “epigenetic*” OR “urinary extracellular vesicles” OR “imaging” OR “MRI” OR “shear wave elastography” OR “digital biomarker*” OR “mobile health” OR “wearable*” OR “machine learning” OR “artificial intelligence” OR “tongue diagnosis” OR “pulse wave” OR “Traditional Chinese Medicine”).This broad search yielded a large number of citations. We therefore further refined the strategy by restricting terms to title/abstract fields and by adding risk/prognostic concepts (e.g., risk, prediction, prognos, risk score, risk model, risk stratification, progression), and we applied filters for humans and English-language publications.

We prioritized original human studies and high-quality systematic reviews that evaluated prediction, early detection, or monitoring of DKD progression using clinically relevant endpoints such as eGFR slope, sustained ≥40% eGFR decline, incident kidney failure, or cardio–renal composite outcomes. Experimental work in cell and animal models, and purely methodological or technical papers without direct clinical application, were included selectively when they clarified mechanisms or key limitations of the technologies. Reference lists of key articles and recent guidelines (e.g., KDIGO 2024 CKD guideline) were screened to identify additional relevant publications. Given the heterogeneity of study designs and outcomes, we did not perform a formal meta-analysis; instead, we synthesized the evidence qualitatively, emphasizing study design, sample size, validation status, and readiness for clinical translation.

2 Rationale for prioritizing DKD in early diagnosis and intervention

DKD represents a strategically valuable entry point for early diagnosis and intervention among diabetes-related complications, owing to its considerable public health burden and modifiable disease trajectory (9). Diabetes is now the leading global cause of chronic kidney disease (CKD) and end-stage kidney disease (ESKD), and constitutes the most common indication for dialysis and kidney transplantation in many countries (10). Early upstream detection and interruption of DKD progression, therefore, have the potential to substantially reduce the societal burden of renal replacement therapy, premature mortality, and long-term disability (9).

In its initial stages, DKD typically progresses silently and without symptoms; however, a therapeutic window still exists during which the disease course can be delayed or even partially reversed through timely, evidence-based intervention (11). Recent clinical trial data have demonstrated that early and sufficiently intensive treatment in high-risk individuals can favorably alter long-term outcomes (6, 12). Notably, the FLOW trial confirmed that semaglutide, a GLP-1 receptor agonist, significantly reduced the incidence of composite kidney outcomes and cardiovascular mortality in patients with type 2 diabetes and CKD, thus reinforcing the value of early detection and intervention (12).

From an implementation standpoint, DKD lends itself to scalable and standardized risk stratification due to its streamlined, guideline-endorsed screening framework. The KDIGO 2024 guidelines recommend that all individuals with high-risk or suspected diabetes undergo simultaneous measurement of eGFR and uACR, with repeat testing advised to exclude acute fluctuations (5). Where resources permit, combined creatinine–cystatin C estimation (eGFRcr-cys) is encouraged to improve CKD staging precision and guide therapeutic decisions.

Subsequent risk categorization is based on an eGFR–uACR heatmap, complemented by validated risk equations such as the Kidney Failure Risk Equation (KFRE) to quantify absolute prognosis (13). For example, in CKD stages G3–G5, a 5-year risk ≥3–5% warrants nephrology referral, while a 2-year risk >10% indicates the need for multidisciplinary management. This “screen–confirm–stratify–refer/follow-up” loop provides a unified clinical framework that accommodates the integration of novel biomarkers, imaging metrics, and digital phenotypes on top of existing standard-of-care pathways (14).

Importantly, when high-risk stratification is directly linked to disease-modifying therapy, it evolves from a descriptive label to an actionable decision node. Within the same calibration and reclassification architecture, clinicians may layer additional modalities—such as imaging features, circulating biomarkers, and wearable-derived metrics—to refine individual risk prediction and provide early warnings during longitudinal follow-up (15–17). Among these, tongue and pulse characteristics rooted in Traditional Chinese Medicine (TCM) diagnostic paradigms are being translated into quantifiable digital phenotypes. These may serve as weak labels or correction terms in composite risk models (18, 19). Specific methodologies and supporting evidence are discussed in detail in subsequent sections.

3 Emerging technologies for early risk stratification of DKD

To extend the detection window of diabetic kidney disease (DKD), it is essential to enhance the current KDIGO risk stratification framework—based on eGFR, uACR, and the Kidney Failure Risk Equation (KFRE)—through the systematic integration of molecular, tissue, and continuous exposure-level data. These emerging approaches must be evaluated against externally validated hard clinical endpoints, such as annual eGFR slope, sustained ≥40% eGFR decline, progression to end-stage kidney disease (ESKD), or composite kidney–cardiovascular outcomes, to determine their true predictive and clinical utility (20, 21). This section summarizes key technologies, application scenarios, and methodological considerations across a “molecular–tissue–exposure–algorithm” continuum.

3.1 Genetics and epigenetics

Large-scale biobanks and multi-ethnic meta-analyses have significantly accelerated the identification of susceptibility loci and mechanistic insights related to DKD (22, 23). Recent studies in type 1 diabetes (T1D) cohorts have highlighted robust loci such as protective variants in COL4A3 (e.g., rs55703767), as well as signals in AFF3, FRMD3, and the RGMA–MCTP2 region, which are consistently associated with albuminuria, eGFR decline, and ESKD (24, 25). Multi-ancestry initiatives such as the FIND study demonstrate directional consistency across diverse populations, a critical feature for translational generalizability (25).

By contrast, the greater phenotypic heterogeneity in type 2 diabetes (T2D)—due to comorbidities such as hypertension, obesity, and dyslipidemia—has limited the detection of genome-wide significant loci (26). This underscores the need for refined phenotyping and larger, multi-ethnic cohorts to improve signal stability. Functional annotation strategies that integrate GWAS signals with kidney-specific eQTL, TWAS, and cell-type-resolved transcriptomic data are increasingly bridging the gap between statistical association and mechanistic interpretation.

3.1.1 Polygenic risk scores (PRS)

A 2024 systematic review suggests that polygenic risk scores (PRS) derived from existing GWAS datasets can identify individuals at elevated risk for DKD at the population level (27). However, predictive performance varies with ancestry, phenotype definition, and baseline risk distributions. In real-world settings—particularly those involving mixed ancestry, comorbid cardiovascular disease, or variable medication exposure—model calibration may drift, reducing clinical reliability (28).

Best practices for PRS development and deployment include multi-ancestry training, independent external validation, pre-registered thresholds, and transparent clinical interpretability (27). In this context, PRS may be best utilized not for population-wide screening or as replacements for KDIGO models, but rather as tools for mechanistic stratification or enrichment of clinical trial populations—especially for identifying genetically high-risk individuals with borderline or subclinical phenotypes.

3.1.2 Epigenetic reprogramming and single cell multi-omics

Epigenetic modifications associated with “metabolic memory”—including DNA methylation and histone remodeling—can persist despite glycemic normalization and contribute to podocyte injury, interstitial fibrosis, oxidative stress, and inflammation (29). Recent reviews (2024–2025) have mapped key epigenetic alterations in renal parenchymal cells, including dynamic shifts in DNA methylation and H3K4/H3K27 histone marks (30, 31). Although pharmacologic targeting of epigenetic pathways is promising, robust human data linking these features to hard renal outcomes remain scarce (32).

In parallel, single-cell and spatial multi-omics technologies are rapidly advancing nephrology research (33, 34). Platforms such as scRNA-seq and scATAC-seq allow for precise spatiotemporal mapping of cell-type-specific processes, linking genetic predisposition and epigenetic reprogramming to molecular phenotypes and microenvironmental signals (35).

In the near term, a practical application may lie in multimodal predictive modeling that incorporates epigenetic and transcriptomic features alongside clinical variables, imaging, and fluid biomarkers to refine early risk stratification and subtype identification (34). However, these models require validation in multi-center prospective cohorts with hard endpoints (e.g., eGFR slope, ESKD, cardiorenal events) to establish their incremental predictive value.

3.1.3 Integrative perspective

Current evidence supports an interactive model in which genetic predisposition, environmental exposures, and epigenetic reprogramming converge to shape DKD trajectory (27, 36). PRS may serve as tools for mechanistic stratification or high-risk trial recruitment, while epigenetic and single-cell approaches guide pathway-based intervention and target prioritization. Successful clinical translation will depend on external multi-ancestry validation, alignment with human kidney tissue data, and demonstration of added predictive utility anchored to meaningful clinical outcomes.

3.2 Optimization and standardization of core laboratory markers

eGFR and uACR remain the cornerstone biomarkers for early detection and longitudinal monitoring of DKD (37). However, their clinical utility depends on strict standardization of measurement and interpretation to ensure accuracy and consistency across diverse healthcare settings.

According to KDIGO 2024, serum creatinine–based eGFR (eGFRcr) is recommended for initial screening, while combined creatinine–cystatin C equations (eGFRcr-cys) should be prioritized in decision-critical scenarios when available (5, 38). All estimation equations should exclude race-based coefficients (39). For uACR, abnormal values must be confirmed on repeat testing, and risk stratification should be based on the joint interpretation of eGFR and uACR.

3.2.1 Systematic bias in creatinine measurement

Despite global efforts in assay standardization, serum creatinine remains susceptible to both methodological and interference-related variability. Discrepancies between enzymatic and Jaffe methods, as well as inter-laboratory variation, can result in inconsistent eGFR values and CKD staging errors. External quality assessments and multi-center evaluations continue to detect residual measurement differences even under standardized protocols (40).

Common analytical interferences include falsely low enzymatic readings due to dopamine or dobutamine administration, pseudo-hypocreatininemia caused by IgM paraproteinemia, and interference from N-acetylcysteine. These factors are especially relevant when eGFR values approach clinical decision thresholds (e.g., for drug dosing or nephrology referral) and warrant confirmatory testing or the use of alternative biomarkers when appropriate (41, 42).

3.2.2 Improved estimation equations

The 2021 CKD-EPI equations, which exclude race and incorporate cystatin C, offer improved accuracy and reduced demographic bias compared to traditional creatinine-only models (39). In 2023, a further refinement was proposed: a cystatin C–C-based formula that also excludes sex, offering a robust alternative in patients with abnormal muscle mass, atypical creatinine kinetics, or near dosing thresholds (43).In clinical practice, eGFRcr-cys or cystatin C–only equations should be considered at decision points where misclassification may have therapeutic consequences, provided testing infrastructure allows (14, 38).

3.2.3 Measured GFR as the analytical gold standard

When eGFR estimates are discordant with clinical presentation, or in cases involving extreme body composition, abnormal creatinine/cystatin C kinetics, or critical research/drug-dosing scenarios, direct measurement of GFR (mGFR) is warranted. Standardized plasma clearance techniques using non-radioactive tracers (e.g., iohexol) have demonstrated feasibility and reproducibility in both clinical and research settings, and serve as the gold standard for GFR assessment (44).

3.2.4 Standardization of uACR measurement

Like eGFR, uACR measurement requires rigorous standardization (45). Reference methods such as LC-MS/MS and external proficiency testing have revealed systematic discrepancies between routine immunoassays and isotope dilution mass spectrometry (IDMS) (46). Reference materials—such as NIST SRM 2925 (recombinant human albumin)—enable traceability and help reduce inter-laboratory variability, thereby improving comparability in longitudinal studies and multicenter trials (47). Clinical laboratories should disclose the traceability and uncertainty of their assays. In research contexts, preference should be given to externally validated and calibrated platforms.

3.2.5 Analytical calibration as a prerequisite for biomarker integration

Robust analytical calibration at the foundational biomarker level is a prerequisite for the meaningful integration of novel molecular, imaging, or digital phenotyping tools. Without such calibration, perceived gains in prediction may reflect the correction of baseline measurement error rather than the true biological signal. Therefore, technical rigor in eGFR and uACR assessment is essential to support the valid clinical deployment of emerging diagnostic technologies.

3.3 Inflammatory and tubular injury biomarkers

Beyond eGFR and uACR, circulating biomarkers reflecting inflammation and tubular injury have shown consistent and reproducible associations with DKD progression (48, 49). Among these, soluble tumor necrosis factor receptors 1 and 2 (TNFR1/2) and kidney injury molecule-1 (KIM-1) represent the most robust biomarker axis with established clinical relevance (50, 51).

