Research progress of multimodal biomarkers in the early diagnosis of mild cognitive impairment in Parkinson’s disease

Parkinson’s disease (PD), as a common neurodegenerative disorder, its associated mild cognitive impairment (PD-mild cognitive impairment, PD-MCI) is a key prodromal stage in the progression to PD-dementia (PDD). Cognitive dysfunction seriously affects the quality of life of patients. Early diagnosis of PD-MCI is crucial for the prognosis of the disease. At present, traditional clinical diagnosis mainly relies on neuropsychological tests, but it has limitations such as low sensitivity and being easily interfered by subjective factors. It is difficult to achieve early and accurate identification of PD-MCI, which greatly affects the intervention timing and prognosis of the disease. This article summarizes the research progress of multimodal biomarkers in the early diagnosis of PD-MCI, mainly including the comprehensive application of neuroimaging biomarkers, humoral biomarkers, genetic and molecular biomarkers, digital biomarkers, and clinical assessment, providing new theoretical basis and technical paths for promoting the early diagnosis of PD-MCI. It is of great clinical significance and social value to assist in the formulation of individualized intervention strategies, delay the progression of diseases to dementia, improve the quality of life of patients, and reduce the medical burden on families and society.

1 Introduction

Mild cognitive impairment (PD-mild cognitive impairment, PD-MCI) in Parkinson’s disease is the prodromal stage of PD-dementia (PDD), and its diagnosis and treatment are crucial for improving the prognosis of patients. Relevant epidemiological research shows that over 80% of PD patients have cognitive function changes, and approximately 30% of them will eventually develop into PDD (1). Between 20 and 33% of patients have PD-MCI at the time of diagnosis of PD, and as many as 57% of PD patients develop PD-MCI successively within 5 years after the first diagnosis (2, 3). In recent years, with the rapid development of big data, artificial intelligence and machine learning technologies, the application of multimodal biomarkers has increasingly gained favor in clinical research, providing powerful tools for the early diagnosis and prognosis of diseases. By aggregating and integrating multi-dimensional data such as neuroimaging, humoral biomarkers, genetics, digital phenotypes, and clinical assessment tools, the pathophysiological changes of PD-MCI can be reflected from different perspectives, providing support for conducting high-quality clinical research and precise diagnosis and treatment. The realization of precise identification of risk factors for PD-MCI, early intervention of the disease and improvement of prognosis provides evidence-based medical evidence, which is conducive to the formulation of effective prevention and treatment strategies in clinical practice. This article reviews the application of multimodal biomarkers in the diagnosis of PD-MCI, explores their different roles in clinical diagnosis, and looks forward to their potential value in the assessment of PD-MCI, aiming to provide a more comprehensive diagnostic strategy for clinical practice and offer references for future research directions (Figure 1).

Figure 1

Figure 1. Classification and techniques of multimodal biomarkers.

2 Neuroimaging biomarkers

Imaging plays a crucial role in the early diagnosis of PD-MCI (Parkinson’s Disease with Mild Cognitive Impairment). Through imaging techniques such as Structural Magnetic Resonance Imaging (sMRI), Functional Magnetic Resonance Imaging (fMRI), Single Photon Emission Computed Tomography (SPECT), and Positron Emission Tomography (PET), researchers can identify structural and functional changes associated with PD-MCI. For instance, MRI can reveal atrophy in brain regions related to motor control, while PET is used to detect changes in dopamine transporter activity in the brain. These imaging biomarkers are not only helpful for early diagnosis but also provide key evidence for monitoring disease progression. Studies have shown that imaging-related indicators, such as changes in the basal ganglia, are closely correlated with the severity of clinical symptoms, making imaging an important tool for assessing PD-MCI patients (4).

2.1 Structural magnetic resonance imaging

Structural Magnetic Resonance Imaging (sMRI), utilizing surface-based morphometry (SBM) techniques, quantifies parameters such as cortical thickness, surface area, volume, and curvature, revealing the characteristic gray matter atrophy and white matter damage in PD-MCI patients. Changes in these parameters are closely related to dysfunction in specific cognitive domains, as reflected by abnormal striatal volume alterations in PD-MCI patients. A multicenter study found that cortical thickness in the frontal lobe (dorsolateral prefrontal cortex, DLPFC) and parietal lobe (inferior parietal cortex) of PD-MCI patients was significantly reduced compared to healthy controls, with the degree of atrophy positively correlating with declines in executive function (e.g., Stroop test) and working memory scores (5). A longitudinal cohort study showed that volume reduction in the temporal lobe (parahippocampal gyrus, entorhinal cortex) independently predicts episodic memory impairment in PD-MCI patients, with an atrophy rate 1.5 times faster than in PD patients without cognitive impairment (6). Compared to early PD patients without MCI, the PD-MCI group exhibited more pronounced white matter microstructural abnormalities. sMRI evidence indicated widespread reductions in white matter fiber tract integrity, along with significantly increased volumes of white matter hyperintensity lesions in periventricular and deep brain regions. Moreover, expansion of white matter hyperintensities in the left frontal and temporal lobes was significantly associated with worsening performance on free recall tests, suggesting that white matter damage in specific brain regions may underlie PD-MCI pathology (7–10).

2.2 Functional magnetic resonance imaging

Resting-state fMRI studies have shown (11) that PD-MCI patients exhibit significant weakening in the functional connectivity (FC) of the default mode network (DMN) between the hippocampus-inferior frontal gyrus, posterior cingulate cortex-posterior parietal lobe, anterior temporal lobe-inferior frontal gyrus, and middle frontal gyrus-middle temporal gyrus. Compared to controls, PD-MCI patients show markedly reduced bilateral prefrontal cortex intra-FC, characterized by decreased connectivity strength between the caudate nucleus and precuneus (12). Besides widespread FC reduction, studies have reported abnormal increases in FC between the midbrain-posterior cingulate cortex and left hippocampus-right cerebellar hemisphere in PD-MCI patients (13, 14). Apart from FC, local consistency (ReHo) values within the DMN positively correlate with cognitive function in PD and continuously decline with disease progression (15). The dynamic FC changes in the DMN, frontoparietal network (FPN), and salience network are more sensitive than structural imaging, with continuous alterations in the salience network showing the strongest association with PD cognitive impairment (PD-CI). Multiple task-based fMRI studies have found (16, 17) that during working memory tasks, PD-MCI patients exhibit significantly reduced activation in bilateral caudate nuclei, right putamen, and dorsolateral/ventrolateral prefrontal cortex, indicating dopaminergic-cortical circuit dysfunction. Additionally, cerebellar functional activity is generally weakened, suggesting damage to cerebello-cortical pathways or their involvement in cognitive regulation.

