<?xml version="1.0" encoding="utf-8"?>
    <rss version="2.0">
      <channel xmlns:content="http://purl.org/rss/1.0/modules/content/">
        <title>Frontiers in Medical Technology | Medtech Data Analytics section | New and Recent Articles</title>
        <link>https://www.frontiersin.org/journals/medical-technology/sections/medtech-data-analytics</link>
        <description>RSS Feed for Medtech Data Analytics section in the Frontiers in Medical Technology journal | New and Recent Articles</description>
        <language>en-us</language>
        <generator>Frontiers Feed Generator,version:1</generator>
        <pubDate>2026-05-12T15:51:22.960+00:00</pubDate>
        <ttl>60</ttl>
        <item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2026.1800134</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2026.1800134</link>
        <title><![CDATA[Economic evaluation of hearing aid use and quality of life in older adults with hearing impairment in India]]></title>
        <pubdate>2026-05-12T00:00:00Z</pubdate>
        <category>Brief Research Report</category>
        <author>Rakesh Kumar Sahoo</author><author>Krushna Chandra Sahoo</author><author>Urmi Pattanayak</author><author>Abhinav Sinha</author><author>Abhisek Jena</author><author>Lanu Wanboy Aimol</author><author>Kavitha Rajsekar</author><author>Debdutta Bhattacharya</author><author>Sanghamitra Pati</author>
        <description><![CDATA[Hearing impairment is a growing public health challenge among older adults in India, with low uptake of hearing aids despite potential benefits for communication, social participation, and health-related quality of life (HRQoL). This study evaluated the cost–utility of hearing aid provision for older adults (≥60 years) with hearing impairment in India to inform policy and financing decisions. A cost–utility analysis compared hearing aid use with no hearing aid over a 3-year decision tree and a 15-year Markov model. Data were collected in 2023 from 636 participants (276 users, 360 non-users). HRQoL was measured using EQ-5D-5L and converted to QALYs with the India value set. Costs were based on the government ceiling price for digital behind-the-ear hearing aids (₹8,000 per ear), modelled for unilateral and bilateral use. ICERs and budget impact were estimated. Mean EQ-5D-5L utility was higher among hearing aid users than non-users (0.832 vs. 0.601), with corresponding QALYs of 5.824 vs. 4.207 (incremental gain 1.617 QALYs). In the 3-year decision tree, ICERs were ₹5,419/QALY (one-year) and ₹10,359/QALY (two-year). In the discounted 15-year Markov model, ICERs were ₹3,076/QALY (one-year) and ₹5,971/QALY (two-year). Budget impact was ₹1,230 crore over 3 years (₹410 crore annually) for the modelled scale-up among older adults. Hearing-aid provision for older adults with hearing impairment in India is highly cost-effective in short- and long-term modelling. Prioritising access within healthy ageing and universal health coverage, alongside strengthened assessment, fitting, counselling, and follow-up, could maximise real-world benefits.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2026.1800307</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2026.1800307</link>
        <title><![CDATA[Powering care at the frontlines: healthcare providers' perspectives on solarised primary health centres in rural Karnataka, India, using the WHO HHFA framework]]></title>
        <pubdate>2026-05-07T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Veeresh Tadahal</author><author>Rinshu Dwivedi</author><author>Ramesh Athe</author>
        <description><![CDATA[IntroductionReliable electricity is fundamental to effective healthcare delivery; however, healthcare facilities in low- and middle-income countries continue to experience chronic energy shortage and frequent power interruptions. In rural India, Primary Health Centres (PHCs) often struggle to provide essential care due to unreliable electricity. This study explored healthcare providers’ perspective on the impact of solar energy interventions on healthcare infrastructure and service delivery in an aspirational district of Karnataka, India.MethodsA qualitative phenomenological study was conducted across the Yadgir district of North Karnataka, India. Focus Group Discussion (FGDs) and In-depth Interviews (IDIs) were carried out with 55 participants (16 men and 39 women), primarily healthcare providers, including medical officers, nurses, technicians, Accredited Social Health Activists (ASHAs), and other support staff. The interviews and discussions were audio-recorded, transcribed, translated into English, and thematically analysed. The study adhered to the Consolidated Criteria for Reporting Qualitative Research (COREQ) guidelines and applied the Harmonized Healthcare Assessment (HHFA) framework.ResultsFour main themes emerged: (1) strengthened service availability; (2) enhanced service readiness; (3) advanced quality of care; and (4) improved data management, costs, operation, and maintenance. Solarisation of healthcare facilities significantly improved service quality, reliability, and resilience by enabling safe maternal deliveries, ensuring reliable vaccine cold storage, and supporting timely emergency care. The perspectives of healthcare providers offer a sustainable pathway to universal health coverage in underserved regions.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2026.1780837</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2026.1780837</link>