Prospective cohort studies across type 1 and type 2 diabetes populations have demonstrated that elevated TNFR1/2 levels—whether at baseline or over time—are independently associated with accelerated eGFR decline, progression to CKD stage 3 or higher, and increased risk of end-stage kidney disease (ESKD), independent of albuminuria and traditional risk factors (50, 52, 53). KIM-1, a marker of proximal tubular damage, has also been consistently linked to adverse renal outcomes in various cohorts (54, 55).

These findings suggest that inflammatory and tubular injury biomarkers can provide additive prognostic information for intermediate- and long-term renal outcomes, beyond standard KDIGO risk stratification based on eGFR and uACR (49, 56).

3.3.1 Clinical integration: The KidneyIntelX.dkd model

KidneyIntelX.dkd is a multi-biomarker machine learning–based risk score that integrates TNFR1, TNFR2, and KIM-1 with key clinical variables such as eGFR, uACR, HbA1c, and blood urea nitrogen (56, 57). This model has received De Novo authorization from the U.S. Food and Drug Administration (FDA) for predicting the 5-year risk of progressive kidney function decline in adults with diabetes and early-stage DKD.

The tool supports individualized care pathways by informing follow-up intensity, referral to nephrology, or clinical trial enrollment (58). Both derivation/validation studies and real-world implementations have shown that combining biomarkers with clinical parameters improves discrimination and facilitates actionable risk stratification (56, 58, 59).

3.3.2 Population-specific considerations and therapeutic responsiveness

Biomarker performance may vary across diabetes subtypes and disease stages (48). For example, in type 1 diabetes, the independent predictive value of urinary KIM-1 for ESKD is attenuated after adjusting for albuminuria, indicating its potential utility in early-stage disease or in conjunction with inflammatory markers (60).

While TNFR1/2 are more robust than many other inflammatory biomarkers, their levels may be affected by systemic inflammation, limiting disease specificity (61). Therefore, clinical interpretation should account for comorbid conditions and be contextualized within the broader phenotypic landscape.

Emerging evidence also suggests that these biomarkers are modifiable by pharmacologic interventions. Randomized trials of sodium–glucose cotransporter 2 (SGLT2) inhibitors have shown reductions in KIM-1 and modest decreases in TNFR1/2, indicating their potential utility for pharmacodynamic monitoring and treatment response assessment (62, 63).

3.3.3 Methodological and implementation considerations

To enable clinical translation of these biomarkers, robust analytical validation, external performance verification, and seamless clinical integration are essential. Assay platforms should clearly define the sample matrix (e.g., serum or plasma), calibration protocols, precision parameters, and dynamic ranges to ensure inter-laboratory comparability (64). Predictive models must undergo external validation across ethnically and geographically diverse cohorts using standardized outcomes such as ≥40% sustained eGFR decline, ESKD, or composite renal/cardiovascular events (65). Reporting should include discrimination (C-statistic, AUC), calibration (slope, Brier score), and clinical utility metrics (e.g., NRI, IDI, decision curve analysis) (66). Integration into routine care should align with the KDIGO 2024 framework, with predefined trigger thresholds for nephrology referral and follow-up intensity. Post-implementation strategies must include recalibration procedures and monitoring for model performance drift (67).

3.3.4 Summary and clinical implications

Inflammatory (TNFR1/2) and tubular injury (KIM-1) biomarkers offer independent prognostic value beyond traditional markers and can aid in early identification of high-risk patients, individualized care planning, and enrichment of clinical trial cohorts. However, widespread clinical adoption will require harmonization of assay platforms, validation across diverse populations, and the accumulation of real-world evidence demonstrating improvement in patient reclassification and clinical outcomes.

3.4 Urinary proteomics and the CKD273 classifier

Urinary proteomics has emerged as a promising modality for the early detection and risk stratification of DKD (68). Among the various platforms, the 273 urinary peptides (CKD273) classifier—developed using capillary electrophoresis coupled with mass spectrometry (CE-MS)—is currently the most extensively validated urinary peptide panel for kidney risk assessment (69).

In individuals with type 2 diabetes and normoalbuminuria at baseline, elevated CKD273 scores have been shown to predict progression to microalbuminuria within 2–3 years, independent of changes in uACR (70). This prognostic value has been confirmed in large-scale studies, including the PRIORITY trial and the DIRECT-Protect 2 sub-study, supporting its utility as a preclinical risk identifier that precedes overt albuminuria and facilitates early screening and enrichment in clinical trials (70, 71).

3.4.1 Responsiveness to pharmacologic interventions

In a randomized, double-blind, crossover trial, the addition of dapagliflozin for 12 weeks on top of background renin–angiotensin system (RAS) inhibition significantly reduced CKD273 scores and shifted a subset of individuals from a high-risk to a low-risk peptide profile. These findings position CKD273 as a potential pharmacodynamic marker of treatment response (72). Similar proteomic modulation has been observed with empagliflozin, suggesting a class effect of SGLT2 inhibitors on the urinary peptide signature (73). Conversely, the addition of spironolactone in patients with high CKD273 scores failed to prevent microalbuminuria onset, indicating limited evidence for mineralocorticoid receptor antagonist (MRA) therapy guided by this biomarker at present (70). Overall, CKD273 appears more suited as an early “trigger” and a treatment response indicator, rather than a diagnostic surrogate (69).

3.4.2 Methodological considerations for clinical translation

The development and validation of CKD273 have predominantly relied on CE-MS (74). Despite its strong intra- and inter-assay reproducibility, variability introduced by batch effects, platform heterogeneity (e.g., CE-MS vs. LC-MS/MS), and pre-analytical procedures (e.g., centrifugation, desalting, freeze–thaw cycles) remains a major challenge for multicenter deployment (75, 76). To ensure analytical reliability and comparability, strict quality control, batch normalization, and inter-laboratory calibration protocols must be explicitly implemented in both research and clinical settings.

3.4.3 Suggested clinical applications and evaluation framework

The CKD273 classifier may serve as a complementary biomarker layer atop standard KDIGO-based risk stratification (e.g., eGFR × uACR, KFRE), with potential applications in three distinct clinical scenarios (69). First, in patients with borderline or fluctuating uACR levels, a positive CKD273 result could trigger enhanced surveillance strategies such as intensified uACR retesting or ambulatory blood pressure monitoring(ABPM). Second, in high-risk individuals without overt proteinuria, CKD273 can help identify subpopulations prone to rapid progression, thereby improving event rates and statistical power in clinical trials. Third, in therapeutic monitoring, a post-treatment reduction in CKD273 score may serve as a pharmacodynamic marker when interpreted alongside uACR decline and eGFR slope stabilization. To ensure clinical relevance, all studies evaluating CKD273 should report not only discrimination metrics (e.g., C-statistic, AUC), but also calibration (e.g., calibration slope, Brier score) and clinical utility indices (e.g., NRI, IDI, decision curve analysis). Furthermore, external validation should prioritize hard clinical endpoints such as sustained ≥40% eGFR decline, ESKD, or cardio–renal composite events to demonstrate incremental prognostic value beyond existing tools.

3.4.4 Positioning and future directions

CKD273 offers incremental risk information before overt albuminuria and exhibits quantifiable responsiveness to SGLT2 inhibitor therapy. Its optimal role is as an early activation signal and a treatment monitoring tool layered on top of standard care. Future directions include large-scale, multi-center prospective studies with harmonized protocols, validation against hard clinical endpoints, and generation of real-world evidence to demonstrate reclassification benefit and net clinical impact.

3.5 Urinary extracellular vesicles

Urinary extracellular vesicles (uEVs) are emerging as a promising biomarker modality for DKD, offering several advantages for translational research (77). They originate from various nephron segments, are noninvasively and repeatedly collectible, and reflect dynamic renal pathophysiology. The cargo of uEVs—including proteins and microRNAs—provides insight into glomerular and tubular compartments, enabling their use as mechanistic surrogates, early risk indicators, and potential pharmacodynamic markers (78). Podocyte-derived proteins have been linked to albuminuria, while tubular injury markers such as aquaporins and transporters may reflect tubulointerstitial damage.

3.5.1 Evidence and limitations of clinical application

Although uEVs show potential for early stratification and mechanistic mapping, there remains a lack of large-scale, prospective validation studies using hard renal endpoints such as sustained eGFR decline, ESKD, or cardio–renal composite outcomes (79). Current evidence is primarily derived from cross-sectional or small longitudinal cohorts (80). To assess incremental value, standardized phenotype definitions and methodological consistency are essential.

3.5.2 Technical standardization and reporting guidelines

Technical challenges specific to uEVs include variability in sample dilution, low protein concentrations, and contamination from non-renal sources. Recent international guidelines emphasize the importance of unified nomenclature, standardized isolation techniques such as ultracentrifugation and immunocapture, and normalization approaches using urinary creatinine, volume, or vesicle-specific markers (80). Reporting standards should include particle count, protein content, and characterization markers, alongside inter-batch and cross-platform quality control.

3.5.3 Short-term positioning: augmentation layer rather than standalone tool

At the current stage of evidence, uEVs are best regarded as a complementary biomarker layer rather than a replacement for conventional tools (80). When used in conjunction with urinary proteomics or serum markers, uEVs may enhance risk stratification within the KDIGO framework. Evaluation should include not only discrimination metrics but also calibration performance, reclassification indices, and clinical decision impact analyses, particularly concerning testing frequency, referral timing, and therapeutic escalation.

3.5.4 Research priorities and marker candidates

Research should prioritize the development of multi-ethnic prospective cohorts, harmonized pre-analytical workflows, and validation strategies anchored to hard clinical endpoints (78). Candidate biomarkers of interest include podocyte-derived proteins and associated microRNAs, as well as vesicular cargo originating from distinct tubular segments or signaling pathways (81, 82). These may support the delineation of tubulointerstitial-dominant phenotypes and enable more personalized treatment approaches. Ultimately, uEVs hold potential as part of an integrated, multi-modal biomarker strategy for early detection, risk classification, and treatment monitoring in DKD.

3.6 Imaging-based characterization of renal tissue injury

3.6.1 Overview of imaging modalities

Recent advances in non-invasive imaging technologies have enabled the direct visualization and quantification of renal tissue alterations that precede detectable changes in conventional biomarkers such as eGFR and uACR (83). Among these modalities, multiparametric magnetic resonance imaging (MRI) has emerged as a robust tool for assessing renal perfusion, microstructure, and oxygenation (84). Techniques such as arterial spin labeling (ASL) quantify cortical blood flow; diffusion-weighted and diffusion tensor imaging (DWI/DTI) evaluate tissue diffusivity and microstructural integrity; and blood oxygen level–dependent (BOLD) MRI captures tissue oxygenation through deoxyhemoglobin signal variation (85). Collectively, these metrics can detect early renal alterations in individuals with diabetes, even before a measurable decline in kidney function (86).

3.6.2 Standardization and technical challenges

To facilitate clinical translation, initiatives such as PARENCHIMA have proposed standardized acquisition, processing, and reporting protocols across ASL, DWI, BOLD, and T1/T2 mapping (84). These guidelines emphasize the harmonization of pre-scan variables—hydration status, sodium intake, body positioning—and recommend prioritizing cortical regions to improve reproducibility across imaging platforms. Despite these advances, widespread adoption of renal MRI remains limited by high cost, extended scan durations, and inter-vendor variability (87).

3.6.3 Emerging role of shear wave elastography

In addition to MRI, shear wave elastography (SWE) has been investigated as a non-invasive surrogate for renal fibrosis (88). By quantifying cortical stiffness, SWE offers insight into interstitial fibrosis, supported by biopsy-correlated studies and meta-analyses. Combining SWE with conventional biomarkers (e.g., eGFR) may enhance discrimination of fibrosis stages (88). However, SWE is highly operator- and device-dependent and lacks standardized thresholds across platforms, limiting its broad clinical utility.

3.6.4 Suggested clinical applications

These imaging tools are best positioned as adjunctive modalities in selected clinical contexts rather than as general screening instruments. Their integration is particularly valuable in three scenarios (1): enriching clinical trial populations by identifying individuals with greater subclinical tissue injury—such as impaired perfusion, restricted diffusion, or increased cortical stiffness—thereby improving event rates and trial efficiency (2); monitoring therapeutic response in patients receiving renoprotective interventions (e.g., SGLT2 inhibitors, mineralocorticoid receptor antagonists, GLP-1 receptor agonists), where dynamic imaging changes may complement conventional markers like uACR and eGFR slope; and (3) evaluating the incremental prognostic value of imaging biomarkers in prospective cohorts anchored by hard clinical endpoints (e.g., sustained ≥40% eGFR decline, ESKD, or cardio–renal composite outcomes), with emphasis on reclassification performance and net clinical benefit (89).