2.3 Dopaminergic system imaging

Dopamine transporter (DAT) single-photon emission computed tomography (SPECT) demonstrates early dopaminergic dysfunction and its association with cognitive impairment in PD-MCI patients by detecting dopaminergic nerve terminal density in the striatum. A multicenter study showed that DAT binding in the caudate and posterior putamen of PD-MCI patients was 30–40% lower than that of healthy controls, with caudate DAT loss significantly correlated with executive functions (such as Stroop test) and working memory (such as digit span test) scores (18). The Bäckman research team, through longitudinal DAT-SPECT tracking, found that PD patients with an annual striatal DAT binding decline rate exceeding 5% had a 3.2-fold increased risk of progression to MCI within 3 years, indicating that DAT loss could serve as an early predictive biomarker of cognitive decline (19). Serotonin (5-HT) positron emission tomography (PET) utilizes tracers (e.g., [11C] DASB, [18F]altanserin) to visualize serotonin transporter (SERT) or receptor distribution, revealing abnormalities in the raphe 5-HT system and its association with attention and emotional regulation deficits in PD-MCI. Studies show that compared to healthy controls, PD-MCI patients exhibit a 45% decrease in raphe SERT binding, which is positively correlated with Conners’ Continuous Performance Test (CPT) scores. Reduced 5-HT projections from the raphe to the prefrontal cortex may impair attentional control (20). Smith et al.’s longitudinal 5-HT PET study indicated that PD patients with an annual raphe SERT binding decline exceeding 8% have a 2.5-fold increased risk of developing MCI within 2 years (21). The Ballanger team, using the [18F] MPPF tracer, demonstrated elevated 5-HT1A receptor binding in the anterior cingulate cortex of PD-MCI patients, which is associated with depressive symptoms (e.g., BDI scores) and anxiety, suggesting that compensatory receptor upregulation might exacerbate mood disorders (22).

2.4 Metabolism and molecular imaging

Fluorodeoxyglucose positron emission tomography (FDG-PET) reveals the distinctive metabolic patterns of PD-MCI (Parkinson’s Disease Mild Cognitive Impairment) patients by measuring cerebral glucose metabolism rate, which is associated with impairments in specific cognitive domains. Huang et al.’s study demonstrated significantly reduced glucose metabolism in the prefrontal cortex (especially the dorsolateral prefrontal cortex, DLPFC) and the inferior parietal lobule in PD-MCI patients, with the degree of metabolic decline positively correlating with executive function scores (e.g., Stroop test, Digit Symbol Substitution Test) (23). Tau positron emission tomography (Tau-PET), employing specific tracers such as [18F]AV-1451 and [18F]MK-6240, visualizes tau protein deposits, illustrating the spatial characteristics of tau pathology in PD-MCI and their relationship to cognitive decline. Schöll et al. found that tau deposition in the midbrain positively correlates with dopaminergic neuron loss assessed by DaT-SPECT, suggesting the involvement of tau pathology in degeneration of the nigrostriatal pathway (24). Ossenkoppele et al. proposed that tau deposition in PD-MCI primarily occurs in the midbrain and limbic system, whereas in Alzheimer’s disease (AD) it is predominantly distributed in the temporoparietal lobes and neocortex, supporting specificity in pathological distribution (25).

3 Fluid biomarkers

Fluid biomarkers are crucial for the early identification of PD-MCI. Specific biochemical substances detected in blood and cerebrospinal fluid, such as neuron-specific enolase (NSE) and neurofilament light chain (NfL), are considered closely related to the degree of neuronal damage and stages of disease progression. The detection of these biochemical markers not only aids in identifying early PD-MCI patients but also serves to monitor disease progression and evaluate therapeutic efficacy (26). Due to their diversity and sensitivity, biochemical biomarkers hold broad application potential in clinical practice. Research on biomarkers continues to advance, and novel biomarkers, such as changes in $\alpha$ -synuclein and tau protein, have been confirmed to be associated with PD-MCI, offering new avenues for early clinical diagnosis (27).

3.1 Cerebrospinal fluid

In patients with PD-MCI, changes in cerebrospinal fluid (CSF) $\alpha$ -synuclein levels reflect the association between pathological progression and cognitive impairment. Numerous studies have shown that total $\alpha$ -synuclein levels in the CSF of PD-MCI patients are significantly lower than those in healthy controls and PD patients without cognitive impairment. Kang et al. found that decreased CSF $\alpha$ -synuclein levels in PD-MCI patients are related to midbrain atrophy and weakened functional connectivity of the default mode network (DMN), suggesting it may serve as a surrogate marker for neuronal loss (28). Parnetti et al.’s review indicates that reduced CSF $\mathrm{\alpha }$ -synuclein is associated with degenerative changes in the frontostriatal pathway in PD-MCI patients, possibly affecting executive function through synaptic dysfunction (29). Unlike total $\mathrm{\alpha }$ -synuclein, its oligomeric form (oligomeric $\alpha$ -syn) is significantly elevated in PD-MCI patients and correlates with cognitive deterioration. Shahnawaz et al. developed a novel detection method and found that CSF oligomeric $\mathrm{\alpha }$ -syn levels in PD-MCI patients are three times higher than those in healthy controls and negatively correlated with Montreal Cognitive Assessment (MoCA) scores (30). Abnormal phosphorylation of Tau protein (such as p-tau181, p-tau231) is a typical marker of Alzheimer’s disease (AD), and elevated CSF Tau levels are also observed in PD-MCI patients; however, its pathological distribution and clinical significance differ from those in AD-MCI. Alves et al. found that CSF p-tau181 levels in PD-MCI patients are significantly higher than in PD patients without cognitive impairment, but the increase is less pronounced compared to AD-MCI patients and lacks strong correlation with hippocampal atrophy, suggesting that Tau pathology in PD-MCI may be confined to subcortical regions (such as the brainstem and limbic system) (31). Hall et al. demonstrated through proteomic analysis that the Tau phosphorylation patterns (such as p-tau231) in PD-MCI patients differ from those in AD-MCI and may contribute to synaptic toxicity formation together with $\mathrm{\alpha }$ -synuclein (32).