        <title><![CDATA[Enhancing context-aware SARS disorder management: a proposed multi-agent simulation framework with machine learning and bio-sensor data integration]]></title>
        <pubdate>2026-04-15T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author> Abdullah</author><author>Zulaikha Fatima</author><author>Nida Hafeez</author><author>Muhammad Ateeb Ather</author><author>Rolando Quintero Tellez</author><author>Grigori Sidorov</author><author>Carlos Guzmán Sánchez-Mejorada</author><author>Miguel Jesús Torres Ruiz</author>
        <description><![CDATA[In this work, SARS disorder denotes a generic acute severe respiratory distress condition characterized by abnormal respiratory rate, oxygen saturation, fever, and cardiovascular stress indicators, and does not represent a COVID-19 diagnostic system. Our research aims at analyzing a context-aware SARS disorder management system through the implementation of a multi-agent simulation framework using the NetLogo setting. The system relies on the use of interacting agents as well as non-monotonic, context-sensitive reasoning to reduce uncertainty and deal with the possible inconsistencies that happen due to biosensor recordings. A knowledge-based inference component is the combination of physiological sensor outputs and domain specific contextual data to assist in making informed decisions. The research involved the use of several machine-learning classifiers, that is, Naïve Bayes, Multinomial Naïve Bayes, Decision Table, Logistic Regression, and Sequential Minimal Optimization (SMO) so as to evaluate their appropriateness in being incorporated into the developed structure. To measure the system performance, standard evaluation measures were used such as True Positive (TP), False Positive (FP), Precision, Recall, F-Measure, Matthews Correlation Coefficient (MCC), Receiver Operating Characteristic (ROC) curve, and the Precision-Recall curve (PRC). The framework includes a list of physiological, environmental, and contextual variables, such as electrocardiographic parameters, heart-rate parameters, blood-pressure parameters, arterial oxygen saturation parameters, core body temperature, room temperature, the past history of the patient, and parameters that relate to alerts. The classification task is to produce probabilistic forecasts that help to define whether a patient should be alerted or clinical staff members informed in order to facilitate context-specific healthcare response.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2026.1708094</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2026.1708094</link>
        <title><![CDATA[Mamba-enhanced codebook learning with anatomical constraints for liver and tumor segmentation in 3D CT volumes]]></title>
        <pubdate>2026-03-18T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Yanfei Teng</author><author>Xiang Li</author><author>Zhenpeng Chen</author><author>Shunlin Guo</author>
        <description><![CDATA[Precise delineation of the liver and its tumors in 3D CT scans plays a vital role in clinical diagnosis and therapeutic planning. However, current deep learning approaches frequently struggle with tumor heterogeneity, varying lesion sizes, and ambiguous boundaries, which can limit their effectiveness. To address these issues, we propose an end-to-end hierarchical network that effectively integrates multi-scale context modeling, global relational learning, and structured feature representation. First, a multi-scale texture encoder is designed to capture tumor characteristics across different spatial resolutions. To model long-range dependencies across slices, we introduce a global relational representation module built upon the emerging Mamba architecture, enabling efficient and directional context aggregation in 3D volumes. Second, to enhance feature compactness and stability, we propose a learnable codebook module that quantizes high-dimensional features into a finite set of semantic prototypes, promoting discriminative representation learning while suppressing redundancy. Furthermore, anatomical prior knowledge—specifically, the spatial constraint that tumors must reside within the liver—is incorporated via an inclusion loss, which explicitly regularizes the segmentation outputs. Comprehensive experiments on the public LiTS dataset show that our method attains state-of-the-art results, surpassing existing methods in Dice score, volumetric overlap error (VOE), and boundary metrics (ASD and 95HD). Ablation analyses confirm the individual contribution of each module, demonstrating the architecture’s effectiveness for accurate and reliable liver and tumor segmentation.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2026.1748964</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2026.1748964</link>
        <title><![CDATA[IoMT–Fog–Cloud-based AI frameworks for chronic disease diagnosis: updated comparative analysis with recent AI-IoMT models (2020–2025)]]></title>
        <pubdate>2026-01-22T00:00:00Z</pubdate>
        <category>Systematic Review</category>
        <author>Kitsakorn Locharoenrat</author>