3.6.5 Future directions and research needs

Multiparametric renal MRI and shear wave elastography (SWE) offer valuable insights into subclinical tissue-level injury and hold promise as precision imaging biomarkers within emerging nephrology frameworks (90). Their optimal role lies in augmenting, rather than replacing, existing clinical assessments, particularly in early detection and treatment response monitoring. Moving forward, priorities include the development of harmonized acquisition protocols, robust multicenter validation across imaging platforms, and demonstration of additive prognostic value over conventional tools such as eGFR and uACR. Additionally, integration with clinical decision pathways will require health-economic evaluations, workflow optimization, and regulatory recognition to support scalable adoption.

3.7 Machine learning and longitudinal modeling for risk prediction in DKD

The progression of DKD is a time-dependent process, so models that use only a single baseline visit often miss patients who will deteriorate quickly (91). Using information over time—for example eGFR slope, change in uACR, and patterns of medication use—can make risk estimates more stable and more useful for clinical decisions. Continuous data from wearable devices such as CGM and ABPM also describe day-to-day metabolic and blood-pressure stress in much greater detail (92). When prediction models are trained to forecast hard outcomes, such as a sustained ≥40% eGFR decline, ESKD, or major cardio-renal events, they are directly aligned with thresholds that clinicians already recognize as actionable end points (93). Several studies show that eGFR slope is a useful surrogate for CKD progression and an important predictor in prognostic models, while CGM and ABPM provide additional phenotypic information for refining risk stratification (94, 95).

3.7.1 Advantages of longitudinal modeling and data sources

Repeated measurements of eGFR and uACR allow the construction of individual trajectories rather than relying on a single value. Models that use these trajectories generally identify high-risk patients earlier and more accurately than static models based only on baseline measurements (96). Incorporating eGFR slope into prediction models has improved calibration for estimating 2-year risk of ESKD. Joint modeling approaches, which analyze biomarker trajectories and time-to-event outcomes together, further strengthen prediction and reduce bias when progression is variable or follow-up is incomplete (97, 98). In parallel, wearable-derived metrics such as CGM-based glycaemic variability and ABPM-based blood-pressure fluctuations provide high-resolution information on exposure to harmful metabolic and hemodynamic stress, and these metrics correlate with adverse cardio-renal outcomes. For these reasons, longitudinal and high-frequency data streams are increasingly viewed as core components of modern DKD risk models (99, 100).

3.7.2 Modeling paradigms and temporal methods

Several time-aware modeling strategies have been applied to DKD. Landmark or dynamic prediction frameworks update an individual’s risk estimate at predefined time points based on all information accumulated so far, which mirrors routine follow-up visits in outpatient care (101, 102). Time-dependent Cox models and joint models combine evolving biomarker values with survival outcomes, allowing risk to change as kidney function or albuminuria changes and reducing bias from informative dropout or censoring (103). Deep learning methods, including recurrent neural networks and architectures such as RETAIN and BEHRT, can integrate information from many time points and modalities while retaining some level of interpretability for clinicians (104). Regardless of the specific technique, it is essential to define the observation window (what history is used) and the prediction horizon (what time frame is being predicted), and to train and test models using hard clinical end points to avoid data leakage and to keep the results clinically meaningful (105).

3.7.3 Multimodal fusion and missing data handling

In practice, DKD risk models can combine multiple data types—clinical variables, laboratory tests, omics, imaging markers and digital phenotypes. These can be fused at different stages of the modeling pipeline, such as early fusion (simple concatenation of features) or later fusion of intermediate representations. Real-world datasets often contain irregular sampling and missing values, which require appropriate handling through modality-specific masking, imputation strategies or dropout-based methods. Differences between hospitals, laboratories and devices can also degrade performance when a model is moved to a new setting. Domain-adaptation and federated-learning approaches have been proposed to reduce this problem and to allow multi-center model development without sharing raw patient data (106–108). A notable example of successful multimodal fusion is KidneyIntelX.dkd, which combines TNFR1, TNFR2, KIM-1 and routine clinical variables and has received FDA De Novo authorization for assessing DKD progression risk, illustrating that such pipelines can reach clinical-grade implementation.

3.7.4 Validation, reporting, and fairness

Before deployment, DKD risk models require careful validation. This includes internal validation using approaches such as nested or time-split cross-validation, and external validation in independent cohorts that differ by center, device and patient mix. Key performance metrics include discrimination (AUROC, C-index), calibration (calibration slope, Brier score) and measures of clinical usefulness such as net reclassification improvement, integrated discrimination improvement and decision-curve analysis, ideally reported for different prediction horizons using time-dependent AUROCs (106). Model development and reporting should follow established guidelines such as TRIPOD-AI and PROBAST-AI, and should include fairness assessments across sex, age, ethnic and socioeconomic subgroups. Fairness metrics—for example equalized odds, calibration within subgroups and comparison of error rates—help to ensure that model performance is equitable across diverse populations (109).

3.7.5 Post-deployment monitoring and recalibration

After a model is introduced into clinical practice, its performance can change as populations, treatments and measurement practices evolve. Therefore, ongoing monitoring is required to detect data drift, loss of calibration and deterioration in specific subgroups. Monitoring can use simple tools such as stability indices, calibration plots and periodic assessment of net clinical benefit. When pre-specified thresholds are crossed, a stepwise recalibration plan—starting with simple updates to the intercept and slope and escalating to partial or full retraining—should be implemented (110). Governance frameworks, including the FDA’s Good Machine Learning Practice (GMLP) and the NIST AI Risk Management Framework (AI RMF), provide guidance on documentation, audit trails and change-control processes that support safe model updating and lifecycle management (111). Repeated fairness audits and real-world outcome evaluations are also important to maintain trustworthiness over time (112).

3.7.6 Representative applications and case studies

In multicenter CKD cohorts, joint models that incorporate eGFR slope together with time-to-event data have improved both calibration and discrimination for predicting kidney failure compared with simpler approaches (91). KDIGO guidelines support the use of ABPM for CKD risk stratification, and CGM is increasingly used in patients with advanced CKD and dialysis for time-in-range–based glycaemic management, providing rich inputs for risk models. KidneyIntelX.dkd is an example of a regulatory-grade model that combines biomarker and clinical data and has been evaluated in prospective and real-world settings, offering a practical template for future translational efforts (113). These examples show that clinically relevant, externally validated and equity-aware modeling frameworks for DKD risk stratification are feasible in real-world practice.

Overall, machine-learning models for DKD risk prediction should be built on longitudinal data, integrate multiple complementary data sources and be trained on clearly defined, clinically meaningful endpoints. Only models that undergo rigorous validation, fairness assessment and ongoing performance monitoring are likely to deliver sustained benefit in everyday care. Aligning these efforts with clinical guidelines such as KDIGO and KFRE, and with regulatory frameworks such as GMLP and AI RMF, is crucial for technical robustness, clinician confidence and wider adoption across diverse healthcare systems (114).

3.8 Integrating TCM-informed digital phenotypes into DKD risk stratification

3.8.1 Conceptual rationale and digital modalities

Traditional Chinese Medicine (TCM) emphasizes zheng (syndrome) as a reflection of dynamic physiological imbalance (115). In the context of DKD, zheng can be reframed as a quantifiable digital phenotype that complements modern risk stratification (116). Advances in sensor and imaging technologies enable the digitization of classical TCM diagnostics: tongue imaging (inspection), pulse waveform acquisition (palpation), structured symptom questionnaires (inquiry), and potentially voice or breath sensing (listening/smelling) (117). Among these, tongue imaging and pulse signals are the most developed. Standardized tongue images yield features such as coating thickness, texture, and color, associated with inflammation, microcirculation, and metabolic status (118). Pulse waveforms captured via photoplethysmography (PPG) offer non-invasive measures of vascular compliance and autonomic tone (119).

3.8.2 Technical standardization and multi-center reproducibility

To ensure multi-site generalizability, strict acquisition protocols are essential. For tongue imaging, a consistent light source, shooting angle, and exclusion of blurry or overexposed images are required (120, 121). Pulse signals must be acquired in quiet, temperature-controlled environments, with signal quality assurance. Device heterogeneity and environmental variability necessitate pre-specified normalization, calibration pipelines, and population-specific validation. These steps are crucial to establishing reliability before integration into predictive models or clinical workflows (122).

3.8.3 Role in early risk detection and longitudinal monitoring

Digital phenotypes can enhance early DKD risk identification by detecting subclinical changes preceding proteinuria or eGFR decline (123). Their low cost, self-collectibility, and suitability for high-frequency monitoring make them valuable for continuous risk surveillance. When used alongside CGM, ABPM, laboratory indices, and imaging markers, these features may enrich predictive accuracy and personalize follow-up intensity. They also provide actionable signals in remote or resource-limited settings, enabling early alerts and behavioral interventions.

3.8.4 Modeling strategy and real-world applications

TCM-derived digital features can be incorporated into multimodal machine learning models targeting hard renal endpoints such as sustained eGFR decline or dialysis initiation (123, 124). Studies using tongue image datasets have shown predictive utility for metabolic traits, while pulse waveform analyses correlate with arterial stiffness—a known predictor of renal outcomes (125, 126). These signals should be evaluated not only by AUROC but also by calibration, reclassification, and clinical decision utility. Promising use cases include screening, remote monitoring, and adjunctive response evaluation. Their deployment must ensure external validation, interpretability, and fairness across subpopulations.

3.8.5 Limitations and comparison with established digital tools

While TCM-derived digital phenotypes offer an attractive, low-cost avenue for continuous monitoring, their current evidence base is substantially weaker than that of established digital health tools such as pulse wave analysis, ABPM, and continuous glucose monitoring (CGM). Most tongue and pulse studies are single-center, cross-sectional, or small longitudinal cohorts with heterogeneous inclusion criteria and limited adjustment for confounders. Acquisition protocols vary widely across devices and environments, which can introduce substantial measurement noise and reduce cross-site reproducibility.

Moreover, cross-cultural generalizability remains a key challenge. Tongue color, coating, and pulse characteristics can be affected by diet, oral hygiene, skin tone, and comorbid conditions, all of which may differ across populations. In contrast, CGM and ABPM have undergone large-scale, randomized and observational validation, are incorporated into international guidelines, and have well-defined clinical thresholds. At present, therefore, TCM-derived digital phenotypes should be regarded as exploratory signals that may enrich multimodal models or support remote hypothesis generation, but they cannot replace validated digital tools for risk stratification or treatment guidance.

3.8.6 Ethical, fairness, and implementation considerations

Factors such as skin tone, oral hygiene, comorbidities, and acquisition environment can introduce bias. Models must report performance stratified by age, sex, ethnicity, and device type. Fairness-aware modeling strategies (e.g., reweighting, subgroup-specific training) are critical to mitigate disparities (120). Data privacy, auditability, and version control should be embedded through localized or federated learning pipelines. Ultimately, the value of these digital phenotypes will be determined by prospective validation, cost-effectiveness, and their ability to accelerate time-to-intervention within DKD care pathways.

3.9 Comparative appraisal of emerging technologies

To move beyond descriptive listing of individual modalities, Table 2 provides a comparative overview of key emerging technologies for early DKD risk stratification. For each modality, we summarize typical study designs, approximate sample sizes, validation status, and current readiness for clinical translation. Overall, urinary proteomics (e.g., CKD273) and inflammatory/tubular biomarkers integrated into clinical-grade models such as KidneyIntelX.dkd are supported by the most consistent prospective data, including external validation and some regulatory authorization. By contrast, urinary extracellular vesicles, multimodal renal MRI, shear wave elastography, and TCM-derived digital phenotypes remain at an earlier stage, with promising mechanistic insights but limited large-scale, endpoint-anchored validation and scarce health-economic evaluation.

Table 2

Table 2. Emerging technologies for DKD risk stratification: study characteristics, validation status, and clinical readiness.

This comparative view underscores that many of the technologies discussed in this review are still largely hypothesis-generating. Only a subset have progressed to the level of clinically actionable tools that can be embedded into KDIGO 2024–aligned care pathways. Future research should therefore prioritize prospective, multi-center studies that benchmark novel markers directly against established anchors—eGFR, uACR, and KFRE—and report incremental value using calibration, reclassification, and clinical utility metrics.

