Smart Health最新文献

{"title":"Enhancing latent space representation in Adversarial Autoencoders for exercise recognition: A data augmentation perspective using low-cost sensors","authors":"Vincent Hernandez,&nbsp;Gentiane Venture","doi":"10.1016/j.smhl.2026.100635","DOIUrl":"10.1016/j.smhl.2026.100635","url":null,"abstract":"<div><div>Exercise monitoring in the context of Human Activity Recognition (HAR) is essential for delivering immediate feedback and facilitating the analysis of movement. When combined with data dimensionality reduction, it can offer deeper insights into movement patterns, thereby aiding in the development of training programs. This study investigates the impact of data augmentation and the number of participants in the training data on the accuracy of 2D latent space representations generated by an Adversarial AutoEncoder (AAE).</div><div>In this study, data from the Wii Balance Board (WiiBB) and Inertial Measurement Units (IMUs) placed on each forearm and hip were collected from 20 participants. Experiments were performed for upper and lower body exercises, with the accuracy of the latent space representation analyzed by varying the number of participants in the training set from 2 to 12 with and without data augmentation.</div><div>The results demonstrate that the incorporation of data augmentation significantly improves the accuracy of the latent space representation of AAE. For example, using only two participants in the training set, data augmentation improves test accuracy by 10.2% and 4.4% for WiiBB data and IMU, respectively, for lower body exercises, while upper body exercises showed improvements of 6.8% and 3.1% respectively.</div><div>These findings show how data augmentation can mitigate the limitations of small training datasets significantly improving latent space representations for HAR applications. This study emphasizes the importance of combining data augmentation strategies and sensor types to achieve reliable and interpretable results in remote rehabilitation systems.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"40 ","pages":"Article 100635"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146081798","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Cold stress test potential in diabetic foot diagnosis using a thermal camera and Artificial Intelligence","authors":"Asma Aferhane ,&nbsp;Hafid Elfahimi ,&nbsp;Hassan Douzi ,&nbsp;Rachid Harba","doi":"10.1016/j.smhl.2026.100644","DOIUrl":"10.1016/j.smhl.2026.100644","url":null,"abstract":"<div><div>This paper investigates the potential use of a cold stress test for diabetic foot (DF) diagnosis. Thermal foot images were taken freehandly before immersion in cold water (image <span><math><msub><mrow><mi>T</mi></mrow><mrow><mn>0</mn></mrow></msub></math></span>), followed by a second image after 10 min after removing the feet from the cold water (image <span><math><msub><mrow><mi>T</mi></mrow><mrow><mn>0</mn></mrow></msub></math></span> and <span><math><msub><mrow><mi>T</mi></mrow><mrow><mn>10</mn></mrow></msub></math></span>), using a low-cost thermal camera integrated into a smartphone. A fully automated AI-based process was employed to calculate the mean absolute point-to-point temperature difference between the two feet. The AI system segmented both feet and performed affine registration for contralateral and multitemporal alignment (<span><math><msub><mrow><mi>T</mi></mrow><mrow><mn>0</mn></mrow></msub></math></span> and <span><math><msub><mrow><mi>T</mi></mrow><mrow><mn>10</mn></mrow></msub></math></span>). A total of 145 diabetic subjects without a history of ulcers were recruited at Hospital National Dos de Mayo. Each participant received a verbal description of the test objectives and underwent a comprehensive foot examination of both feet assigned to one of the following three groups: the low-risk group (R0), the medium-risk group (R1), and the high-risk group (R2). Results show that multitemporal temperature variation effectively differentiates between R0 and R1 patients, achieving a sensitivity of 82% and a specificity of 70% (<span><math><mrow><mi>p</mi><mo>&lt;</mo><mn>0</mn><mo>.</mo><mn>01</mn></mrow></math></span>). When combined with contralateral variation, the sensitivity increases to 86%. These findings underscore the value of multitemporal analysis in distinguishing risk levels for ulcer development. The system, consisting of a smartphone, a dedicated low-cost thermal camera, and fully automated AI software, shows great promise as a practical, low-cost tool for diabetic foot screening.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"40 ","pages":"Article 100644"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146191303","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Revisiting NLP applications using MIMIC data: A systematic literature review toward advancing clinical processes and workflows in the Era of LLMs","authors":"Jiawei Wu ,&nbsp;Nazmus Sakib ,&nbsp;Fahim Islam Anik ,&nbsp;K M Sajjadul Islam ,&nbsp;Kevin Chovanec ,&nbsp;Praveen Madiraju ,&nbsp;Sheikh Iqbal Ahamed","doi":"10.1016/j.smhl.2026.100636","DOIUrl":"10.1016/j.smhl.2026.100636","url":null,"abstract":"<div><div>This review comprehensively examines the role of Natural Language Processing (NLP) in medical decision-making using the MIMIC dataset. It focuses on three critical dimensions: exploring the diverse NLP use cases that utilize MIMIC data to support medical decision-making, identifying the key input factors that enhance the predictive performance of NLP models in clinical contexts, and evaluating how current