        <description><![CDATA[Chronic diseases such as diabetes and cardiovascular disease require frequent monitoring and timely clinical feedback to prevent complications. Internet of Medical Things (IoMT) systems increasingly combine near-patient sensing with Fog and Cloud computing so that time-critical preprocessing and inference can run close to the patient while compute-intensive training and population-level analytics remain in the Cloud. This review synthesizes primary studies published between 2020 and 2025 that implement AI-enabled IoMT, with an emphasis on systems that report both diagnostic performance and network quality-of-service (QoS). Following PRISMA 2020, we screened database records and included 14 primary studies; we focus the joint performance–QoS synthesis on six IoMT–Fog–Cloud frameworks for diabetes and cardiovascular disease and compare them with two recent multi-disease AI-IoMT models (DACL and TasLA). Diabetes-oriented implementations commonly report accuracy around 95%–96% using explainable or ensemble deep learning, whereas some cardiovascular frameworks report >99% accuracy in controlled settings; we therefore discuss plausible sources of optimistic performance, including small datasets, class imbalance, curated benchmarks, and potential leakage/overfitting in simulation-based evaluations. Across IoMT–Fog–Cloud studies, placing preprocessing and/or inference at the Fog layer repeatedly reduces end-to-end latency for streaming biosignals, but multi-Fog provisioning can increase energy and power demands. To support more reproducible comparisons, we organize 14 extracted metrics into (i) diagnostic performance (accuracy, precision, recall, F1-score, sensitivity, specificity) and (ii) system/network QoS (latency, jitter, throughput, bandwidth utilization, processing/execution time, network usage, energy consumption, power consumption), and we translate the evidence into study-linked design recommendations for future deployments.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2025.1748577</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2025.1748577</link>
        <title><![CDATA[Ventricular suction detection algorithm designed for ventricular assist devices]]></title>
        <pubdate>2026-01-13T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Yijiao Wu</author><author>Yuzhuo Yang</author><author>Xudong Pan</author><author>Shunzhou Yu</author>
        <description><![CDATA[BackgroundVentricular assist devices (VADs) are an effective treatment for end-stage heart failure and can significantly improve patients' quality of life. However, when the rotational speed of the VAD does not match the intraventricular blood volume, ventricular suction may occur. Severe suction can lead to ventricular collapse, making accurate and real-time suction detection critically important.MethodsTwo statistical features and two frequency-domain features were extracted from the pump flow signal to build a classification and regression tree (CART) model. Additionally, a secondary decision-making process was applied using a time-domain threshold.ResultsThe proposed method was validated using both in vivo and in vitro experimental data. Experimental results show that, compared to existing suction detection techniques, the proposed approach not only reduces computational complexity but also achieves higher detection accuracy and enhanced algorithmic stability.ConclusionsThe proposed method provides a more efficient and reliable solution for real-time ventricular suction detection, which is crucial for the safe operation of VADs in clinical settings.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2025.1678192</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2025.1678192</link>
        <title><![CDATA[Comparison of Nissen vs. Toupet fundoplication in laparoscopic hiatal hernia repair for gastroesophageal reflux disease with extra-esophageal symptoms]]></title>
        <pubdate>2026-01-09T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Qingchao Zhu</author><author>Nengquan Sheng</author><author>Zhigang Wang</author><author>Yang Xia</author>
        <description><![CDATA[ObjectiveThis study aims to evaluate the clinical efficacy of laparoscopic hiatal hernia repair (LHHR) in treating gastroesophageal reflux disease (GERD) and to through the therapeutic effect of total (360°) and partial (270°) laparoscopic fundoplication.MethodsThis retrospective observational study enrolled 100 patients, both with and without documented extra-oesophageal symptoms of GERD. Data were extracted from medical records, covering basic information, symptoms, treatments, and follow-up. Symptom relief and quality of life were assessed via GERD-Q score, Reflux Symptom Index (RSI), and EORTC QLQ-C30 scale, offering a foundation for comprehensive GERD patient management and treatment evaluation in clinical practice.ResultsThe DeMeester index significantly decreased postoperatively in both the laparoscopic Nissen fundoplication (LNF) group (from 55.23 ± 25.12 to 11.45 ± 10.20, p < 0.05) and the laparoscopic Toupet fundoplication (LTF) group (from 60.51 ± 28.40 to 11.70 ± 9.65, p < 0.05). The RSI scores improved at 12 months postoperatively in both groups: LNF group (from 23.1 ± 15.4 to 13.7 ± 9.6, p < 0.05) and LTF group (from 21.9 ± 15.8 to 12.8 ± 8.2, p < 0.05). The GERD scores also improved postoperatively: LNF group (from 13 ± 5.0 to 10 ± 4.4, p < 0.05) and LTF group (from 10 ± 4.7 to 7.5 ± 4.5, p < 0.05).ConclusionOur report demonstrates that LHHR significantly improved GERD regarding symptom frequency, acid reflux occurrences and DeMeester score. Both LNF and LTF provide good results.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2025.1702201</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2025.1702201</link>
        <title><![CDATA[X-ray contrast-adjustable 3D printing for multimodal fusion of microCT and histology]]></title>