4 Risk-driven therapeutic intensification and response monitoring

4.1 Foundational therapies: RAAS blockade and SGLT2 inhibitors

Early risk identification in DKD must be translated into the timely initiation of disease-modifying therapy. Across major guidelines, sodium–glucose cotransporter 2 inhibitors (SGLT2i) are now considered foundational for patients with CKD and albuminuria, regardless of diabetes status (5). Landmark trials such as DAPA-CKD and EMPA-KIDNEY confirmed consistent kidney-protective effects across diabetic and non-diabetic subgroups (6). These results support a risk-based rather than disease-centric approach, establishing SGLT2i as first-line agents when albuminuria is present and eGFR is within a safe prescribing range.

4.2 Add-on strategies for residual risk: non-steroidal MRAs and GLP-1 receptor agonists

For patients remaining at high risk—particularly those in the red/orange zones of eGFR-uACR grids or with persistent proteinuria—further intensification is warranted. Finerenone, a non-steroidal mineralocorticoid receptor antagonist (MRA), has demonstrated additive renal and cardiovascular benefit when added to background RAAS inhibition, as shown in FIDELIO-DKD and FIGARO-DKD (7). While hyperkalemia is a concern, it is manageable with serum potassium monitoring and stepwise titration. Recent data suggest that SGLT2i-MRA co-administration yields synergistic albuminuria reduction (128).

In parallel, glucagon-like peptide-1 receptor agonists (GLP-1RA) have shown renal benefit. The FLOW trial demonstrated that weekly semaglutide reduced the risk of composite kidney outcomes by 24%, alongside slowing eGFR decline and improving cardiovascular and all-cause mortality (12). These findings support GLP-1RA as a third-line agent for patients with persistent risk, obesity, or atherosclerotic disease despite optimized RAAS and SGLT2i therapy (129).

4.3 Monitoring efficacy and safety: risk-anchored protocols

KDIGO 2024 provides a framework for risk-anchored therapeutic monitoring. A modest eGFR drop (≤30%) following initiation of RAAS blockers or SGLT2i is expected and not grounds for discontinuation (130). However, greater declines necessitate evaluation of volume status, hypotensive burden, or drug–drug interactions. For finerenone, serum potassium and creatinine should be checked post-initiation and during titration (131). Reduction in UACR and flattening of eGFR slope are actionable intermediate endpoints, correlated with long-term outcome improvement, and suitable for outpatient decision-making.

4.4 A structured treatment pathway for DKD

A structured, risk-stratified treatment algorithm provides a pragmatic approach to therapeutic intensification in DKD. Initial risk classification can be achieved using eGFR–uACR heatmaps and the Kidney Failure Risk Equation (KFRE), enabling clinicians to identify patients requiring immediate intervention (5). For those with moderate to high risk, first-line therapy should combine SGLT2 inhibitors with optimized RAAS blockade (132). In cases of residual proteinuria or high total risk, non-steroidal MRAs such as finerenone may be added, followed by GLP-1 receptor agonists in patients with concurrent obesity, atherosclerotic disease, or inadequate risk control (133). Treatment effectiveness should be dynamically assessed using trends in UACR, eGFR slope, ambulatory blood pressure, and CGM-derived metrics (in diabetic individuals), with therapy adjusted accordingly. Importantly, transient eGFR declines post-initiation—particularly <30%—should not prompt discontinuation, but rather warrant evaluation of hemodynamic factors and reassessment of medication interactions to preserve long-term renoprotective benefit.

4.5 Compatibility with early detection and digital phenotyping tools

This risk-based therapeutic framework is inherently synergistic with the early detection technologies and TCM-informed digital phenotypes outlined in prior sections. Emerging tools such as multiparametric renal MRI, tongue/pulse signal analytics, and machine learning-based longitudinal prediction models can enhance the precision of risk assessment, allowing for earlier identification of high-risk individuals who may otherwise be missed using conventional markers alone. Moreover, continuous tracking of digital phenotypes and imaging biomarkers enables earlier detection of favorable treatment responses, providing a rationale for adaptive therapy escalation or de-escalation. Such integration supports more efficient resource allocation—targeting follow-up intensity and specialist referral toward those most likely to benefit—and reinforces a future-ready model of precision DKD management that bridges traditional and emerging paradigms.

4.6 Alignment with KDIGO 2024 and current care models

From a practical standpoint, only a subset of the technologies discussed in this review are currently ready for broad implementation in routine care. Within the KDIGO 2024 framework, annual or risk-based measurement of eGFR and uACR, combined with the eGFR×uACR heat map and KFRE, remains the mandatory foundation for DKD screening and risk stratification in primary care, endocrinology, and nephrology settings. Foundational therapies—optimized RAAS blockade, SGLT2 inhibitors, and, where appropriate, non-steroidal MRAs and GLP-1 receptor agonists—can be deployed today using existing guideline algorithms.

In contrast, urinary proteomics (CKD273), inflammatory/tubular biomarkers, multiparametric renal MRI, uEVs, and TCM-derived digital phenotypes are best viewed as adjunctive layers whose use should currently be confined to specialized centers, research protocols, or carefully designed implementation studies. When considered for clinical use, these tools should be embedded explicitly within guideline-based workflows—for example, as enrichment criteria for early nephrology referral, triggers for closer follow-up, or pharmacodynamic markers of therapy response—rather than as stand-alone replacements for established markers.

5 Ethics, regulation, and data governance

To enable the real-world implementation of multimodal early detection and risk-driven management frameworks, robust and replicable frameworks for ethics and data governance must be established from the outset. As modalities such as tongue imaging, pulse waveform acquisition, wearable monitoring, and home-based data collection extend data generation beyond clinical settings into daily life, the exposure surface for privacy risks increases substantially (134). Therefore, it is essential to adopt foundational data governance principles early in the research and deployment lifecycle. These include data minimization, purpose limitation, de-identification, and secure local storage. Furthermore, clear access controls and audit mechanisms must be established for data crossing national borders or interfacing with third-party algorithmic services.

When algorithms are integrated into clinical workflows as decision-support tools, human-machine collaboration must remain reversible and auditable (135). All automated outputs should include uncertainty indicators and interpretable insights, with the final clinical judgment retained by healthcare professionals (127). To ensure equitable utility, the model’s performance should be reported across subpopulations stratified by skin tone, language, health literacy, and device conditions. When disparities are observed, localized guidance and interface adaptations should be implemented to bridge the digital divide.

In addition to general principles of data protection, multimodal risk prediction systems for DKD often fall under the category of software as a medical device (SaMD). In the United States, the U.S. Food and Drug Administration (FDA) has published SaMD and Good Machine Learning Practice (GMLP) guidance documents that emphasize clear intended use, locked versus adaptive algorithms, predefined change-control plans, and ongoing post-market surveillance. Multimarker models such as KidneyIntelX.dkd provide an early example of a DKD-focused SaMD that has undergone De Novo review and authorization, highlighting the level of analytical and clinical validation expected for similar tools.

In Europe, AI-enabled diagnostic systems are subject to the Medical Device Regulation and the emerging European AI Act, which introduce risk-based classification, transparency requirements, and obligations for robustness, fairness, and human oversight. Systems used for high-stakes decisions in nephrology—such as predicting kidney failure or guiding treatment escalation—are likely to be categorized as “high-risk” AI, and therefore must undergo rigorous conformity assessment, documentation of training data provenance, and continuous performance monitoring.

At the level of data protection, region-specific frameworks such as the EU General Data Protection Regulation (GDPR), the U.S. Health Insurance Portability and Accountability Act (HIPAA), and national health data or personal information protection laws (e.g., China’s Personal Information Protection Law) mandate purpose limitation, data minimization, and strong safeguards for cross-border data transfer. For DKD-focused digital phenotyping, this implies that tongue images, pulse signals, and wearable data should be processed under explicit consent, with clear data retention policies and technical safeguards such as encryption, federated or on-device learning, and robust audit trails.

From a regulatory perspective, the evidence pathway for traditional biomarkers and imaging metrics typically progresses through three stages: analytical validity, clinical validity, and clinical utility. For software-as-a-medical-device (SaMD) systems, including multimodal risk prediction models, additional requirements apply (136). These include pre-specified protocols for change management, performance recalibration, and post-market surveillance. Regulatory approval and broad adoption are contingent upon a full chain of evidence demonstrating that the technology is safe, effective, interpretable, and economically viable. Only then can such integrated systems be ethically deployed and scaled in routine healthcare.

6 Limitations of current evidence and implementation considerations

Despite substantial progress, several limitations of the current evidence base must be acknowledged. First, many emerging biomarkers and imaging modalities have been evaluated primarily in small, single-center cohorts with limited ethnic diversity, short follow-up, and heterogeneous DKD phenotypes. For several technologies—particularly urinary extracellular vesicles, multiparametric renal MRI, shear wave elastography, and TCM-derived digital phenotypes—robust, multi-center studies anchored to hard endpoints (sustained eGFR decline, kidney failure, or cardio–renal composite outcomes) are still lacking.

Second, head-to-head comparisons between novel modalities and established markers are rare. As a result, it is often unclear whether a new test provides true incremental prognostic value beyond eGFR, uACR, and KFRE, or merely refines risk estimation in a statistically modest but clinically marginal way. Third, the majority of multimodal machine learning models have been developed and validated retrospectively in high-resource settings, with limited data on performance drift, equity, and generalizability in low- and middle-income countries.

Finally, health-economic and resource considerations are underexplored. High-throughput proteomics, multi-analyte biomarker panels, and advanced imaging techniques carry substantial per-patient costs and infrastructure requirements, which may not be feasible for routine use in many health systems. In contrast, simpler tools such as optimized eGFR and uACR measurement, structured risk scores, and guideline-directed pharmacotherapy are relatively inexpensive and widely implementable. Future research should therefore incorporate cost-effectiveness analyses and implementation science frameworks to determine which multimodal combinations deliver sufficient clinical benefit to justify real-world deployment.

7 Conclusion and future perspectives

The paradigm of DKD detection and management is undergoing a fundamental transformation—from static, late-stage biomarkers to an integrated, multimodal framework capable of capturing subclinical progression and guiding personalized intervention. Core to this evolution are validated anchors such as the eGFR×uACR grid and the Kidney Failure Risk Equation (KFRE), which now serve as the foundational layer upon which additional biological (e.g., blood and urine biomarkers), structural (e.g., functional renal imaging), and physiological (e.g., wearable-derived metrics) signals can be layered. Complementing these are digitally quantified TCM phenotypes—such as tongue features and pulse waveforms—which offer a low-cost, longitudinally accessible augmentation layer that captures subtle deviations from homeostasis.

Crucially, these technologies are not designed to replace current guideline-based workflows but to enhance them—particularly in the “gray zone” of early DKD where traditional indicators may be insensitive. When deployed within a risk-driven, closed-loop clinical pathway, these tools can shorten the lag between risk emergence and therapeutic intensification, enabling earlier and more targeted interventions. However, the translational focus must now shift from algorithmic discrimination to clinical impact. Future studies should prioritize hard outcomes such as sustained eGFR decline, ESKD, cardiovascular events, and health-economic metrics, underpinned by rigorous external validation across populations, settings, and acquisition devices.

Sustaining real-world utility will also require continuous model recalibration, data quality assurance, and bias mitigation strategies to ensure equitable performance across diverse subgroups. If implemented responsibly, this integrated and ethically grounded framework can deliver on the long-standing promise of precision nephrology: to identify high-risk individuals early, intervene proactively, and ultimately reduce the burden of DKD through individualized, scalable, and cost-effective care pathways.

Author contributions

LM: Funding acquisition, Writing – original draft. ZL: Data curation, Investigation, Writing – review & editing. LX: Investigation, Methodology, Writing – review & editing. FW: Data curation, Writing – review & editing. HJ: Data curation, Writing – review & editing. LZ: Project administration, Supervision, Writing – review & editing. AZ: Funding acquisition, Methodology, Writing – review & editing. ZZ: Funding acquisition, Supervision, Writing – review & editing.

Funding

The author(s) declared financial support was received for this work and/or its publication. This work was supported by the General Project of Social Development in Yangzhou City in 2024(YZ2024150), the National Natural Science Foundation of China (No. 82505424), 2025 Yangzhou Key R&D Program – Social Development Project (YZ2025444).