3.2 Blood

The levels of α-synuclein and synaptic proteins such as GAP43 and SNAP25 in blood exosomes change and can serve as non-invasive biomarkers. Abnormal oligomers of α-synuclein can spread between neurons via exosomes, accelerating pathological propagation. Stuendl et al. observed a significant increase of α-synuclein oligomers in the blood exosomes of PD-MCI patients, which positively correlated with oligomer concentrations in cerebrospinal fluid (CSF), indicating it can replace CSF testing (33). Shi et al. used single molecule array (Simoa) technology to show that exosomal α-synuclein oligomer levels were significantly associated with reduced executive function scores (e.g., Stroop test) in PD-MCI patients (34). Mechanistic studies by Chung et al. demonstrated that exosome-mediated α-synuclein propagation triggers microglial inflammatory responses, leading to synaptic loss in the frontal cortex and consequent cognitive impairment (35). Growth-associated protein 43 (GAP43) and synaptosomal-associated protein 25 (SNAP25) are key proteins for synaptic plasticity and neurotransmitter release; changes in their exosomal levels reflect synaptic integrity. Svenningsson et al. reported a 40% decrease in GAP43 content in blood exosomes of PD-MCI patients compared to healthy controls, correlating with hippocampal volume reduction (36). Bacioglu et al. found significantly reduced exosomal SNAP25 in PD-MCI patients, associated with weakened functional connectivity in the default mode network (DMN), indicating decreased synaptic transmission efficiency (37). Chen et al., combining exosomal proteomics with structural MRI, identified that GAP43/SNAP25 levels and frontal cortex thickness together predict the risk of PD-MCI progressing to dementia (38).

Neurofilament Light Chain (NfL) is a cytoskeletal protein released following neuronal axonal damage. Elevated plasma $\mathrm{NfL}$ levels reflect the activity of neurodegeneration. Backstrm and colleagues, after a 5-year follow-up, found that initial plasma NfL levels in PD-MCI patients were higher than in PD patients without cognitive impairment, and $\mathrm{NfL}$ for each increase of $10\mathrm{pg}/\mathrm{mL}$ , the risk of dementia conversion increased 2.3-fold (39). Byrne et al. demonstrated that combining plasma NfL with midbrain atrophy MRI can predict the rate of cognitive decline in PD-MCI patients (40). The rate of white matter atrophy and ventricular enlargement are also considered key early predictive factors for MCI, further indicating that plasma NfL combined with midbrain atrophy MRI has potential in predicting the rate of cognitive decline in PD-MCI patients. Hansson et al. through autopsy research, found that plasma $\mathrm{NfL}$ levels positively correlate with the degree of nigral dopaminergic neuron loss and $\mathrm{\alpha }$ -synuclein pathological burden in the brains of $\mathrm{PD}$ - $\mathrm{MCI}$ patients (41).

3.3 Saliva/urine

The amount of salivary α-synuclein is positively correlated with its concentration in CSF (cerebrospinal fluid), demonstrating potential screening value. Vivacqua et al. showed that the content of $\mathrm{\alpha }$ -synuclein oligomers in the saliva of PD (Parkinson’s disease) patients is significantly positively correlated with its concentration in CSF, and that PD-MCI (Parkinson’s disease with mild cognitive impairment) patients have higher salivary oligomer levels than PD patients without cognitive impairment (42). Stewart et al. used ultrasensitive single molecule detection technology (Simoa) to prove that the total α-synuclein level in saliva highly matches the total $\mathrm{\alpha }$ -synuclein concentration in CSF, and that salivary $\mathrm{\alpha }$ -synuclein is reduced by approximately 30% in PD-MCI patients compared to healthy controls (43). Kang et al.’s longitudinal cohort study showed that PD patients with high baseline saliva $\mathrm{\alpha }$ -synuclein levels are more likely to develop MCI within 2 years, demonstrating its predictive role (44). In PD-MCI patients, oxidative stress participates in cognitive decline by damaging neurons and synaptic function; elevated urinary 8-hydroxy-2′-deoxyguanosine (8-OHdG) levels can serve as a non-invasive pathological marker. A cross-sectional study reported that urinary 8-OHdG levels are approximately 2.5 times higher in PD-MCI patients than in healthy controls, and are significantly negatively correlated with Montreal Cognitive Assessment (MoCA) scores; patients with high 8-OHdG levels have lower executive function and memory scores (45). A five-year cohort study showed that PD patients with baseline urinary 8-OHdG levels above the median have a 2.8-fold increased risk of progression to MCI, and the rate of increase in 8-OHdG levels positively correlates with the rate of cognitive decline (46). At present, there are relatively few studies on saliva/urine biomarkers of PD-MCI. Subsequently, it is necessary to optimize the detection technology, standardize the research process, and deeply explore its value in the early diagnosis of the disease, disease monitoring and mechanism interpretation.

3.4 Skin/sweat

$\mathrm{\alpha }$ -synuclein abnormal deposition in cutaneous nerve endings and sweat glands is a core pathological feature of PD, and its quantitative detection provides a non-invasive method for early diagnosis of PD-MCI. Donadio et al. found through skin biopsy that the positivity rate of phosphorylated $\mathrm{\alpha }$ -synuclein (p- $\mathrm{\alpha }$ -syn) deposition in the epidermal small nerve fibers of PD patients reached 90%, and the deposition density in PD-MCI patients was significantly higher than that in PD patients without cognitive impairment. The deposition density was negatively correlated with the MoCA score, indicating its association with cognitive decline (47). Gibbons et al. quantitatively analyzed $\mathrm{\alpha }$ -synuclein aggregates around sweat glands using confocal microscopy and found that the aggregate load in PD-MCI patients was three times that of healthy controls and positively correlated with CSF $\mathrm{\alpha }$ -synuclein oligomer levels (48). Wang’s team multicenter study showed that skin p-α-syn detection distinguished PD-MCI from healthy controls with an AUC of 0.92, outperforming traditional blood biomarkers such as NfL (49).

4 Genetic biomarkers

4.1 Genetics

Cohort studies by Sidransky et al. showed that PD patients carrying the heterozygous GBA N370S mutation have a 2.8-fold increased risk of developing MCI compared to non-carriers, with L444P mutation carriers exhibiting an even higher risk (50). Polymeropoulos et al. found that 80% of PD patients carrying the SNCA A53T mutation progressed to MCI within 3 years after diagnosis, significantly higher than non-carriers (51). Kiely et al. reported that the SNCA E46K mutation accelerates $\mathrm{\alpha }$ -synuclein fibril formation, which is associated with temporal cortical atrophy and memory impairment (52). Irwin et al. found that PD-MCI patients carrying APOE $\epsilon 4$ exhibit significantly elevated cerebrospinal fluid p-tau181 levels, indicating aggravated tau pathology (53). Zhang et al.’s fMRI study showed that the Val/Val genotype weakens functional connectivity between the DLPFC and the striatum, exacerbating deficits in cognitive flexibility (54).