        <pubdate>2026-01-09T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Philipp Nolte</author><author>Chris Johann Ackurat</author><author>Marcel Brettmacher</author><author>Marius Reichardt</author><author>Marieke Stammes</author><author>Christoph Rußmann</author><author>Christian Dullin</author>
        <description><![CDATA[IntroductionPhantoms and reference structures are essential tools for calibration and correlative imaging in pre-clinical and research applications of X-ray-based imaging. They serve as reference standards, ensuring consistency and accuracy in imaging results. However, generating individual phantoms often involves a complex creation process, high production costs, and significant time investment.Material and methodsConic reference structures were 3D printed using a mixture of UV-curable resin and X-ray contrast agents. These structures were then embedded together with lung specimens of SARS-CoV-2-infected rhesus macaques in a methyl methacrylate-based solution. The polymerized blocks were scanned using propagation-based phase-contrast microCT, a method chosen for its superior ability to enhance contrast, especially in low-absorbing biological samples. Utilizing the conic reference structures, subsequently performed histological sections were co-registered into the 3D context of the microCT datasets.ResultsThe produced 3D printed models were highly visible in terms of contrast and detail in both imaging methods, allowing for a precise co-registration of microCT and histological imaging.ConclusionsThe novel methodology of using contrast agents and resin in 3D printing enables the generation of customizable, contrast-specific phantoms and reference structures. These can be straightforwardly segmented from the embedding material, significantly simplifying and enhancing the workflow of multimodal imaging processes. In this study, 3D printed conic reference structures were effectively used to automate and streamline the precise multimodal fusion of microCT and histological imaging.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2025.1732580</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2025.1732580</link>
        <title><![CDATA[Explainable machine learning for predicting postoperative length of stay after gastrectomy: a nationwide study using XGBoost and SHAP]]></title>
        <pubdate>2025-12-05T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Tsunehiko Maruyama</author><author>Kazuto Ikezawa</author><author>Hideo Suzuki</author><author>Tomohiro Kurokawa</author><author>Yoshimasa Akashi</author><author>Tatsuya Oda</author>
        <description><![CDATA[BackgroundGastric cancer remains a major cause of cancer-related morbidity and mortality. Despite advances in surgical and perioperative care, prolonged hospitalization continues to strain healthcare systems. Predicting postoperative length of stay (LOS) could support personalized care and efficient resource allocation. Japan's nationwide Diagnosis Procedure Combination (DPC) database provides real-world data for large-scale analysis, but no study has applied machine learning to predict LOS after gastrectomy.MethodsThis retrospective study included 26,097 patients who underwent gastrectomy between 2017 and 2022 at 472 hospitals in Japan. Using XGBoost, we developed a predictive model based on 1,433 admission-time variables extracted from the DPC database. Model performance was evaluated using Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) in a five-fold cross-validation. SHAP values were used to interpret feature importance.ResultsThe final model achieved an RMSE of 3.74 and MAE of 2.82 days. Key predictors of LOS included surgical procedure (laparoscopic distal gastrectomy and open total gastrectomy), designated cancer hospital, hospital size, peritoneal dissemination, and admission ADL score. SHAP analysis revealed that Laparoscopic distal gastrectomy and higher hospital volume were associated with shorter LOS, while open total gastrectomy was associated with longer LOS.ConclusionsWe developed a machine learning model that predicts postoperative length of stay with an error range of 2–4 days using admission data. This proof-of-concept study demonstrates the feasibility of predicting length of stay from admission data, showing that explainable AI can replicate intuitive patterns in surgical oncology while simultaneously identifying unexpected insights from administrative data. These findings highlight the clinical potential of explainable AI for perioperative workflow optimization.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2025.1699821</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2025.1699821</link>
        <title><![CDATA[An advanced multimodal image fusion model for accurate detection of Alzheimer's disease using MRI and PET]]></title>
        <pubdate>2025-12-02T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Arshiya S. Ansari</author><author>Mohammad Sajid Mohammadi</author><author>Carlo Cattani</author><author>Asifa Tassaddiq</author>