NLP applications can be critically reimagined to transform clinical workflows, especially with the advent of Large Language Models (LLMs). Through a systematic literature search across IEEE, PubMed, Scopus, and ACM, we identified peer-reviewed studies employing NLP to address critical medical challenges, including risk prediction, disease progression modeling, patient phenotyping, and drug concept extraction. These studies are further categorized into domains such as patient representation, topic modeling, knowledge extraction, sentiment analysis, and text classification, showcasing a variety of approaches, predominantly deep learning and traditional machine learning techniques. Our findings highlight the critical role of input factors—such as temporal trends, demographic variables, and clinical events—in shaping the predictive capabilities of NLP models. Furthermore, the review underscores the importance of interrogating existing NLP applications to ensure alignment with clinical workflows, emphasizing interpretability, equity, and practical integration into healthcare systems. By exploring the transformative potential of NLP and LLMs, this study offers insights into optimizing medical decision-making and advancing precision medicine, ultimately paving the way for innovative, data-driven healthcare solutions.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"40 ","pages":"Article 100636"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146191302","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Novel ECG signal classification based on Minkowski distance to enhance intelligent arrhythmia detection systems","authors":"Rawaa R. Rfys ,&nbsp;Dhiah Al-Shammary ,&nbsp;Mustafa Noaman Kadhim ,&nbsp;Ayman Ibaida","doi":"10.1016/j.smhl.2026.100645","DOIUrl":"10.1016/j.smhl.2026.100645","url":null,"abstract":"<div><div>In this study, an efficient approach utilizing the Minkowski distance metric is introduced for the identification of arrhythmias through the examination of electrocardiogram (ECG) signals. Current machine learning classifiers, such as Naive Bayes (NB), Decision Trees (DT), Support Vector Machines (SVM), K-Nearest Neighbors (KNN), and Random Forest (RF), often face challenges in handling high-dimensional features and capturing complex relationships within signals, which are critical for accurate arrhythmia detection. The proposed method employs the Minkowski metric as a classifier to measure the distance between ECG signals, leveraging its sensitivity to capture relationships and matching patterns within the data. This approach enables the accurate determination of the state of ECG signals, facilitating precise arrhythmia detection. Particle Swarm Optimization (PSO) is employed to select and optimize features, improving the proposed classifier's capability to detect subtle patterns within ECG signals. The study utilizes the widely recognized MIT-BIH Arrhythmia dataset for evaluation. Performance metrics, including accuracy, precision, recall, and F1-score, are employed to assess the effectiveness of the proposed models. Experimental results demonstrate that the proposed approach outperforms traditional classifiers such as NB, SVM, KNN, RF, and DT, achieving accuracy rates of up to 91.25% and 93.75% without and with PSO, respectively. These findings highlight the potential of the Minkowski-based technique in improving arrhythmia detection and addressing the limitations of traditional ML methods. With its ability to support early arrhythmia diagnosis, the proposed approach presents a valuable and efficient solution for medical practitioners, contributing to better clinical outcomes.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"40 ","pages":"Article 100645"},"PeriodicalIF":0.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146191383","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Mental health risk prediction in autoimmune patients in primary care: A class-sensitive machine learning approach","authors":"Mariachiara Di Cosmo ,&nbsp;Sara Campanella ,&nbsp;Michele Bernardini ,&nbsp;Adriano Mancini ,&nbsp;Lorenzo Palma","doi":"10.1016/j.smhl.2026.100634","DOIUrl":"10.1016/j.smhl.2026.100634","url":null,"abstract":"<div><div>Patients with autoimmune diseases are at increased risk of developing psychological conditions such as anxiety and depression. Timely detection of these mental health disorders is essential for effective intervention, yet remains difficult due to the overlap and subtle presentation of symptoms. In this study, we propose a machine learning (ML) framework for early prediction of anxiety and depression in autoimmune patients using routine clinical and demographic features in primary care. We develop an XGBoost-based model enhanced with custom loss functions specifically designed to handle class imbalance and semantic similarity between mental health outcomes. Experimental results demonstrate that the proposed penalty terms improve standard XGBoost performance in identifying minority classes and minimizing confusion between similar categories, by reaching a maximum statistically significant macro-Recall gain of 4.03% (p<span><math><mo>&lt;</mo></math></span>0.05) for the Logit Distance Exponential Penalty term. This work contributes to developing data-driven tools for mental health risk assessment, with potential applications in personalized care and digital screening for vulnerable clinical populations.