4.2 Epigenetics

Abnormal methylation of genes such as HDAC4 and HOXC5 is present in the peripheral blood of PD-MCI patients and is associated with neuroinflammation and synaptic plasticity. Studies have found hypermethylation in the promoter region of HDAC4 in the peripheral blood of PD-MCI patients, leading to decreased expression, which is also linked to reduced synaptophysin in the frontal cortex. Low expression of HDAC4 impairs synaptic plasticity by inhibiting the BDNF (brain-derived neurotrophic factor) signaling pathway (55). Another study using epigenome-wide association study (EWAS) showed hypomethylation in the gene body of HOXC5 in PD-MCI patients, resulting in increased expression. Overexpression of HOXC5 promotes peripheral monocytes to release IL-6 and TNF- $\mathrm{\alpha }$ , exacerbating blood–brain barrier permeability and central inflammation (56). Abnormal expression of miR-124 and miR-132 regulates tau phosphorylation and synaptic function. Relevant studies found a 50% reduction of miR-124 levels in the plasma of PD-MCI patients compared with controls, with a negative correlation between its expression and CSF p-tau181. Experimental evidence shows that overexpression of miR-124 can inhibit GSK-3β activity, thereby reducing $\mathrm{tau}$ phosphorylation (57). Additionally, studies have found significantly reduced miR-132 levels in exosomes from PD-MCI patients, with its expression positively correlated with hippocampal volume and memory scores. Overexpression of miR-132 can rescue synaptic plasticity deficits (58).

5 Digital biomarkers

Digital biomarkers leverage technologies such as wearable devices, voice analysis, and cognitive task interactions to collect real-time, non-invasive physiological, behavioral, and environmental multidimensional data, greatly enhancing early diagnostic accuracy for PD-MCI. Research shows that digital features based on gait dynamics (such as stride variability) and micro-expression recognition can precisely capture neurodegenerative changes in the prodromal phase of PD-MCI (59). Moreover, smartphone applications (e.g., “PD-Brain”) can diagnose risks of hippocampal and prefrontal cortex functional decline by real-time monitoring of cognitive-motor dual tasks (60). Longitudinal studies indicate that baseline speech articulation deficits in PD patients (e.g., prolonged consonant-vowel transition time) can predict the risk of developing MCI within 2 years, with sensitivity comparable to hippocampal volume reduction (61). These technological advances overcome the subjective limitations of traditional neuropsychological scales, providing objective and dynamic quantitative tools for precise stratification and intervention in PD-MCI.

6 Clinical and neuropsychological assessment

Clinical assessment indicators, as a key part of early diagnosis of PD-MCI (Parkinson’s Disease Mild Cognitive Impairment), are typically implemented using standardized clinical evaluation tools. These tools include the Unified Parkinson’s Disease Rating Scale (UPDRS), Montreal Cognitive Assessment (MoCA), and Stroop Color-Word Test, which effectively assess patients’ motor function and cognitive status. Research shows that clinical assessment indicators have significant correlations with imaging and biochemical biomarkers, providing critical information on disease progression. For example, integrating UPDRS scores with imaging results can more accurately diagnose patients’ disease progression and quality of life (62). By combining clinical assessments with other biomarker information, physicians can develop personalized treatment plans better suited to patients’ needs to improve their quality of life.

7 Multimodal integration models

The application of machine learning algorithms in the medical field is becoming increasingly widespread, especially in the early prediction of diseases. Common machine learning algorithms include categories such as Support Vector Machine (SVM), Random Forest (RF), Gradient Boosting Tree (GBM), and deep learning models. These algorithms identify biomarkers and potential therapeutic targets of PD-MCI by analyzing different dimensions of the data, and provide new ideas for the comprehensive diagnosis and personalized treatment of PD-MCI. As algorithms continue to be optimized and datasets become increasingly rich, the accuracy and reliability of early machine learning predictions are constantly improving. Machine learning helps identify PD-MCI patients by analyzing specific proteins or molecules in samples such as blood and cerebrospinal fluid. Zhang et al. constructed support vector machine (SVM), decision tree (DT), and logistic regression (LR) models based on multimodal markers of quantitative electroencephalogram (qEEG) and structural magnetic resonance imaging (sMRI). The study found that among 23 pairs of PD-MCI and non-dementia Parkinson’s disease (PD-ND) patients matched by propensity scores, qEEG features are superior to sMRI features in terms of prediction accuracy and area under the curve (AUC). Among them, the theta wave in lead P3 of qEEG features has a significant impact on the classification model, suggesting that multimodal “composite markers” have potential value for individualized diagnosis of PD-MCI (63). Zhu et al. integrated clinical features, plasma markers, functional connectivity of the default mode network and cortical thickness to construct a model and found that the plasma phosphorylated tau181 level and the ratio of phosphorylated tau181/β-amyloid protein 1–42 in the PD-MCI group were significantly higher than those in the PD with normal cognition (PD-NC) group. Moreover, in terms of functional connectivity, the connection between the left posterior cingulate cortex and the left parahippocampal gyrus in the PD-MCI group was enhanced (64). Andrew et al. found that by using support vector machine (SVM) to classify clinical data, T1-weighted magnetic resonance imaging (MRI), resting-state functional MRI (Rs-fMRI), and plasma nerve fiber light chain (NfL) and glial fibrillary acidic protein (GFAP) level data, the results showed that The classifier combining clinical data, Rs-fMRI and NfL biomarkers performed the best, highlighting the importance of multimodal data fusion in improving the predictive ability of PD-MCI (65). A study analyzed the cognitive function assessment of MCI patients using machine learning algorithms. The results showed that a multimodal model integrating multi-source data biomarkers such as neuroimaging, humoral biomarkers, and clinical features demonstrated higher sensitivity and specificity in predicting cognitive decline in patients compared to single biomarkers or traditional diagnostic methods. It has laid a solid foundation for achieving precise and early diagnosis of PD-MCI and promoting the implementation of individualized intervention strategies (66).

8 Conclusion

PD-MCI is an intermediate transformation stage from normal cognition to PDD. As the duration of the disease increases, patients gradually lose their autonomy, increasing the mental and economic burden on society and families. To improve the prognosis of patients with PD-MCI, enhance the quality of life and reduce the social burden, the early identification, diagnosis and treatment of PD-MCI are of vital importance. However, there are still difficulties in the early and accurate identification of PD-MCI. This study integrates multi-dimensional data of multimodal biomarkers, including imaging biomarkers, humoral biomarkers, genetic and molecular biomarkers, digital biomarkers, and clinical evaluations. With the help of machine learning algorithms to mine potential disease patterns, it can more comprehensively assess the patient’s condition and improve the sensitivity and specificity of early diagnosis of PD-MCI. Provide quantitative basis for individualized risk stratification.