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. The author(s) verify and take full responsibility for the use of generative AI in the preparation of this manuscript. Generative AI was used to assist with language editing, translation from Chinese to English, and stylistic refinement to improve clarity and readability for the intended audience. All scientific content, interpretations, and conclusions are solely the responsibility of the authors.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Glossary

ABPM: ambulatory blood pressure monitoring

AE: adverse event

AI: artificial intelligence

AKI: acute kidney injury

ASCVD: atherosclerotic cardiovascular disease

ASL: arterial spin labeling

AUC: area under the receiver-operating characteristic curve

BOLD: blood oxygen level–dependent

CGM: continuous glucose monitoring

CKD: chronic kidney disease

CKD-EPI: Chronic Kidney Disease Epidemiology Collaboration

CKD273: 273 urinary peptides

DKD: diabetic kidney disease

EHR: electronic health record

ERBP: European Renal Best Practice

ESKD: end-stage kidney disease

EV: extracellular vesicle

EVs: extracellular vesicles

FDA: U.S. Food and Drug Administration

GFR: glomerular filtration rate

GDPR: General Data Protection Regulation

GLP-1: glucagon-like peptide-1

GLP-1RA: glucagon-like peptide-1 receptor agonist

GMLP: Good Machine Learning Practice

GWAS: genome-wide association study

HbA1c: glycated hemoglobin A1c

HIPAA: Health Insurance Portability and Accountability Act

IDMS: isotope dilution mass spectrometry

KDIGO: Kidney Disease Improving Global Outcomes

KFRE: Kidney Failure Risk Equation

KIM-1: kidney injury molecule-1

LC-MS/MS: liquid chromatography–tandem mass spectrometry

MAFLD: metabolic dysfunction-associated fatty liver disease

MISEV2023: Minimal Information for Studies of Extracellular Vesicles 2023

ML: machine learning

MRA: mineralocorticoid receptor antagonist

MRI: magnetic resonance imaging

ns-MRA: non-steroidal mineralocorticoid receptor antagonist

PPG: photoplethysmography

PIPL: Personal Information Protection Law of the People’s Republic of China

PRS: polygenic risk score

PRIORITY: Proteomic prediction and Renin–Angiotensin system Inhibition prevention Of early diabetic nephRopathy In TYpe 2 diabetes trial

RAAS: renin–angiotensin–aldosterone system

RAS: renin–angiotensin system

RCT: randomized controlled trial

SaMD: software as a medical device

SGLT2: sodium–glucose cotransporter 2

SGLT2i: sodium–glucose cotransporter 2 inhibitor

SWE: shear wave elastography

T1D: type 1 diabetes

T2D: type 2 diabetes

TCM: Traditional Chinese Medicine

TNF: tumor necrosis factor

TNFR1: tumor necrosis factor receptor 1

TNFR2: tumor necrosis factor receptor 2

TRIPOD: Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis

eGFR: estimated glomerular filtration rate

ns-MRA: non-steroidal mineralocorticoid receptor antagonist

uACR: urine albumin-to-creatinine ratio

uEV: urinary extracellular vesicle

uEVs: urinary extracellular vesicles.

References

1. Sun H, Saeedi P, Karuranga S, Pinkepank M, Ogurtsova K, Duncan BB, et al. IDF Diabetes Atlas: Global, regional and country-level diabetes prevalence estimates for 2021 and projections for 2045. Diabetes Res Clin Pract. (2022) 183:109119. doi: 10.1016/j.diabres.2021.109119

PubMed Abstract | Crossref Full Text | Google Scholar

2. Ong KL, Stafford LK, McLaughlin SA, Boyko EJ, Vollset SE, Smith AE, et al. Global, regional, and national burden of diabetes from 1990 to 2021, with projections of prevalence to 2050: a systematic analysis for the Global Burden of Disease Study 2021. Lancet. (2023) 402:203–34. doi: 10.1016/S0140-6736(23)01301-6

PubMed Abstract | Crossref Full Text | Google Scholar

3. Murray CJL and GBD 2021 Collaborators. Findings from the global burden of disease study 2021. Lancet. (2024) 403:2259–62. doi: 10.1016/S0140-6736(24)00769-4

PubMed Abstract | Crossref Full Text | Google Scholar

4. Butt MD, Ong SC, Rafiq A, Kalam MN, Sajjad A, Abdullah M, et al. A systematic review of the economic burden of diabetes mellitus: contrasting perspectives from high and low middle-income countries. J Pharm Policy Pract. (2024) 17(1):2322107. doi: 10.1080/20523211.2024.2322107

PubMed Abstract | Crossref Full Text | Google Scholar

5. Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work Group. KDIGO 2024 clinical practice guideline for the evaluation and management of chronic kidney disease. Kidney Int. (2024) 105:S117–314. doi: 10.1016/j.kint.2023.10.018

PubMed Abstract | Crossref Full Text | Google Scholar

6. Heerspink HJL, Stefánsson BV, Correa-Rotter R, Chertow GM, Greene T, Hou F-F, et al. Dapagliflozin in patients with chronic kidney disease. New Engl J Med. (2020) 383:1436–46. doi: 10.1056/NEJMoa2024816

PubMed Abstract | Crossref Full Text | Google Scholar

7. Bakris GL, Agarwal R, Anker SD, Pitt B, Ruilope LM, Rossing P, et al. Effect of finerenone on chronic kidney disease outcomes in type 2 diabetes. New Engl J Med. (2020) 383:2219–29. doi: 10.1056/NEJMoa2025845

PubMed Abstract | Crossref Full Text | Google Scholar

8. Pitt B, Filippatos G, Agarwal R, Anker SD, Bakris GL, Rossing P, et al. Cardiovascular events with finerenone in kidney disease and type 2 diabetes. New Engl J Med. (2021) 385:2252–63. doi: 10.1056/NEJMoa2110956

PubMed Abstract | Crossref Full Text | Google Scholar

9. Francis A, Harhay MN, Ong ACM, Tummalapalli SL, Ortiz A, Fogo AB, et al. Chronic kidney disease and the global public health agenda: an international consensus. Nat Rev Nephrol. (2024) 20:473–85. doi: 10.1038/s41581-024-00820-6

PubMed Abstract | Crossref Full Text | Google Scholar

10. Deng L, Guo S, Liu Y, Zhou Y, Liu Y, Zheng X, et al. Global, regional, and national burden of chronic kidney disease and its underlying etiologies from 1990 to 2021: a systematic analysis for the Global Burden of Disease Study 2021. BMC Public Health. (2025) 25:636. doi: 10.1186/s12889-025-21851-z

PubMed Abstract | Crossref Full Text | Google Scholar

11. Zhao M, Cao Y, and Ma L. New insights in the treatment of DKD: recent advances and future prospects. BMC Nephrol. (2025) 26:72. doi: 10.1186/s12882-025-03953-3

PubMed Abstract | Crossref Full Text | Google Scholar

12. Perkovic V, Tuttle KR, Rossing P, Mahaffey KW, Mann JFE, Bakris G, et al. Effects of semaglutide on chronic kidney disease in patients with type 2 diabetes. New Engl J Med. (2024) 391:109–21. doi: 10.1056/NEJMoa2403347

PubMed Abstract | Crossref Full Text | Google Scholar

13. Tangri N, Stevens LA, Griffith J, Tighiouart H, Djurdjev O, Naimark D, et al. A predictive model for progression of chronic kidney disease to kidney failure. Clinicoecon Outcomes Res. (2021) 13:475–486. doi: 10.2147/CEOR.S313857

PubMed Abstract | Crossref Full Text | Google Scholar

14. Iatridi F, Carrero JJ, Cornec-Le Gall E, Kanbay M, Luyckx V, Shroff R, et al. KDIGO 2024 clinical practice guideline for the evaluation and management of chronic kidney disease in children and adults: a commentary from the european renal best practice (ERBP). Nephrol Dial Transplant. (2024) 40:273–82. doi: 10.1093/ndt/gfae209

PubMed Abstract | Crossref Full Text | Google Scholar

15. Lioufas N, Hawley CM, Cameron JD, and Toussaint ND. Chronic kidney disease and pulse wave velocity: A narrative review. Int J Hypertens. (2019) 2019:9189362. doi: 10.1155/2019/9189362

PubMed Abstract | Crossref Full Text | Google Scholar

16. Liu J-J, Liu S, Lee J, Gurung RL, Yiamunaa M, Ang K, et al. Aortic pulse wave velocity, central pulse pressure, augmentation index, and chronic kidney disease progression in individuals with type 2 diabetes: a 3-year prospective study. BMC Nephrol. (2020) 21:359. doi: 10.1186/s12882-020-02024-z

PubMed Abstract | Crossref Full Text | Google Scholar

17. Wieringa FP, Broers NJH, Kooman JP, van der Sande FM, and Van Hoof C. Wearable sensors: can they benefit patients with chronic kidney disease? Expert Rev Med Devices. (2017) 14:505–19. doi: 10.1080/17434440.2017.1342533

PubMed Abstract | Crossref Full Text | Google Scholar

18. Thirunavukkarasu U, Umapathy S, Ravi V, and Alahmadi TJ. Tongue image fusion and analysis of thermal and visible images in diabetes mellitus using machine learning techniques. Sci Rep. (2024) 14:14571. doi: 10.1038/s41598-024-64150-0

PubMed Abstract | Crossref Full Text | Google Scholar

19. Hsu P-C, Chen J-M, Chang C-C, Chang Y-J, Chiu P-F, Chiang JY, et al. Exploring the pivotal variables of tongue diagnosis between patients with chronic kidney disease and health participants. Front Big Data. (2025) 7:1443646. doi: 10.3389/fdata.2024.1443646

PubMed Abstract | Crossref Full Text | Google Scholar

20. Levey AS, Gansevoort RT, Coresh J, Inker LA, Heerspink HL, Grams ME, et al. Change in albuminuria and GFR as end points for clinical trials in early stages of CKD: A scientific workshop sponsored by the national kidney foundation in collaboration with the US food and drug administration and european medicines agency. Am J Kidney Dis. (2020) 75:84–104. doi: 10.1053/j.ajkd.2019.06.009

PubMed Abstract | Crossref Full Text | Google Scholar

21. The EMPA-KIDNEY Collaborative Group, Herrington WG, Staplin N, Wanner C, Green JB, Hauske SJ, et al. Empagliflozin in patients with chronic kidney disease. N Engl J Med. (2023) 388:117–27. doi: 10.1056/NEJMoa2204233

PubMed Abstract | Crossref Full Text | Google Scholar

22. Brownlee M. Biochemistry and molecular cell biology of diabetic complications. Nature. (2001) 414:813–20. doi: 10.1038/414813a

PubMed Abstract | Crossref Full Text | Google Scholar

23. Kato M and Natarajan R. Epigenetics and epigenomics in diabetic kidney disease and metabolic memory. Nat Rev Nephrol. (2019) 15:327–45. doi: 10.1038/s41581-019-0135-6

PubMed Abstract | Crossref Full Text | Google Scholar

24. Salem RM, Todd JN, Sandholm N, Cole JB, Chen W-M, Andrews D, et al. Genome-wide association study of diabetic kidney disease highlights biology involved in glomerular basement membrane collagen. J Am Soc Nephrol. (2019) 30:2000–16. doi: 10.1681/ASN.2019030218

PubMed Abstract | Crossref Full Text | Google Scholar

25. Sandholm N, Salem RM, McKnight AJ, Brennan EP, Forsblom C, Isakova T, et al. New susceptibility loci associated with kidney disease in type 1 diabetes. PloS Genet. (2012) 8:e1002921. doi: 10.1371/journal.pgen.1002921

PubMed Abstract | Crossref Full Text | Google Scholar

26. Suzuki K, Hatzikotoulas K, Southam L, Taylor HJ, Yin X, Lorenz KM, et al. Genetic drivers of heterogeneity in type 2 diabetes pathophysiology. Nature. (2024) 627:347–57. doi: 10.1038/s41586-024-07019-6

PubMed Abstract | Crossref Full Text | Google Scholar

27. Ali AS, Pham C, Morahan G, and Ekinci EI. Genetic risk scores identify people at high risk of developing diabetic kidney disease: A systematic review. J Clin Endocrinol Metab. (2024) 109:1189–97. doi: 10.1210/clinem/dgad704

PubMed Abstract | Crossref Full Text | Google Scholar

28. Vy HMT, Coca SG, Sawant A, Sakhuja A, Gutierrez OM, Cooper R, et al. Genome-wide polygenic risk score for CKD in individuals with APOL1 high-risk genotypes. Clin J Am Soc Nephrol. (2024) 19:374. doi: 10.2215/CJN.0000000000000379