        <description><![CDATA[The accurate detection of Alzheimer's disease (AD), a progressive and irreversible neurodegenerative disorder, remains a critical challenge in clinical neuroscience. The research aims to develop an advanced multimodal image fusion model for the accurate detection of AD using positron emission tomography (PET) and magnetic resonance imaging (MRI) techniques. The proposed method leverages structural MRI and functional 18-fluorodeoxyglucose PET (FDG-PET) information derived from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). After preprocessing, including Gaussian filtering, skull stripping, and intensity normalization, voxel-based morphometry (VBM) is applied to extract gray matter (GM) features relevant to AD progression. A GM mask generated from MRI is used to isolate corresponding metabolic activity in the PET scans. These features are then integrated using a mask-coding strategy to construct a unified representation that captures both anatomical and functional characteristics. For classification, the model introduces a Glowworm Swarm-Optimized Spatial Multimodal Attention-Enriched Convolutional Neural Network (GWS-SMAtt-ECNN), where the optimization enhances both feature selection and network parameter tuning. The Python was implemented, and the result demonstrates that the proposed multimodal image fusion strategy outperforms traditional unimodal and basic fusion approaches in terms of F1-score (94.22%), recall (96.73%), and accuracy (98.70%). These results highlight the therapeutic usefulness of the suggested improved fusion architecture in facilitating immediate and accurate AD detection by MRI and PET.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2025.1685088</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2025.1685088</link>
        <title><![CDATA[Explainable multi-modal machine learning for predicting occult pulmonary metastases in differentiated thyroid cancer: a SHAP-based approach prior to radioactive iodine scans]]></title>
        <pubdate>2025-11-28T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Yuqi Su</author><author>Yuhuang Cai</author><author>Shui Jin</author><author>Xuemei Ye</author><author>Jaesik Jeong</author><author>Ye Yuan</author><author>Heqing Yi</author>
        <description><![CDATA[BackgroundPatients with differentiated thyroid cancer (DTC) may have occult lung metastases before 131iodine (131I) treatment. Identifying occult lung metastases before 131I treatment is of great clinical value for the correct staging of patients and the establishment of 131I treatment plans. Our research is of great significance in establishing statistical models for clinical data using machine learning algorithms to study the prediction of lung metastasis before 131I treatment.MethodsPatients were selected from Zhejiang cancer hospital and data was from two groups of DTC patients treated with 131I, where the experimental group consisted of 55 patients who showed no lung metastases on CT but tested positive on 131I-whole body scan (131I-WBS). The control group included 316 patients who tested negative for metastases across CT, ultrasound, and 131I-WBS. Six machine learning algorithms such as Support Vector Machines (SVM), Decision Trees (DT), Random Forests (RF), Logistic Regression (LR), Extreme Gradient Boosting (XGBoost), and K-Nearest Neighbors (KNN) were employed to predict models and AUC, sensitivity, accuracy, precision, specificity, F1 Score were used to compare the performance between each models. Finally, the SHAP algorithm was used to explain the importance rank of the features.ResultsA total of 371 thyroid cancer patients were included in this study, 55 patients with occult lung metastasis and 316 patients in the control group. The data is divided into a training set and a testing set in a 7:3 ratio. Eleven acceptable variables analyzed including gender, age, T stage, N stage, tumor size, degree of invasion, number of lymph node metastases count, Thyroid Stimulating Hormone (TSH), thyroglobulin (Tg), Thyroglobulin antibodies (Tgab), and administrated activity were screened out by multivariate Cox regression. Evaluation indicators of the best model- LR were as following: accuracy (0.91), recall rate (0.64), precision (0.92), F1-s core (0.70), Area Under Curve (AUC) value (0.93), and the Specificity score (0.96).ConclusionThe logistic model (LR) showed the best performance in predicting occult lung metastases of thyroid cancer patients before 131I-WBS. Lymph nodes metastases and throglobulin have the most significant impact on the prediction.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2025.1712952</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2025.1712952</link>
        <title><![CDATA[DGA-Net: a dual-branch group aggregation network for liver tumor segmentation in medical images]]></title>
        <pubdate>2025-11-26T00:00:00Z</pubdate>
        <category>Methods</category>
        <author>Lin Zhu</author><author>Shuyan Liu</author>
        <description><![CDATA[Hepatocellular carcinoma (HCC) is one of the most common malignant tumors worldwide. Due to its high invasiveness and poor prognosis, it ranks among the top three causes of cancer-related deaths globally. Accurate segmentation of the liver and lesion areas is crucial. It provides key support for diagnosis, surgical planning, and rehabilitation therapy. Deep learning technologies have been applied to the automatic segmentation of the liver and tumors. However, several issues remain, such as insufficient utilization of inter-pixel relationships, lack of refined processing after fusing high-level and low-level features, and high computational costs. To address insufficient inter-pixel modeling and high parameter costs, we propose DGA-Net (Dual-branch Group Aggregation Network for Liver Tumor Segmentation in Medical Images), a dual-branch architecture that includes two main components, i.e., a dual-branch encoder and a decoder with a specific module. The dual-branch encoder consists of the Fourier Spectral Learning Multi-Scale Fusion (FSMF) branch and the Multi-Axis Aggregation Hadamard