Although multimodal biomarkers have shown great potential in improving the accuracy, efficiency and personalization of the early diagnosis of PD-MCI, there are still many bottlenecks that need to be urgently broken through in their clinical transformation: (1) The cognitive barriers and trust deficits of clinicians toward emerging technologies stem from the insufficient update of interdisciplinary knowledge; (2) The update of clinical practice guidelines lags behind the progress of scientific research, and there is a lack of standardized procedures for interpreting multimodal data. (3) Insufficient allocation of interdisciplinary research resources has restricted the efficiency of large-scale cohort validation and technology transfer. Multimodal biomarkers still need to be further optimized and validated through clinical tests. Future research directions include: improving data quality on the basis of multimodal data fusion, constructing more generalized integrated models based on neurophysiological and pathological mechanisms, building larger-scale multi-center standardized databases, and promoting clinical transformation and precision diagnosis and treatment through industry-university-research collaboration. By continuously promoting the development of multimodal biomarkers in the diagnosis of PD-MCI, it is expected that in the future, through strengthening interdisciplinary cooperation, such as close collaboration among clinicians, data scientists and biomedical researchers, relying on large-scale cohort validation and standardized processes to promote the clinical application of multimodal biomarkers, more accurate diagnosis and more effective treatment can be provided for PD-MCI patients. In addition, with the continuous advancement of technology, the application of artificial intelligence and machine learning will also play a key role in this process. Utilizing artificial intelligence for multimodal data analysis and pattern recognition can accurately identify disease states and patient needs, providing support for the optimization of treatment plans. In conclusion, although there are many challenges at present, with the deepening of research and technological innovation, the application prospects of multimodal biomarkers in the diagnosis of PD-MCI will be very broad.

Author contributions

HG: Funding acquisition, Writing – original draft, Writing – review & editing. XS: Funding acquisition, Writing – review & editing. ML: Funding acquisition, Writing – original draft. GZ: Funding acquisition, Writing – original draft. CC: Funding acquisition, Writing – original draft. BX: Funding acquisition, Writing – original draft. YX: Funding acquisition, Writing – original draft, Writing – review & editing.

Funding

The author(s) declare that financial support was received for the research and/or publication of this article. This work was supported by the National Natural Science Foundation of China Project (81303011); Key Project jointly supported by the National Health Commission and Provincial Ministries (S BGJ202102187); Henan Province Key Scientific and Technological Tackling Project (242102311250); Henan Province “Double First-Class” Establishment Engineering Projects (No. HSRP-DFCTCM-2023-8-27, HSRP-DFCTCM-2023-8-44).

Conflict of interest

The authors declare that the research 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 authors declare that no Gen AI was used in the creation of this manuscript.

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.

References

1. Monastero, R, Cicero, CE, Baschi, R, Davì, M, Luca, A, Restivo, V, et al. Mild cogni-tive impairment in Parkinson's disease: the Parkinson's diseasecognitive study(PACOS). J Neurol. (2018) 265:1050–8. doi: 10.1007/s00415-018-8800-4

PubMed Abstract | Crossref Full Text | Google Scholar

2. Lawson, RA, Yarnall, AJ, Duncan, GW, Breen, DP, Khoo, TK, Williams-Gray, CH, et al. Stability of mild cognitiveimpairment in newly diagnosed Parkinson's disease. J Neurol Neurosurg Psychiatry. (2017) 88:648–52. doi: 10.1136/jnnp-2016-315099

Crossref Full Text | Google Scholar

3. Goldman, JG, Vernaleo, ВA, Camicioli, R, Dahodwala, N, Dobkin, RD, Ellis, T, et al. Cognitive impairment inParkinson's disease:a report from a multidisciplinary symposium onunmet needs and future directions to maintain cognitive health. NPJ Parkinsons Dis. (2018) 4:19. doi: 10.1038/s41531-018-0055-3

Crossref Full Text | Google Scholar

4. Liang, J, Li, ZW, Sun, ZN, Bi, Y, Cheng, H, Zeng, T, et al. Latent space search based multimodal optimization with personalized edge-network biomarker for multi-purpose early disease prediction. Brief Bioinform. (2023) 24:bbad364. doi: 10.1093/bib/bbad364

PubMed Abstract | Crossref Full Text | Google Scholar

5. Rong, S, Li, Y, Li, B, Nie, K, Zhang, P, Cai, T, et al. Meynert nucleus-related cortical thinning in Parkinson's disease with mild cognitive impairment. Quant Imaging Med Surg. (2021) 11:1554–66. doi: 10.21037/qims-20-444

PubMed Abstract | Crossref Full Text | Google Scholar

6. Xu, H, Liu, Y, Wang, L, Zeng, X, Xu, Y, and Wang, Z. Role of hippocampal subfields in neurodegenerative disease progression analyzed with a multi-scale attention-based network. Neuroimage Clin. (2023) 38:103370. doi: 10.1016/j.nicl.2023.103370

PubMed Abstract | Crossref Full Text | Google Scholar

7. Choi, SA, Evidente, VG, Caviness, JN, Shill, HA, Sabbagh, MN, Connor, DJ, et al. Are there differences in cerebral white matter lesion burdens between Parkinson's disease patients with or without dementia? Acta Neuropathol. (2020) 119:147–9. doi: 10.1007/s00401-009-0620-2

Crossref Full Text | Google Scholar

8. Zhu, Y, Yang, B, Zhou, C, Gao, C, Hu, Y, Yin, WF, et al. Cortical atrophy is associated with cognitive impairment in Parkinson's disease: a combined analysis of cortical thickness and functional connectivity. Brain Imaging Behav. (2022) 16:2586–600. doi: 10.1007/s11682-022-00714-w

PubMed Abstract | Crossref Full Text | Google Scholar

9. Zhao, W, Cheng, B, Zhu, T, Cui, Y, Shen, Y, Fu, X, et al. Effects of white matter hyperintensity on cognitive function in PD patients: a meta-analysis. Front Neurol. (2023) 14:1203311. doi: 10.3389/fneur.2023.1203311

PubMed Abstract | Crossref Full Text | Google Scholar

10. Pu, W, Shen, X, Huang, M, Li, Z, Zeng, X, Wang, R, et al. Assessment of white matter lesions in Parkinson's disease: voxelbased analysis and tract-based spatial statistics analysis of Parkinson's disease with mild cognitive impairment. Curr Neurovasc Res. (2020) 17:480–6. doi: 10.2174/1567202617666200901181842

PubMed Abstract | Crossref Full Text | Google Scholar

11. Hou, Y, Yang, J, Luo, C, Song, W, Ou, R, Liu, W, et al. Dysfunction of the default mode network in drug-naïve Parkinson's disease with mild cognitive impairments: a resting-state fMRI study. Front Aging Neurosci. (2016) 8:247. doi: 10.3389/fnagi.2016.00247