PubMed Abstract | Crossref Full Text | Google Scholar

29. Chen Z, Malek V, and Natarajan R. Update: the role of epigenetics in the metabolic memory of diabetic complications. Am J Physiol Renal Physiol. (2024) 327:F327–39. doi: 10.1152/ajprenal.00115.2024

PubMed Abstract | Crossref Full Text | Google Scholar

30. Zhou LY, Sofer T, Horimoto ARVR, Talavera GA, Lash JP, Cai J, et al. Polygenic risk scores and kidney traits in the Hispanic/Latino population: The Hispanic Community Health Study/Study of Latinos. HGGADVANCE. (2023) 4:100177. doi: 10.1016/j.xhgg.2023.100177

PubMed Abstract | Crossref Full Text | Google Scholar

31. Qu P, Li L, Jin Q, Liu D, Qiao Y, Zhang Y, et al. Histone methylation modification and diabetic kidney disease: Potential molecular mechanisms and therapeutic approaches (Review). Int J Mol Med. (2024) 54:104. doi: 10.3892/ijmm.2024.5428

PubMed Abstract | Crossref Full Text | Google Scholar

32. Liebisch M and Wolf G. Role of epigenetic changes in the pathophysiology of diabetic kidney disease. Glomerular Dis. (2024) 4:211–26. doi: 10.1159/000541923

PubMed Abstract | Crossref Full Text | Google Scholar

33. Tan W, Chen J, Wang Y, Xiang K, Lu X, Han Q, et al. Single-cell RNA sequencing in diabetic kidney disease: a literature review. Ren Fail. (2024) 46(2):2387428. doi: 10.1080/0886022X.2024.2387428

PubMed Abstract | Crossref Full Text | Google Scholar

34. Li H, Li D, Ledru N, Xuanyuan Q, Wu H, Asthana A, et al. Transcriptomic, epigenomic, and spatial metabolomic cell profiling redefines regional human kidney anatomy. Cell Metab. (2024) 36:1105–1125.e10. doi: 10.1016/j.cmet.2024.02.015

PubMed Abstract | Crossref Full Text | Google Scholar

35. Wilson PC, Muto Y, Wu H, Karihaloo A, Waikar SS, and Humphreys BD. Multimodal single cell sequencing implicates chromatin accessibility and genetic background in diabetic kidney disease progression. Nat Commun. (2022) 13:5253. doi: 10.1038/s41467-022-32972-z

PubMed Abstract | Crossref Full Text | Google Scholar

36. Dong H, Sun Y, Nie L, Cui A, Zhao P, Leung WK, et al. Metabolic memory: mechanisms and diseases. Sig Transduct Target Ther. (2024) 9:38. doi: 10.1038/s41392-024-01755-x

PubMed Abstract | Crossref Full Text | Google Scholar

37. Biljak VR, Honović L, Matica J, Krešić B, and Vojak SŠ. The role of laboratory testing in detection and classification of chronic kidney disease: national recommendations. Biochem Med (Zagreb). (2017) 27:153–76. doi: 10.11613/BM.2017.019

PubMed Abstract | Crossref Full Text | Google Scholar

38. Navaneethan SD, Bansal N, Cavanaugh KL, Chang A, Crowley S, Delgado C, et al. KDOQI US commentary on the KDIGO 2024 clinical practice guideline for the evaluation and management of CKD. Am J Kidney Dis. (2025) 85:135–76. doi: 10.1053/j.ajkd.2024.08.003

PubMed Abstract | Crossref Full Text | Google Scholar

39. Inker LA, Eneanya ND, Coresh J, Tighiouart H, Wang D, Sang Y, et al. New creatinine- and cystatin C–based equations to estimate GFR without race. New Engl J Med. (2021) 385:1737–49. doi: 10.1056/NEJMoa2102953

PubMed Abstract | Crossref Full Text | Google Scholar

40. Lundgren M, Ridefelt P, Kristoffersson C, Samuelsson I, and Larsson A. Inter-laboratory variation for urine albumin among laboratories in a Swedish external quality assessment scheme 2005–2023. Clin Biochem. (2024) 133–134:110825. doi: 10.1016/j.clinbiochem.2024.110825

PubMed Abstract | Crossref Full Text | Google Scholar

41. McCudden C, Clark EG, Akbari A, Kong J, Kanji S, and Hiremath S. N-acetylcysteine interference with creatinine measurement: an in vitro analysis. Kidney Int Rep. (2021) 6:1973–6. doi: 10.1016/j.ekir.2021.04.006

PubMed Abstract | Crossref Full Text | Google Scholar

42. Dao T and Young BY. Falsely low serum creatinine in an AKI patient on dopamine drip: SA-PO034. J Am Soc Nephrol. (2022) 33:611. doi: 10.1681/ASN.20223311S1611b

Crossref Full Text | Google Scholar

43. Pottel H, Björk J, Rule AD, Ebert N, Eriksen BO, Dubourg L, et al. Cystatin C-based equation to estimate GFR without the inclusion of race and sex. N Engl J Med. (2023) 388:333–43. doi: 10.1056/NEJMoa2203769

PubMed Abstract | Crossref Full Text | Google Scholar

44. Ebert N, Schaeffner E, Seegmiller JC, van Londen M, Bökenkamp A, Cavalier E, et al. Iohexol plasma clearance measurement protocol standardization for adults: a consensus paper of the European Kidney Function Consortium. Kidney Int. (2024) 106:583–96. doi: 10.1016/j.kint.2024.06.029

PubMed Abstract | Crossref Full Text | Google Scholar

45. Miller WG, Kaufman HW, Levey AS, Straseski JA, Wilhelms KW, Yu H-YE, et al. National kidney foundation laboratory engagement working group recommendations for implementing the CKD-EPI 2021 race-free equations for estimated glomerular filtration rate: practical guidance for clinical laboratories. Clin Chem. (2022) 68:511–20. doi: 10.1093/clinchem/hvab278

PubMed Abstract | Crossref Full Text | Google Scholar

46. Jacobson BE, Seccombe DW, Katayev A, and Levin A. A study examining the bias of albumin and albumin/creatinine ratio measurements in urine. Clin Chem Lab Med. (2015) 53:1737–43. doi: 10.1515/cclm-2014-1105

PubMed Abstract | Crossref Full Text | Google Scholar

47. Green AB, Bunk DM, Zhang NF, and Alejo W. Certification of Standard Reference Material^® 2925: Recombinant Human Serum Albumin Solution (Primary Reference Calibrator for Urine Albumin) (Frozen). gaithersburg maryland: NIST (2020). Available online at: https://www.nist.gov/publications/certification-standard-reference-materialr-2925-recombinant-human-serum-albumin.

Google Scholar

48. Jung C-Y and Yoo T-H. Pathophysiologic mechanisms and potential biomarkers in diabetic kidney disease. Diabetes Metab J. (2022) 46:181–97. doi: 10.4093/dmj.2021.0329

PubMed Abstract | Crossref Full Text | Google Scholar

49. Oshima M, Hara A, Toyama T, Jun M, Pollock C, Jardine M, et al. Comparison of circulating biomarkers in predicting diabetic kidney disease progression with autoantibodies to erythropoietin receptor. Kidney Int Rep. (2021) 6:284–95. doi: 10.1016/j.ekir.2020.10.039

PubMed Abstract | Crossref Full Text | Google Scholar

50. Pavkov ME, Nelson RG, Knowler WC, Cheng Y, Krolewski AS, and Niewczas MA. Elevation of circulating TNF receptors 1 and 2 increases the risk of end-stage renal disease in American Indians with type 2 diabetes. Kidney Int. (2015) 87:812–9. doi: 10.1038/ki.2014.330

PubMed Abstract | Crossref Full Text | Google Scholar

51. Niewczas MA, Gohda T, Skupien J, Smiles AM, Walker WH, Rosetti F, et al. Circulating TNF receptors 1 and 2 predict ESRD in type 2 diabetes. J Am Soc Nephrol. (2012) 23:507–15. doi: 10.1681/ASN.2011060627

PubMed Abstract | Crossref Full Text | Google Scholar

52. Gohda T, Niewczas MA, Ficociello LH, Walker WH, Skupien J, Rosetti F, et al. Circulating TNF receptors 1 and 2 predict stage 3 CKD in type 1 diabetes. J Am Soc Nephrol. (2012) 23:516–24. doi: 10.1681/ASN.2011060628

PubMed Abstract | Crossref Full Text | Google Scholar

53. Krolewski AS, Skupien J, Rossing P, and Warram JH. Fast renal decline to ESRD: an unrecognized feature of nephropathy in diabetes. Kidney Int. (2017) 91:1300–11. doi: 10.1016/j.kint.2016.10.046

PubMed Abstract | Crossref Full Text | Google Scholar

54. Nowak N, Skupien J, Niewczas MA, Yamanouchi M, Major M, Croall S, et al. Increased plasma Kidney Injury Molecule-1 suggests early progressive renal decline in non-proteinuric patients with Type 1 diabetes. Kidney Int. (2016) 89:459–67. doi: 10.1038/ki.2015.314

PubMed Abstract | Crossref Full Text | Google Scholar

55. Amatruda JG, Katz R, Sarnak MJ, Gutierrez OM, Greenberg JH, Cushman M, et al. Biomarkers of kidney tubule disease and risk of end-stage kidney disease in persons with diabetes and CKD. Kidney Int Rep. (2022) 7:1514–23. doi: 10.1016/j.ekir.2022.03.033

PubMed Abstract | Crossref Full Text | Google Scholar

56. Chan L, Nadkarni GN, Fleming F, McCullough JR, Connolly P, Mosoyan G, et al. Derivation and validation of a machine learning risk score using biomarker and electronic patient data to predict progression of diabetic kidney disease. Diabetologia. (2021) 64:1504–15. doi: 10.1007/s00125-021-05444-0

PubMed Abstract | Crossref Full Text | Google Scholar

57. Chauhan K, Nadkarni GN, Fleming F, McCullough J, He CJ, Quackenbush J, et al. Initial validation of a machine learning-derived prognostic test (KidneyIntelX) integrating biomarkers and electronic health record data to predict longitudinal kidney outcomes. Kidney360. (2020) 1:731–9. doi: 10.34067/KID.0002252020

PubMed Abstract | Crossref Full Text | Google Scholar

58. Tokita J, Vega A, Sinfield C, Naik N, Rathi S, Martin S, et al. Real world evidence and clinical utility of kidneyIntelX on patients with early-stage diabetic kidney disease: interim results on decision impact and outcomes. J Prim Care Community Health. (2022) 13:21501319221138196. doi: 10.1177/21501319221138196

PubMed Abstract | Crossref Full Text | Google Scholar

59. Nadkarni GN, Stapleton S, Takale D, Edwards K, Moran K, Mosoyan G, et al. Derivation and independent validation of kidneyintelX.dkd: A prognostic test for the assessment of diabetic kidney disease progression. Diabetes Obes Metab. (2023) 25:3779–87. doi: 10.1111/dom.15273

PubMed Abstract | Crossref Full Text | Google Scholar

60. Panduru NM, Sandholm N, Forsblom C, Saraheimo M, Dahlström EH, Thorn LM, et al. Kidney injury molecule-1 and the loss of kidney function in diabetic nephropathy: A likely causal link in patients with type 1 diabetes. Diabetes Care. (2015) 38:1130–7. doi: 10.2337/dc14-2330

PubMed Abstract | Crossref Full Text | Google Scholar

61. Gohda T, Murakoshi M, Shibata T, Suzuki Y, Takemura H, Tsuchiya K, et al. Circulating TNF receptor levels are associated with estimated glomerular filtration rate even in healthy individuals with normal kidney function. Sci Rep. (2024) 14:7245. doi: 10.1038/s41598-024-57265-x

PubMed Abstract | Crossref Full Text | Google Scholar

62. Sen T, Li J, Neuen BL, Neal B, Arnott C, Parikh CR, et al. Effects of the SGLT2 inhibitor canagliflozin on plasma biomarkers TNFR-1, TNFR-2 and KIM-1 in the CANVAS trial. Diabetologia. (2021) 64:2147–58. doi: 10.1007/s00125-021-05512-5

PubMed Abstract | Crossref Full Text | Google Scholar

63. Cho J, Doo SW, Song N, Lee M, Lee H, Kim H, et al. Dapagliflozin reduces urinary kidney injury biomarkers in chronic kidney disease irrespective of albuminuria level. Clin Pharmacol Ther. (2024) 115:1441–9. doi: 10.1002/cpt.3237