Attention (MAHA) branch. The decoder is equipped with a Group Multi-Head Cross-Attention Aggregation (GMCA) module. The FSMF branch uses a Fourier network to learn amplitude and phase information. This helps capture richer features and details. The MAHA branch combines spatial information to enhance discriminative features. At the same time, it effectively reduces computational costs. The GMCA module merges features from different branches. This not only improves localization capabilities but also establishes long-range inter-pixel dependencies. We conducted experiments on the public LiTS2017 liver tumor dataset. Experiments on the public LiTS2017 liver tumor dataset show that the proposed method outperforms existing state-of-the-art approaches, achieving Dice-per-case (DPC) scores of 94.84% for liver and 69.51% for tumors, outperforming competing methods such as PVTFormer by 0.72% (liver) and 1.68% (tumor), and AGCAF-Net by 0.97% (liver) and 2.59% (tumor). We also carried out experiments on the 3DIRCADb dataset. The method still delivers excellent results, which highlights its strong generalization ability.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2025.1674343</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2025.1674343</link>
        <title><![CDATA[More than just a heatmap: elevating XAI with rigorous evaluation metrics]]></title>
        <pubdate>2025-10-28T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Dost Muhammad</author><author>Malika Bendechache</author>
        <description><![CDATA[BackgroundMagnetic Resonance Imaging (MRI) and ultrasound are central to tumour diagnosis and treatment planning. Although Deep learning (DL) models achieve strong prediction performance, high computational demand and limited explainability can hinder clinical adoption. Common post hoc Explainable Artificial Intelligence (XAI) methods namely Grad-CAM, LIME, and SHAP often yield fragmented or anatomically misaligned saliency maps.MethodsWe propose SpikeNet, a hybrid framework that combines Convolutional Neural Networks (CNNs) for spatial feature encoding with Spiking Neural Networks (SNNs)for efficient, event driven processing. SpikeNet includes a native saliency module that produces explanations during inference. We also introduce XAlign, a metric that quantifies alignment between explanations and expert tumour annotations by integrating regional concentration, boundary adherence, and dispersion penalties. Evaluation follows patient level cross validation on TCGA–LGG (MRI, 22 folds) and BUSI (ultrasound, 5 folds), with slice level predictions aggregated to patient level decisions and BUSI treated as a three class task. We report per image latency and throughput alongside accuracy, precision, recall, F1, AUROC, and AUPRC.ResultsSpikeNet achieved high prediction performance with tight variability across folds. On TCGA–LGG it reached 97.12±0.63% accuracy and 97.43±0.60% F1; on BUSI it reached 98.23±0.58% accuracy and 98.32±0.50% F1. Patient level AUROC and AUPRC with 95% confidence intervals further support these findings. On a single NVIDIA RTX 3090 with batch size 16 and FP32 precision, per image latency was about 31 ms and throughput about 32 images per second, with the same settings applied to all baselines. Using XAlign, SpikeNet produced explanations with higher alignment than Grad-CAM, LIME, and SHAP on both datasets. Dataset level statistics, paired tests, and sensitivity analyses over XAlign weights and explanation parameters confirmed robustness.ConclusionSpikeNet delivers accurate, low latency, and explainable analysis for MRI and ultrasound by unifying CNN based spatial encoding, sparse spiking computation, and native explanations. The XAlign metric provides a clinically oriented assessment of explanation fidelity and supports consistent comparison across methods. These results indicate the potential of SpikeNet and XAlign for trustworthy and efficient clinical decision support.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2025.1621158</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2025.1621158</link>
        <title><![CDATA[Using machine learning methods to investigate the impact of comorbidities and clinical indicators on the mortality rate of COVID-19]]></title>
        <pubdate>2025-09-22T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Yueh-Chen Hsieh</author><author>Sin Chen</author><author>Shu-Yu Tsao</author><author>Jiun-Ruey Hu</author><author>Wan-Ting Hsu</author><author>Chien-Chang Lee</author>
        <description><![CDATA[BackgroundThis study aims to develop a machine learning model to predict the 30-day mortality risk of hospitalized COVID-19 patients while leveraging federated learning to enhance data privacy and expand the model's applicability. Additionally, SHapley Additive exPlanations (SHAP) values were utilized to assess the impact of comorbidities on mortality.MethodsA retrospective analysis was conducted on 6,321 clinical records of hospitalized COVID-19 patients between January 2021 and October 2022. After excluding cases involving patients under 18 years of age and non-Omicron infections, a total of 4,081 records were analyzed. Key features included three demographic data, six vital signs at admission, and 79 underlying comorbidities. Four machine learning models were compared, including Lasso, Random Forest, XGBoost, and TabNet, with XGBoost demonstrating superior performance. Federated learning was implemented to enable collaborative model training across multiple medical institutions while