PubMed Abstract | Crossref Full Text | Google Scholar

12. Amboni, M, Tessitore, A, Esposito, F, Santangelo, G, Picillo, M, Vitale, C, et al. Restingstate functional connectivity associated with mild cognitive impairment in Parkinson's disease. J Neurol. (2015) 262:425–34. doi: 10.1007/s00415-014-7591-5

PubMed Abstract | Crossref Full Text | Google Scholar

13. Owens-Walton, C, Jakabek, D, Power, BD, Walterfang, M, Hall, S, van Westen, D, et al. Structural and functional neuroimaging changes associated with cognitive impairment and dementia in Parkinson's disease. Psychiatry Res Neuroimaging. (2021) 312:111273. doi: 10.1016/j.pscychresns.2021.111273

PubMed Abstract | Crossref Full Text | Google Scholar

14. Lang, S, Yoon, EJ, Kibreab, M, Kathol, I, Cheetham, J, Hammer, T, et al. Mild behavioral impairment in Parkinson's disease is associated with altered corticostriatal connectivity. NeuroImage: Clinical. (2020) 26:102252. doi: 10.1016/j.nicl.2020.102252

PubMed Abstract | Crossref Full Text | Google Scholar

15. Guo, W, Jin, W, Li, N, Gao, J, Wang, J, Chang, Y, et al. Brain activity alterations in patients with Parkinson's disease with cognitive impairment based on resting-state functional MRI. Neurosci Lett. (2021) 747:135672. doi: 10.1016/j.neulet.2021.135672

Crossref Full Text | Google Scholar

16. Ay, U, and Gürvit, IH. Alterations in large-scale intrinsic connectivity networks in the Parkinson's diseaseassociated cognitive impairment continuum: a systematic review. Noro Psikiyatr Ars. (2022) 59:S57–66. doi: 10.29399/npa.28209

PubMed Abstract | Crossref Full Text | Google Scholar

17. Putcha, D, Ross, RS, Cronin-Golomb, A, Janes, AC, and Stern, CE. Salience and default mode network coupling predicts cognition in aging and Parkinson's disease. J Int Neuropsychol Soc. (2016) 22:205–15. doi: 10.1017/S1355617715000892

PubMed Abstract | Crossref Full Text | Google Scholar

18. Demailly, A, Moreau, C, and Devos, D. Effectiveness of continuous dopaminergic therapies in Parkinson's disease: a review of L-DOPA pharmacokinetics/pharmacodynamics. J Parkinsons Dis. (2024) 14:925–39. doi: 10.3233/JPD-230372

PubMed Abstract | Crossref Full Text | Google Scholar

19. Seo, S, Yoon, YJ, Lee, S, Lim, H, Choo, K, Kim, D, et al. Striatal dopamine transporter uptake predicts neuronal hypometabolism and visuospatial function in Parkinson's disease. Eur J Nucl Med Mol Imaging. (2025) 52:2307–16. doi: 10.1007/s00259-025-07137-x

PubMed Abstract | Crossref Full Text | Google Scholar

20. Lee, Y, Ha, S, Yoo, SW, Joo, HO, Ie, RY, and Joong-Seok, K. Decreased raphe connectivity preceding changes of raphe-regional perfusion and SERT expression in drug naïve PD patients with depressive mood. J Nucl Med. (2022) 63:2948

Google Scholar

21. Frouni, I, Kwan, C, Belliveau, S, and Huot, P. Cognition and serotonin in Parkinson's disease. Prog Brain Res. (2022) 269:373–403. doi: 10.1016/bs.pbr.2022.01.013

Crossref Full Text | Google Scholar

22. Smith, GS, Kuwabara, H, Yan, H, Nassery, N, Yoon, M, Kamath, V, et al. Serotonin degeneration and amyloid-β deposition in mild cognitive impairment: relationship to cognitive deficits. J Alzheimer's Dis. (2023) 96:215–27. doi: 10.3233/JAD-230570

PubMed Abstract | Crossref Full Text | Google Scholar

23. Zhihui, S, Yinghua, L, Hongguang, Z, Yuyin, D, Xiaoxiao, D, Lulu, G, et al. Correlation analysis between 18F-fluorodeoxyglucose positron emission tomography and cognitive function in first diagnosed Parkinson’s disease patients. Front Neurol. (2023) 14:1195576. doi: 10.3389/fneur.2023.1195576

PubMed Abstract | Crossref Full Text | Google Scholar

24. Devignes, Q, Lopes, R, and Dujardin, K. Neuroimaging outcomes associated with mild cognitive impairment subtypes in Parkinson's disease: a systematic review. Parkinsonism Relat Disord. (2022) 95:122–37. doi: 10.1016/j.parkreldis.2022.02.006

PubMed Abstract | Crossref Full Text | Google Scholar

25. Zhang, J, Jin, J, Su, D, Feng, T, and Zhao, H. Tau-PET imaging in Parkinson's disease: a systematic review and meta-analysis. Front Neurol. (2023) 14:1145939. doi: 10.3389/fneur.2023.1145939

PubMed Abstract | Crossref Full Text | Google Scholar

26. Park, B, Kim, Y, Park, J, Choi, H, Kim, SE, Ryu, H, et al. Integrating biomarkers from virtual reality and magnetic resonance imaging for the early detection of mild cognitive impairment using a multimodal learning approach: validation study. J Med Internet Res. (2024) 26:e54538. doi: 10.2196/54538

PubMed Abstract | Crossref Full Text | Google Scholar

27. Galper, J, Dean, NJ, Pickford, R, Lewis, SJG, Halliday, GM, Kim, WS, et al. Lipid pathway dysfunction is prevalent in patients with Parkinson's disease. Brain. (2022) 145:3472–87. doi: 10.1093/brain/awac176

PubMed Abstract | Crossref Full Text | Google Scholar

28. Fan, TS, Liu, SCH, and Wu, RM. Alpha-synuclein and cognitive decline in Parkinson disease. Life. (2021) 11:1239. doi: 10.3390/life11111239

PubMed Abstract | Crossref Full Text | Google Scholar

29. Papaliagkas, V, Kalinderi, K, Vareltzis, P, Moraitou, D, Papamitsou, T, and Chatzidimitriou, M. CSF biomarkers in the early diagnosis of mild cognitive impairment and Alzheimer’s disease. Int J Mol Sci. (2023) 24:8976. doi: 10.3390/ijms24108976

PubMed Abstract | Crossref Full Text | Google Scholar

30. Brockmann, K, Lerche, S, Baiardi, S, Rossi, M, Wurster, I, Quadalti, C, et al. CSF α-synuclein seed amplification kinetic profiles are associated with cognitive decline in Parkinson’s disease. NPJ Parkinsons Dis. (2024) 10:24. doi: 10.1038/s41531-023-00627-5