PubMed Abstract | Crossref Full Text | Google Scholar

64. McDonnell T, Söderberg M, Taal MW, Vuilleumier N, Kalra PA, and NURTuRE-CKD academic steering group. Plasma and urinary KIM-1 in chronic kidney disease: prognostic value, associations with albuminuria, and implications for kidney failure and mortality. Am J Nephrol. (2025), 1–14. doi: 10.1159/000547867

PubMed Abstract | Crossref Full Text | Google Scholar

65. Collins GS, Reitsma JB, Altman DG, and Moons KGM. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMJ. (2015) 350:g7594. doi: 10.1136/bmj.g7594

PubMed Abstract | Crossref Full Text | Google Scholar

66. Vickers AJ, van Calster B, and Steyerberg EW. A simple, step-by-step guide to interpreting decision curve analysis. Diagn Prognostic Res. (2019) 3:18. doi: 10.1186/s41512-019-0064-7

PubMed Abstract | Crossref Full Text | Google Scholar

67. Efthimiou O, Seo M, Chalkou K, Debray T, Egger M, and Salanti G. Developing clinical prediction models: a step-by-step guide. BMJ. (2024) 386:e078276. doi: 10.1136/bmj-2023-078276

PubMed Abstract | Crossref Full Text | Google Scholar

68. Joshi N, Garapati K, Ghose V, Kandasamy RK, and Pandey A. Recent progress in mass spectrometry-based urinary proteomics. Clin Proteomics. (2024) 21:14. doi: 10.1186/s12014-024-09462-z

PubMed Abstract | Crossref Full Text | Google Scholar

69. Pontillo C and Mischak H. Urinary peptide-based classifier CKD273: towards clinical application in chronic kidney disease. Clin Kidney J. (2017) 10:192–201. doi: 10.1093/ckj/sfx002

PubMed Abstract | Crossref Full Text | Google Scholar

70. Tofte N, Lindhardt M, Adamova K, Bakker SJL, Beige J, Beulens JWJ, et al. Early detection of diabetic kidney disease by urinary proteomics and subsequent intervention with spironolactone to delay progression (PRIORITY): a prospective observational study and embedded randomised placebo-controlled trial. Lancet Diabetes Endocrinol. (2020) 8:301–12. doi: 10.1016/S2213-8587(20)30026-7

PubMed Abstract | Crossref Full Text | Google Scholar

71. Lindhardt M, Persson F, Zürbig P, Stalmach A, Mischak H, de Zeeuw D, et al. Urinary proteomics predict onset of microalbuminuria in normoalbuminuric type 2 diabetic patients, a sub-study of the DIRECT-Protect 2 study. Nephrol Dial Transplant. (2017) 32(11):1866–73. doi: 10.1093/ndt/gfw292

PubMed Abstract | Crossref Full Text | Google Scholar

72. Curovic VR, Eickhoff MK, Rönkkö T, Frimodt-Møller M, Hansen TW, Mischak H, et al. Dapagliflozin improves the urinary proteomic kidney-risk classifier CKD273 in type 2 diabetes with albuminuria: A randomized clinical trial. Diabetes Care. (2022) 45:2662–8. doi: 10.2337/dc22-1157

PubMed Abstract | Crossref Full Text | Google Scholar

73. Cherney D, Perkins BA, Lytvyn Y, Heerspink H, Rodríguez-Ortiz ME, and Mischak H. The effect of sodium/glucose cotransporter 2 (SGLT2) inhibition on the urinary proteome. PloS One. (2017) 12:e0186910. doi: 10.1371/journal.pone.0186910

PubMed Abstract | Crossref Full Text | Google Scholar

74. Good DM, Zürbig P, Argilés À, Bauer HW, Behrens G, Coon JJ, et al. Naturally occurring human urinary peptides for use in diagnosis of chronic kidney disease*. Mol Cell Proteomics. (2010) 9:2424–37. doi: 10.1074/mcp.M110.001917

PubMed Abstract | Crossref Full Text | Google Scholar

75. Mavrogeorgis E, Mischak H, Latosinska A, Siwy J, Jankowski V, and Jankowski J. Reproducibility evaluation of urinary peptide detection using CE-MS. Molecules. (2021) 26:7260. doi: 10.3390/molecules26237260

PubMed Abstract | Crossref Full Text | Google Scholar

76. Yu Y, Mai Y, Zheng Y, and Shi L. Assessing and mitigating batch effects in large-scale omics studies. Genome Biol. (2024) 25:254. doi: 10.1186/s13059-024-03401-9

PubMed Abstract | Crossref Full Text | Google Scholar

77. Schnobrich L and Castrop H. Urinary microvesicles: a window into the kidney. Clin Kidney J. (2025) 18(7):sfaf189. doi: 10.1093/ckj/sfaf189

PubMed Abstract | Crossref Full Text | Google Scholar

78. Erdbrügger U, Blijdorp CJ, Bijnsdorp IV, Borràs FE, Burger D, Bussolati B, et al. Urinary extracellular vesicles: A position paper by the Urine Task Force of the International Society for Extracellular Vesicles. J Extracell Vesicles. (2021) 10:e12093. doi: 10.1002/jev2.12093

PubMed Abstract | Crossref Full Text | Google Scholar

79. Saenz-Pipaon G, Echeverria S, Orbe J, and Roncal C. Urinary extracellular vesicles for diabetic kidney disease diagnosis. J Clin Med. (2021) 10:2046. doi: 10.3390/jcm10102046

PubMed Abstract | Crossref Full Text | Google Scholar

80. Welsh JA, Goberdhan DCI, O’Driscoll L, Buzas EI, Blenkiron C, Bussolati B, et al. Minimal information for studies of extracellular vesicles (MISEV2023): From basic to advanced approaches. J Extracellular Vesicles. (2024) 13:e12404. doi: 10.1002/jev2.12404

PubMed Abstract | Crossref Full Text | Google Scholar

81. Kondapi K, Kumar NL, Moorthy S, and Silambanan S. A study of association of urinary nephrin with albuminuria in patients with diabetic nephropathy. Indian J Nephrol. (2021) 31:142–8. doi: 10.4103/ijn.IJN_305_19

PubMed Abstract | Crossref Full Text | Google Scholar

82. Han L, Wang S, Li J, Zhao L, and Zhou H. Urinary exosomes from patients with diabetic kidney disease induced podocyte apoptosis via microRNA-145-5p/Srgap2 and the RhoA/ROCK pathway. Exp Mol Pathol. (2023) 134:104877. doi: 10.1016/j.yexmp.2023.104877

PubMed Abstract | Crossref Full Text | Google Scholar

83. Srivastava A, Cai X, Lee J, Li W, Larive B, Kendrick C, et al. Kidney functional magnetic resonance imaging and change in eGFR in individuals with CKD. Clin J Am Soc Nephrol. (2020) 15:776–83. doi: 10.2215/CJN.13201019

PubMed Abstract | Crossref Full Text | Google Scholar

84. Mendichovszky I, Pullens P, Dekkers I, Nery F, Bane O, Pohlmann A, et al. Technical recommendations for clinical translation of renal MRI: a consensus project of the Cooperation in Science and Technology Action PARENCHIMA. MAGMA. (2020) 33:131–40. doi: 10.1007/s10334-019-00784-w

PubMed Abstract | Crossref Full Text | Google Scholar

85. Ljimani A, Caroli A, Laustsen C, Francis S, Mendichovszky IA, Bane O, et al. Consensus-based technical recommendations for clinical translation of renal diffusion-weighted MRI. MAGMA. (2020) 33:177–95. doi: 10.1007/s10334-019-00790-y

PubMed Abstract | Crossref Full Text | Google Scholar

86. Mora-Gutiérrez JM, Garcia-Fernandez N, Slon Roblero MF, Páramo JA, Escalada FJ, Wang DJ, et al. Arterial spin labeling MRI is able to detect early hemodynamic changes in diabetic nephropathy. J Magn Reson Imaging. (2017) 46:1810–7. doi: 10.1002/jmri.25717

PubMed Abstract | Crossref Full Text | Google Scholar

87. Selby NM, Blankestijn PJ, Boor P, Combe C, Eckardt K-U, Eikefjord E, et al. Magnetic resonance imaging biomarkers for chronic kidney disease: a position paper from the European Cooperation in Science and Technology Action PARENCHIMA. Nephrol Dial Transplant. (2018) 33:ii4–ii14. doi: 10.1093/ndt/gfy152

PubMed Abstract | Crossref Full Text | Google Scholar

88. Cao H, Ke B, Lin F, Xue Y, and Fang X. Shear wave elastography for assessment of biopsy-proven renal fibrosis: A systematic review and meta-analysis. Ultrasound Med Biol. (2023) 49:1037–48. doi: 10.1016/j.ultrasmedbio.2023.01.003

PubMed Abstract | Crossref Full Text | Google Scholar

89. Friedli I, Baid-Agrawal S, Unwin R, Morell A, Johansson L, and Hockings PD. Magnetic resonance imaging in clinical trials of diabetic kidney disease. J Clin Med. (2023) 12:4625. doi: 10.3390/jcm12144625

PubMed Abstract | Crossref Full Text | Google Scholar

90. Francis ST, Selby NM, and Taal MW. Magnetic resonance imaging to evaluate kidney structure, function, and pathology: moving toward clinical application. Am J Kidney Dis. (2023) 82:491–504. doi: 10.1053/j.ajkd.2023.02.007

PubMed Abstract | Crossref Full Text | Google Scholar

91. van den Brand JAJG, Dijkstra TMH, Wetzels J, Stengel B, Metzger M, Blankestijn PJ, et al. Predicting kidney failure from longitudinal kidney function trajectory: A comparison of models. PloS One. (2019) 14:e0216559. doi: 10.1371/journal.pone.0216559

PubMed Abstract | Crossref Full Text | Google Scholar

92. Kosinski L, Frey E, Klein A, O’Doherty I, Romero K, Stegall M, et al. Longitudinal estimated glomerular filtration rate (eGFR) modeling in long-term renal function to inform clinical trial design in kidney transplantation. Clin Transl Sci. (2023) 16:1680–90. doi: 10.1111/cts.13579

PubMed Abstract | Crossref Full Text | Google Scholar

93. Cherney DZI, Dagogo-Jack S, McGuire DK, Cosentino F, Pratley R, Shih WJ, et al. Kidney outcomes using a sustained ≥40% decline in eGFR: A meta-analysis of SGLT2 inhibitor trials. Clin Cardiol. (2021) 44:1139–43. doi: 10.1002/clc.23665

PubMed Abstract | Crossref Full Text | Google Scholar

94. Aoki J, Kaya C, Khalid O, Kothari T, Silberman MA, Skordis C, et al. CKD progression prediction in a diverse US population: A machine-learning model. Kidney Med. (2023) 5:100692. doi: 10.1016/j.xkme.2023.100692

PubMed Abstract | Crossref Full Text | Google Scholar

95. Fritz BA. Virtual care in nephrology: an in-depth retrospective analysis of outcomes using the reset kidney health model. J Clin Med. (2023) 13:66. doi: 10.3390/jcm13010066

PubMed Abstract | Crossref Full Text | Google Scholar

96. Li L, Astor BC, Lewis J, Hu B, Appel LJ, Lipkowitz MS, et al. Longitudinal progression trajectory of GFR among patients with CKD. Am J Kidney Dis. (2012) 59:504–12. doi: 10.1053/j.ajkd.2011.12.009

PubMed Abstract | Crossref Full Text | Google Scholar

97. Wang J-L and Zhong Q. Joint modeling of longitudinal and survival data. Annu Rev Stat Its Appl. (2025) 12:449–76. doi: 10.1146/annurev-statistics-112723-034334

Crossref Full Text | Google Scholar

98. Komaba H, Imaizumi T, Hamano T, Murotani K, Hasegawa T, Fujii N, et al. Predicting ESKD risk and time to RRT initiation based on past slope and current value of eGFR: the CKD-JAC study: PO2344. J Am Soc Nephrol. (2021) 32:718. doi: 10.1681/ASN.20213210S1718a

Crossref Full Text | Google Scholar

99. Ling J, Ng JKCC, Lau ESH, Ma RCW, Kong APS, Luk AOY, et al. Continuous glucose monitoring metrics in the assessment of glycemia in moderate-to-advanced CKD in diabetes. Kidney Int Rep. (2022) 7:1354–63. doi: 10.1016/j.ekir.2022.03.029

PubMed Abstract | Crossref Full Text | Google Scholar

100. Wakasugi S, Mita T, Katakami N, Okada Y, Yoshii H, Osonoi T, et al. Associations between continuous glucose monitoring-derived metrics and diabetic retinopathy and albuminuria in patients with type 2 diabetes. BMJ Open Diabetes Res Care. (2021) 9:e001923. doi: 10.1136/bmjdrc-2020-001923

PubMed Abstract | Crossref Full Text | Google Scholar

101. Xie D, Yang W, Jepson C, Roy J, Hsu JY, Shou H, et al. Statistical methods for modeling time-updated exposures in cohort studies of chronic kidney disease. Clin J Am Soc Nephrol. (2017) 12:1892–9. doi: 10.2215/CJN.00650117

PubMed Abstract | Crossref Full Text | Google Scholar

102. Zhu Y, Li L, and Huang X. Landmark linear transformation model for dynamic prediction with application to a longitudinal cohort study of chronic disease. J R Stat Soc Ser C Appl Stat. (2019) 68:771–91. doi: 10.1111/rssc.12334

PubMed Abstract | Crossref Full Text | Google Scholar

103. Chesnaye NC, Tripepi G, Dekker FW, Zoccali C, Zwinderman AH, and Jager KJ. An introduction to joint models—applications in nephrology. Clin Kidney J. (2020) 13(2):143–9. doi: 10.1093/ckj/sfaa024

PubMed Abstract | Crossref Full Text | Google Scholar

104. Lyu W, Dong X, Wong R, Zheng S, Abell-Hart K, Wang F, et al. A multimodal transformer: fusing clinical notes with structured EHR data for interpretable in-hospital mortality prediction. AMIA Annu Symp Proc. (2023) 2022:719–28.