maintaining data security. SHAP values were applied to interpret the contribution of each comorbidity to the model's predictions.ResultsA subset of 2,156 records from the Taipei branch was used to evaluate model performance. XGBoost achieved the highest AUC of 0.96 and a sensitivity of 0.94. Two versions of the XGBoost model were trained: one incorporating vital signs, suitable for emergency room applications where patients come in with unstable vital signs, and another excluding vital signs, optimized for outpatient settings where we encounter patients with multiple comorbidities. After implementing federated learning, the AUC of the Taipei cohort decreased to 0.90, while the performance of other cohorts improved to meet the required standards. SHAP analysis identified comorbidities including diabetes mellitus, cerebrovascular disease, and chronic lung disease to have a neutral or even protective association with 30-day mortality.ConclusionXGBoost outperformed other models making it a viable tool for both emergency and outpatient settings. The study underscores the importance of chronic disease assessment in predicting COVID-19 mortality, revealing some comorbidities such as diabetes mellitus, cerebrovascular disease and chronic lung disease to have protective association with 30-day mortality. These findings suggest potential refinements in current treatment guidelines, particularly concerning high-risk conditions. The integration of federated learning further enhances the model's clinical applicability while preserving patient privacy.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2025.1621922</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2025.1621922</link>
        <title><![CDATA[ATTNFNET: feature aware depth-to-pressure translation with cGAN training]]></title>
        <pubdate>2025-09-16T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Neevkumar Manavar</author><author>Hanno Gerd Meyer</author><author>Joachim Waßmuth</author><author>Barbara Hammer</author><author>Axel Schneider</author>
        <description><![CDATA[Excessive pressure and shear forces on bedridden patients can lead to pressure injuries, particularly on those with existing ulcers. Monitoring pressure distribution is crucial for preventing such injuries by identifying high-risk areas. To address this challenge, we propose Attention Feature Network (AttnFnet), a self-attention-based deep neural network that generates pressure distribution maps from single-depth images using Conditional Generative Adversarial Network (cGAN) training. We introduce a mixed-domain SSIML2 loss function, combining structural similarity and pixel-level accuracy, along with adversarial loss, to enhance the prediction of pressure distributions for subjects lying in a bed. Evaluation results from the benchmark dataset demonstrate that the AttnFnet outperforms existing methods in terms of Structural Similarity Index Measure (SSIM) and quality analysis, providing accurate pressure distribution estimation from a single depth image.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2025.1655199</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2025.1655199</link>
        <title><![CDATA[Automated identification of early to mid-stage Parkinson’s disease using deep convolutional neural networks on static facial images]]></title>
        <pubdate>2025-09-15T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Ni Yang</author><author>Jing Liu</author><author>Lin Wang</author><author>Jiajun Ding</author><author>Lingzhi Sun</author><author>Xianghua Qi</author><author>Yitong Lu</author><author>Wei Yan</author>
        <description><![CDATA[ObjectiveThis study investigates deep convolutional neural networks (CNNs) for automated detection of early to mid-stage Parkinson's disease (PD) from static facial images, aiming to explore non-invasive, cost-effective approaches for early diagnosis and remote monitoring.Methods2,000 facial images were collected from PD patients and healthy controls, followed by data augmentation to expand the dataset to 6,000 images. After randomly dividing the dataset into training and test sets according to 8:2, five CNN architectures were fine-tuned and assessed. Model performance was assessed by accuracy, precision, recall, specificity, F1 score, and area under the ROC and PR curve (AUC). Grad-CAM visualization techniques were applied to identify the discriminative facial regions associated with PD.ResultsResNet18 achieved the best overall performance, yielding an F1 score of 99.67% across metrics. MobileNetV3 also performed robustly, particularly excelling in recall (99.00%), suggesting its suitability for high-sensitivity screening applications. EfficientNetV2 demonstrated stable convergence and competitive classification performance (F1 score: 96.30%), while VGG16 exhibited balanced performance with rapid convergence. Inception-v4 showed relatively lower accuracy and greater variability, indicating a potential risk of overfitting. Grad-CAM heatmaps revealed that the most predictive facial regions across models were concentrated around the eyes, lips, and nose, consistent with PD-related hypomimia.ConclusionCNNs, particularly ResNet18 and MobileNetV3, exhibit significant potential for the automated identification of PD from facial imagery. These models offer promising avenues for developing scalable, non-invasive screening tools suitable for early detection and remote healthcare delivery, providing significant clinical and social value in the context of aging populations.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2025.1644384</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2025.1644384</link>