PubMed Abstract | Crossref Full Text | Google Scholar

31. van der Gaag, BL, Deshayes, NAC, Breve, JJP, Bol, JGJM, Jonker, AJ, Hoozemans, JJM, et al. Distinct tau and alpha-synuclein molecular signatures in Alzheimer’s disease with and without Lewy bodies and Parkinson’s disease with dementia. Acta Neuropathol. (2024) 147:14. doi: 10.1007/s00401-023-02657-y

PubMed Abstract | Crossref Full Text | Google Scholar

32. Zhang, N, Tang, C, Ma, Q, Wang, W, Shi, M, Zhou, X, et al. Comprehensive serum metabolic and proteomic characterization on cognitive dysfunction in Parkinson’s disease. Ann Transl Med. (2021) 9:559–9. doi: 10.21037/atm-20-4583

PubMed Abstract | Crossref Full Text | Google Scholar

33. Chung, CC, Chan, L, Chen, JH, Chung, C‐C, Chen, J‐H, Bamodu, OA, et al. Plasma extracellular vesicles tau and β-amyloid as biomarkers of cognitive dysfunction of Parkinson's disease. FASEB J. (2021) 35:e21895. doi: 10.1096/fj.202100787R

PubMed Abstract | Crossref Full Text | Google Scholar

34. Yan, YQ, Pu, JL, Zheng, R, Yan, Y‐Q, Pu, J‐L, Fang, Y, et al. Different patterns of exosomal α-synuclein between Parkinson's disease and probable rapid eye movement sleep behavior disorder. Eur J Neurol. (2022) 29:3590–9. doi: 10.1111/ene.15537

Crossref Full Text | Google Scholar

35. Dai, M, Yan, L, Yu, H, Chen, C, and Xie, Y. TNFRSF10B is involved in motor dysfunction in Parkinson's disease by regulating exosomal α-synuclein secretion from microglia. J Chem Neuroanat. (2023) 129:102249. doi: 10.1016/j.jchemneu.2023.102249

Crossref Full Text | Google Scholar

36. Chen, J, Liang, C, Wang, F, Zhu, Y, Zhu, L, Chen, J, et al. Potential biofluid markers for cognitive impairment in Parkinson’s disease. Neural Regen Res. (2025) 21:281–95. doi: 10.4103/NRR.NRR-D-24-00592

Crossref Full Text | Google Scholar

37. Wang, W, Gao, W, Zhang, L, Xia, Z, and Zhao, B. SNAP25 ameliorates postoperative cognitive dysfunction by facilitating PINK1-dependent mitophagy and impeding caspase-3/GSDME-dependent pyroptosis. Exp Neurol. (2023) 367:114463. doi: 10.1016/j.expneurol.2023.114463

PubMed Abstract | Crossref Full Text | Google Scholar

38. Badr, MY, Amer, RA, Kotait, MA, Shoeib, SM, and Reda, AM. Role of multimodal advanced biomarkers as potential predictors of cognitive and psychiatric aspects of Parkinson's disease. Egypt J Neurol Psychiatr Neurosurg. (2023) 59:58. doi: 10.1186/s41983-023-00662-2

Crossref Full Text | Google Scholar

39. Tang, Y, Han, L, Li, S, Hu, T, Xu, Z, Fan, Y, et al. Plasma GFAP in Parkinson’s disease with cognitive impairment and its potential to predict conversion to dementia. NPJ Parkinsons Dis. (2023) 9:23. doi: 10.1038/s41531-023-00447-7

PubMed Abstract | Crossref Full Text | Google Scholar

40. Mohammadi, R, Ng, SYE, Tan, JY, Ng, ASL, Deng, X, Choi, X, et al. Machine learning for early detection of cognitive decline in Parkinson's disease using multimodal biomarker and clinical data. Biomedicine. (2024) 12:2758. doi: 10.3390/biomedicines12122758

PubMed Abstract | Crossref Full Text | Google Scholar

41. Lin, YS, Lee, WJ, Wang, SJ, and Fuh, JL. Levels of plasma neurofilament light chain and cognitive function in patients with Alzheimer or Parkinson disease. Sci Rep. (2018) 8:17368. doi: 10.1038/s41598-018-35766-w

PubMed Abstract | Crossref Full Text | Google Scholar

42. Pawlik, P, and Błochowiak, K. The role of salivary biomarkers in the early diagnosis of Alzheimer’s disease and Parkinson’s disease. Diagnostics. (2021) 11:371. doi: 10.3390/diagnostics11020371

PubMed Abstract | Crossref Full Text | Google Scholar

43. Kharel, S, Ojha, R, Bist, A, Joshi, SP, Rauniyar, R, and Yadav, JK. Salivary alpha-synuclein as a potential fluid biomarker in Parkinson's disease: a systematic review and meta-analysis. Aging Med. (2022) 5:53–62. doi: 10.1002/agm2.12192

PubMed Abstract | Crossref Full Text | Google Scholar

44. Muddaloor, P, Farinango, M, Ansary, A, Dakka, A, Nazir, Z, Shamim, H, et al. Prospect of alpha-Synuclein (A-Syn) isolation from saliva as a promising diagnostic biomarker alternative in Parkinson's disease (PD): a systematic review. Cureus. (2022) 14:e29880. doi: 10.7759/cureus.29880

PubMed Abstract | Crossref Full Text | Google Scholar

45. Ouyang, Q, Xu, L, Zhang, Y, Huang, L, Li, L, and Yu, M. Nonlinear relationship between homocysteine and mild cognitive impairment in early Parkinson’s disease: a cross-sectional study. Neuropsychiatr Dis Treat. (2024) 20:913–21. doi: 10.2147/NDT.S460938

PubMed Abstract | Crossref Full Text | Google Scholar

46. Margiana, R, Hammid, AT, Ahmad, I, Alsaikhan, F, Jalil, AT, Tursunbaev, F, et al. Current progress in aptasensor for ultra-low level monitoring of Parkinson’s disease biomarkers. Crit Rev Anal Chem. (2024) 54:617–32. doi: 10.1080/10408347.2022.2091920

Crossref Full Text | Google Scholar

47. Wang, Z, Becker, K, Donadio, V, Siedlak, S, Yuan, J, Rezaee, M, et al. Skin α-synuclein aggregation seeding activity as a novel biomarker for Parkinson disease. JAMA Neurol. (2021) 78:120. doi: 10.1001/jamaneurol.2020.3311

Crossref Full Text | Google Scholar

48. Majbour, N, Aasly, J, Abdi, I, Ghanem, S, Erskine, D, van de Berg, W, et al. Disease-associated α-synuclein aggregates as biomarkers of Parkinson disease clinical stage. Neurology. (2022) 99:e2417–27. doi: 10.1212/WNL.0000000000201199