PubMed Abstract | Google Scholar

105. Heagerty PJ, Lumley T, and Pepe MS. Time-dependent ROC curves for censored survival data and a diagnostic marker. Biometrics. (2000) 56:337–44. doi: 10.1111/j.0006-341x.2000.00337.x

PubMed Abstract | Crossref Full Text | Google Scholar

106. Bradshaw TJ, Huemann Z, Hu J, and Rahmim A. A guide to cross-validation for artificial intelligence in medical imaging. Radiol Artif Intell. (2023) 5:e220232. doi: 10.1148/ryai.220232

PubMed Abstract | Crossref Full Text | Google Scholar

107. Rieke N, Hancox J, Li W, Milletarì F, Roth HR, Albarqouni S, et al. The future of digital health with federated learning. NPJ Digit Med. (2020) 3:119. doi: 10.1038/s41746-020-00323-1

PubMed Abstract | Crossref Full Text | Google Scholar

108. Li X, Gu Y, Dvornek N, Staib LH, Ventola P, and Duncan JS. Multi-site fMRI analysis using privacy-preserving federated learning and domain adaptation: ABIDE results. Med Image Anal. (2020) 65:101765. doi: 10.1016/j.media.2020.101765

PubMed Abstract | Crossref Full Text | Google Scholar

109. Rajkomar A, Hardt M, Howell MD, Corrado G, and Chin MH. Ensuring fairness in machine learning to advance health equity. Ann Intern Med. (2018) 169:866–72. doi: 10.7326/M18-1990

PubMed Abstract | Crossref Full Text | Google Scholar

110. Davis SE, Greevy RA, Lasko TA, Walsh CG, and Matheny ME. Detection of calibration drift in clinical prediction models to inform model updating. J BioMed Inform. (2020) 112:103611. doi: 10.1016/j.jbi.2020.103611

PubMed Abstract | Crossref Full Text | Google Scholar

111. Tabassi E. Artificial Intelligence Risk Management Framework (AI RMF 1.0). Gaithersburg, MD: NIST (2023). Available online at: https://www.nist.gov/publications/artificial-intelligence-risk-management-framework-ai-rmf-10 (Accessed December 29, 2025).

Google Scholar

112. Tanner KT, Keogh RH, Coupland CAC, Hippisley-Cox J, and Diaz-Ordaz K. Dynamic updating of clinical survival prediction models in a changing environment. Diagn Progn Res. (2023) 7:24. doi: 10.1186/s41512-023-00163-z

PubMed Abstract | Crossref Full Text | Google Scholar

113. Veeranki V and Prasad N. Utilising continuous glucose monitoring for glycemic control in diabetic kidney disease. World J Diabetes. (2024) 15:2006–9. doi: 10.4239/wjd.v15.i10.2006

PubMed Abstract | Crossref Full Text | Google Scholar

114. Fahim YA, Hasani IW, Kabba S, and Ragab WM. Artificial intelligence in healthcare and medicine: clinical applications, therapeutic advances, and future perspectives. Eur J Med Res. (2025) 30:848. doi: 10.1186/s40001-025-03196-w

PubMed Abstract | Crossref Full Text | Google Scholar

115. Jiatuo XU, Tao JIANG, and Shi LIU. Research status and prospect of tongue image diagnosis analysis based on machine learning. Digital Chin Med. (2024) 7:3–12. doi: 10.1016/j.dcmed.2024.04.002

Crossref Full Text | Google Scholar

116. Jiang M, Lou X-E, Zhang X, Nan Q, Gao X, and Liu H. Development and validation of the diagnostic scale of traditional Chinese medicine syndrome elements for diabetic kidney disease. Ann Palliative Med. (2021) 10:122912299–122912299. doi: 10.21037/apm-21-3147

PubMed Abstract | Crossref Full Text | Google Scholar

117. Wang L, Tang K, Wang Y, Zhang P, and Li S. Advancements in artificial intelligence-driven diagnostic models for traditional Chinese medicine. Am J Chin Med. (2025) 53:647–73. doi: 10.1142/S0192415X25500259

PubMed Abstract | Crossref Full Text | Google Scholar

118. Almarshad MA, Islam MS, Al-Ahmadi S, and BaHammam AS. Diagnostic features and potential applications of PPG signal in healthcare: A systematic review. Healthcare (Basel). (2022) 10:547. doi: 10.3390/healthcare10030547

PubMed Abstract | Crossref Full Text | Google Scholar

119. Shahrbabaki SS, Linz D, and Baumert M. Nocturnal pulse wave amplitude attenuations are associated with long-term cardiovascular events. Int J Cardiol. (2023) 385:55–61. doi: 10.1016/j.ijcard.2023.05.047

PubMed Abstract | Crossref Full Text | Google Scholar

120. Qi Z, Tu L, Chen J, Hu X, Xu J, and Zhang Z. The classification of tongue colors with standardized acquisition and ICC profile correction in traditional Chinese medicine. BioMed Res Int. (2016) 2016:3510807. doi: 10.1155/2016/3510807

PubMed Abstract | Crossref Full Text | Google Scholar

121. Charlton PH, Paliakaitė B, Pilt K, Bachler M, Zanelli S, Kulin D, et al. Assessing hemodynamics from the photoplethysmogram to gain insights into vascular age: a review from VascAgeNet. Am J Physiol Heart Circ Physiol. (2022) 322:H493–522. doi: 10.1152/ajpheart.00392.2021

PubMed Abstract | Crossref Full Text | Google Scholar

122. Zhou X, Li C, Su H, Tang Z, Li P, Li Y, et al. Intelligent quality control of traditional chinese medical tongue diagnosis images based on deep learning. Technol Health Care. (2024) 32(S1):207–16. doi: 10.3233/THC-248018

PubMed Abstract | Crossref Full Text | Google Scholar

123. Balasubramaniyan S, Jeyakumar V, and Nachimuthu DS. Panoramic tongue imaging and deep convolutional machine learning model for diabetes diagnosis in humans. Sci Rep. (2022) 12:186. doi: 10.1038/s41598-021-03879-4

PubMed Abstract | Crossref Full Text | Google Scholar

124. Lu L, Lu T, Tian C, and Zhang X. AI: bridging ancient wisdom and modern innovation in traditional Chinese medicine. JMIR Med Inf. (2024) 12:e58491. doi: 10.2196/58491

PubMed Abstract | Crossref Full Text | Google Scholar

125. Inserra F, Forcada P, Castellaro A, and Castellaro C. Chronic kidney disease and arterial stiffness: A two-way path. Front Med. (2021) 8:765924. doi: 10.3389/fmed.2021.765924

PubMed Abstract | Crossref Full Text | Google Scholar

126. Dai S, Guo X, Liu S, Tu L, Hu X, Cui J, et al. Application of intelligent tongue image analysis in Conjunction with microbiomes in the diagnosis of MAFLD. Heliyon. (2024) 10:e29269. doi: 10.1016/j.heliyon.2024.e29269

PubMed Abstract | Crossref Full Text | Google Scholar

127. Cross JL, Choma MA, and Onofrey JA. Bias in medical AI: Implications for clinical decision-making. PloS Digit Health. (2024) 3:e0000651. doi: 10.1371/journal.pdig.0000651

PubMed Abstract | Crossref Full Text | Google Scholar

128. Provenzano M, PuChades MJ, Garofalo C, Jongs N, D’Marco L, Andreucci M, et al. Albuminuria-lowering effect of dapagliflozin, eplerenone, and their combination in patients with chronic kidney disease: A randomized crossover clinical trial. J Am Soc Nephrol. (2022) 33:1569–80. doi: 10.1681/ASN.2022020207

PubMed Abstract | Crossref Full Text | Google Scholar

129. Yepes-Cortés CA, Cardenas-Moreno IC, Daza-Arnedo R, Feriz-Bonelo KM, Yama-Mosquera E, Ramirez-Rincón AH, et al. Combining GLP-1 receptor agonists and SGLT2 inhibitors in type 2 diabetes mellitus: A scoping review and expert insights for clinical practice utilizing the nominal group technique. Diabetes Ther. (2025) 16:813–49. doi: 10.1007/s13300-025-01722-x

PubMed Abstract | Crossref Full Text | Google Scholar

130. Kao Y, Chao T, Chen S, Cheng Y, Chan Y, and Chu P. Initial eGFR changes with SGLT2 inhibitor in patients with type 2 diabetes and associations with the risk of abnormal serum potassium level. J Am Heart Assoc. (2024) 13:e033236. doi: 10.1161/JAHA.123.033236

PubMed Abstract | Crossref Full Text | Google Scholar

131. Goulooze SC, Snelder N, Seelmann A, Horvat-Broecker A, Brinker M, Joseph A, et al. Finerenone dose–exposure–serum potassium response analysis of FIDELIO-DKD phase III: the role of dosing, titration, and inclusion criteria. Clin Pharmacokinet. (2022) 61:451–62. doi: 10.1007/s40262-021-01083-1

PubMed Abstract | Crossref Full Text | Google Scholar

132. Zhao L-M, Zhang M, Zhan Z-L, and Qiu M. Does combination therapy with SGLT2 inhibitors and renin–angiotensin system blockers lead to greater reduction in cardiorenal events among patients with type 2 diabetes? Front Cardiovasc Med. (2021) 8:679124. doi: 10.3389/fcvm.2021.679124

PubMed Abstract | Crossref Full Text | Google Scholar

133. DeFronzo RA. Combination therapy with GLP-1 receptor agonist and SGLT2 inhibitor. Diabetes Obes Metab. (2017) 19:1353–62. doi: 10.1111/dom.12982

PubMed Abstract | Crossref Full Text | Google Scholar

134. Madanchian M and Taherdoost H. Ethical theories, governance models, and strategic frameworks for responsible AI adoption and organizational success. Front Artif Intell. (2025) 8:1619029. doi: 10.3389/frai.2025.1619029

PubMed Abstract | Crossref Full Text | Google Scholar

135. Khosravi M, Zare Z, Mojtabaeian SM, and Izadi R. Artificial intelligence and decision-making in healthcare: A thematic analysis of a systematic review of reviews. Health Serv Res Manag Epidemiol. (2024) 11:23333928241234863. doi: 10.1177/23333928241234863

PubMed Abstract | Crossref Full Text | Google Scholar

136. Ratitch B, Rodriguez-Chavez IR, Dabral A, Fontanari A, Vega J, Onorati F, et al. Considerations for analyzing and interpreting data from biometric monitoring technologies in clinical trials. Digit biomark. (2022) 6:83–97. doi: 10.1159/000525897

PubMed Abstract | Crossref Full Text | Google Scholar