        <title><![CDATA[Editorial: mHealth and smartphone apps in patient follow-up]]></title>
        <pubdate>2025-07-07T00:00:00Z</pubdate>
        <category>Editorial</category>
        <author>Uffe Kock Wiil</author>
        <description></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2024.1360280</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2024.1360280</link>
        <title><![CDATA[Detection and counting of Leishmania intracellular parasites in microscopy images]]></title>
        <pubdate>2024-08-23T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Lariza María de la Caridad Portuondo-Mallet</author><author>Niurka Mollineda-Diogo</author><author>Rubén Orozco-Morales</author><author>Juan Valentín Lorenzo-Ginori</author>
        <description><![CDATA[ProblemLeishmaniasis is a disease caused by protozoan parasites of the genus Leishmania and has a high prevalence and impact on global health. Currently, the available drugs for its treatment have drawbacks, such as high toxicity, resistance of the parasite, and high cost. Therefore, the search for new, more effective, and safe drugs is a priority. The effectiveness of an anti-leishmanial drug is analyzed through in vitro studies in which a technician manually counts the intracellular form of the parasite (amastigote) within macrophages, which is slow, laborious, and prone to errors.Objective(s)To develop a computational system that facilitates the detection and counting of amastigotes in microscopy images obtained from in vitro studies using image processing techniques.MethodologySegmentation of objects in the microscope image that might be Leishmania amastigotes was performed using the multilevel Otsu method on the saturation component of the hue, saturation, and intensity color model. In addition, morphological operations and the watershed transform combined with the weighted external distance transform were used to separate clustered objects. Then positive (amastigote) objects were detected (and consequently counted) using a classifier algorithm, the selection of which as well as the definition of the features to be used were also part of this research. MATLAB was used for the development of the system.Results and discussionThe results were evaluated in terms of sensitivity, precision, and the F-measure and suggested a favorable effectiveness of the proposed method.ConclusionsThis system can help researchers by allowing large volumes of images of amastigotes to be counted using an automatic image analysis technique.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2024.1372358</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2024.1372358</link>
        <title><![CDATA[Editorial: New technologies improve maternal and newborn safety]]></title>
        <pubdate>2024-05-30T00:00:00Z</pubdate>
        <category>Editorial</category>
        <author>Jieyun Bai</author><author>Yaosheng Lu</author><author>Huishu Liu</author><author>Fang He</author><author>Xiaohui Guo</author>
        <description></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fmedt.2024.1297552</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fmedt.2024.1297552</link>
        <title><![CDATA[Influence of sports on cortical excitability in patients with spinal cord injury: a TMS study]]></title>
        <pubdate>2024-05-15T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Vanessa N. Frey</author><author>Patrick B. Langthaler</author><author>Nora Renz</author><author>Georg Zimmermann</author><author>Christopher Höhn</author><author>Kerstin Schwenker</author><author>Aljoscha Thomschewski</author><author>Alexander B. Kunz</author><author>Yvonne Höller</author><author>Raffaele Nardone</author><author>Eugen Trinka</author>
        <description><![CDATA[BackgroundPatients with spinal cord injury (SCI) show abnormal cortical excitability that might be caused by deafferentation. We hypothesize a reduced short-interval intracortical inhibition preceding movement in patients with SCI compared with healthy participants. In addition, we expect that neuroplasticity induced by different types of sports can modulate intracortical inhibition during movement preparation in patients with SCI.MethodsWe used a reaction test and paired-pulse transcranial magnetic stimulation to record cortical excitability, assessed by measuring amplitudes of motor-evoked potentials in preparation of movement. The participants were grouped as patients with SCI practicing wheelchair dancing (n = 7), other sports (n = 6), no sports (n = 9), and healthy controls (n = 24).ResultsThere were neither significant differences between healthy participants and the patients nor between the different patient groups. A non-significant trend (p = .238), showed that patients engaged in sports have a stronger increase in cortical excitability compared with patients of the non-sportive group, while the patients in the other sports group expressed the highest increase in cortical excitability.ConclusionThe small sample sizes limit the statistical power of the study, but the trending effect warrants further investigation of different sports on the neuroplasticity in patients with SCI. It is not clear how neuroplastic changes impact the sensorimotor output of the affected extremities in a patient. This needs to be followed up in further studies with a greater sample size.]]></description>
      </item>
      </channel>
    </rss>