PubMed Abstract | Crossref Full Text | Google Scholar

49. Zhao, Y, Luan, M, Liu, J, Wang, Q, Deng, J, Wang, Z, et al. Skin α-synuclein assays in diagnosing Parkinson's disease: a systematic review and meta-analysis. J Neurol. (2025) 272:326. doi: 10.1007/s00415-025-12978-5

Crossref Full Text | Google Scholar

50. Ren, J, Zhou, G, Wang, Y, Zhang, R, Guo, Z, Zhou, H, et al. Association of GBA genotype with motor and cognitive decline in Chinese Parkinson’s disease patients. Front Aging Neurosci. (2023) 15:1091919. doi: 10.3389/fnagi.2023.1091919

PubMed Abstract | Crossref Full Text | Google Scholar

51. Ramezani, M, Mouches, P, Yoon, E, Rajashekar, D, Ruskey, JA, Leveille, E, et al. Investigating the relationship between the SNCA gene and cognitive abilities in idiopathic Parkinson’s disease using machine learning. Sci Rep. (2021) 11:4917. doi: 10.1038/s41598-021-84316-4

PubMed Abstract | Crossref Full Text | Google Scholar

52. Zhao, K, Li, Y, Liu, Z, Long, H, Zhao, C, Luo, F, et al. Parkinson's disease associated mutation E46K of α-synuclein triggers the formation of a distinct fibril structure. Nat Commun. (2020) 11:2643. doi: 10.1038/s41467-020-16386-3

PubMed Abstract | Crossref Full Text | Google Scholar

53. Weigand, AJ, Ortiz, G, Walker, KS, Galasko, DR, Bondi, MW, and Thomas, KR. APOE differentially moderates cerebrospinal fluid and plasma phosphorylated tau181 associations with multi-domain cognition. Neurobiol Aging. (2023) 125:1–8. doi: 10.1016/j.neurobiolaging.2022.10.016

PubMed Abstract | Crossref Full Text | Google Scholar

54. Gordon, EM, Devaney, JM, Bean, S, and Vaidya, CJ. Resting-state striato-frontal functional connectivity is sensitive to DAT1 genotype and predicts executive function. Cereb Cortex. (2015) 25:336–45. doi: 10.1093/cercor/bht229

PubMed Abstract | Crossref Full Text | Google Scholar

55. Hasani, F, Ahmadi, P, Masrour, M, Khamaki, S, Rahmati, R, and Surguchov, A. Epigenetic alterations of cognition functions in Parkinson's disease (2024). doi: 10.20944/preprints202406.0981.v1,

Crossref Full Text | Google Scholar

56. Zhang, Y, and Shen, S. Epigenome-wide DNA methylation analysis of late-stage mild cognitive impairment. Front Cell Dev Biol. (2024) 12:1276288. doi: 10.3389/fcell.2024.1276288

PubMed Abstract | Crossref Full Text | Google Scholar

57. Jiménez-Ramírez, IA, and Castaño, E. Non-coding RNAs in the pathogenesis of Alzheimer’s disease: β-amyloid aggregation, tau phosphorylation and neuroinflammation. Mol Biol Rep. (2025) 52:183. doi: 10.1007/s11033-025-10284-x

PubMed Abstract | Crossref Full Text | Google Scholar

58. Arya, R, Haque, AKMA, Shakya, H, Billah, MM, Parvin, A, Rahman, M-M, et al. Parkinson’s disease: biomarkers for diagnosis and disease progression. Int J Mol Sci. (2024) 25:12379. doi: 10.3390/ijms252212379

PubMed Abstract | Crossref Full Text | Google Scholar

59. Geritz, J, Welzel, J, Hansen, C, Maetzler, C, Hobert, MA, Elshehabi, M, et al. Cognitive parameters can predict change of walking performance in advanced Parkinson's disease - chances and limits of early rehabilitation. Front Aging Neurosci. (2022) 14:1070093. doi: 10.3389/fnagi.2022.1070093

PubMed Abstract | Crossref Full Text | Google Scholar

60. Berron, D, Olsson, E, Andersson, F, Janelidze, S, Tideman, P, Düzel, E, et al. Remote and unsupervised digital memory assessments can reliably detect cognitive impairment in Alzheimer's disease. Alzheimers Dement. (2024) 20:4775–91. doi: 10.1002/alz.13919

PubMed Abstract | Crossref Full Text | Google Scholar

61. Park, SY, and Schott, N. Which motor-cognitive abilities underlie the digital trail-making test? Decomposing various test scores to detect cognitive impairment in Parkinson's disease-pilot study. Appl Neuropsychol Adult. (2025) 32:60–74. doi: 10.1080/23279095.2022.2147837

PubMed Abstract | Crossref Full Text | Google Scholar

62. Klotz, LV, Casjens, S, Johnen, G, Taeger, D, Brik, A, Eichhorn, F, et al. Combination of calretinin, MALAT1, and GAS5 as a potential prognostic biomarker to predict disease progression in surgically treated mesothelioma patients. Lung Cancer. (2024) 192:107802. doi: 10.1016/j.lungcan.2024.107802

Crossref Full Text | Google Scholar

63. Zhang, J, Gao, Y, He, X, Feng, S, Hu, J, Zhang, Q, et al. Identifying Parkinson's disease with mild cognitive impairment by using combined MR imaging and electroencephalogram. Eur Radiol. (2021) 31:7386–94. doi: 10.1007/s00330-020-07575-1

PubMed Abstract | Crossref Full Text | Google Scholar

64. Zhu, Y, Wang, F, Ning, P, Zhu, Y, Zhang, L, Li, K, et al. Multimodal neuroimaging-based prediction of Parkinson’s disease with mild cognitive impairment using machine learning technique. npj Parkinson's Dis. (2024) 10:218. doi: 10.1038/s41531-024-00828-6

Crossref Full Text | Google Scholar

65. Cirincione, A, Lynch, K, Bennett, J, Choupan, J, Varghese, B, Sheikh-Bahaei, N, et al. Prediction of future dementia among patients with mild cognitive impairment (MCI) by integrating multimodal clinical data. Heliyon. (2024) 10:–e36728. doi: 10.1016/j.heliyon.2024.e36728

Crossref Full Text | Google Scholar

66. Sato, M, Kobayashi, T, Soroida, Y, Tanaka, T, Nakatsuka, T, Nakagawa, H, et al. Development of novel deep multimodal representation learning-based model for the differentiation of liver tumors on B-mode ultrasound images. J Gastroenterol Hepatol. (2022) 37:678–84. doi: 10.1111/jgh.15763

PubMed Abstract | Crossref Full Text | Google Scholar