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"39 ","pages":"Article 100634"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146037198","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Patient-specific deep offline artificial pancreas for blood glucose regulation in type 1 diabetes","authors":"Yixiang Deng ,&nbsp;Kevin Arao ,&nbsp;Christos S. Mantzoros ,&nbsp;George Em Karniadakis","doi":"10.1016/j.smhl.2026.100633","DOIUrl":"10.1016/j.smhl.2026.100633","url":null,"abstract":"<div><div>Due to insufficient insulin secretion, patients with type 1 diabetes mellitus (T1DM) are prone to blood glucose fluctuations ranging from hypoglycemia to hyperglycemia. While dangerous hypoglycemia may lead to coma immediately, chronic hyperglycemia increases patients’ risks for cardiorenal and vascular diseases in the long run. In principle, an artificial pancreas – a closed-loop insulin delivery system requiring patients to manually input insulin dosage according to the upcoming meals – could supply exogenous insulin to control the glucose levels and hence reduce the risks from hyperglycemia. However, insulin overdosing in some type 1 diabetic patients, who are physically active, can lead to unexpected hypoglycemia beyond the control of the common artificial pancreas. Therefore, it is important to take into account the glucose decrease due to physical exercise when designing the next-generation artificial pancreas. In this work, we develop a framework integrating systems biology-informed neural networks (SBINN), deep reinforcement learning (RL) algorithms, and T1DM data collected from wearable devices, to automate insulin dosing for patients. In particular, we build patient-specific computational models using SBINN to mimic the glucose-insulin dynamics for a few patients from the dataset, by simultaneously considering patient-specific carbohydrate intake and physical exercise intensity. Our patient-specific artificial pancreas, based on two deep RL algorithms, provided better insulin dosage, leading to safer glucose levels compared to those in the original dataset.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"39 ","pages":"Article 100633"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146037199","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"AI-powered play assessment approach using video language models: A feasibility study","authors":"Amiya Waldman-Levi ,&nbsp;Dengyi Liu ,&nbsp;Chana Cunin ,&nbsp;Vanessa Murad ,&nbsp;Honggang Wang","doi":"10.1016/j.smhl.2025.100632","DOIUrl":"10.1016/j.smhl.2025.100632","url":null,"abstract":"<div><div>Social-behavioral observation-based assessments are integral to clinical practice and research in health and psychology. However, these methods are often time-intensive and prone to human error, bias, and inconsistency. Deep neural networks (DNNs), a class of machine learning models, offer distinct advantages in healthcare assessments due to their advanced ability to process complex data with greater accuracy than traditional approaches, such as tree-based models. We developed innovative AI-powered software that integrates DNNs with computer vision techniques to analyze parent–child joint play interactions. Our objective was to utilize Video Large Language Models (Video LLMs) to automatically score a validated parent–child play scale (PC-SCP). Using convenience sampling, we recruited 37 mother–child dyads, including both neurotypical and neurodiverse children aged 1–6 years. Following eligibility screening and consent procedures, data collection involved recording 10–15-minute videos of parent–child interactions at home, which were then manually scored using the Parent/Caregiver Support of Children’s Playfulness criteria. We trained and evaluated several DNN models, including Qwen2.5VL, to identify and track parental behaviors in video frames and automatically score the interactions based on the PC-SCP guidelines. Among the models evaluated, our fine-tuned Qwen2.5VL achieved an accuracy of 38.2% and a best-5 accuracy of 61.3%, demonstrating promising potential for automating the scoring of social-behavioral assessments. This novel application of AI represents a significant advancement toward more efficient, objective, and consistent behavioral assessments in both clinical and research settings.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"39 ","pages":"Article 100632"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925691","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"From pre-treatment to post-operative care: Leveraging digital twins for precision surgery transformation","authors":"Farnaz Dehghan ,&nbsp;Seyed Mojtaba Hosseini Bamakan ,&nbsp;Mahboubeh Mirzabagheri ,&nbsp;Alireza Naser Sadrabadi","doi":"10.1016/j.smhl.2025.100620","DOIUrl":"10.1016/j.smhl.2025.100620","url":null,"abstract":"<div><div>Digital twin (DT) technology offers transformative potential for the healthcare sector by enabling virtual models that support data-driven diagnostics, treatment planning, and preventive care. This paper presents a robust framework aimed at unlocking the full potential of digital twins across all facets of healthcare delivery. The process begins with the pre-treatment phase, during which patient data from multiple sources are integrated to construct digital twin models, enabling predictive analytics and aiding in surgical planning. Moving into surgery, DTs prove invaluable by simulating procedures, offering real-time guidance, and facilitating remote collaboration among medical experts. In the postoperative phase, they support personalized care through continuous patient monitoring and assessment of interventions using virtual models. However, realizing the complete potential of digital replicas in surgery faces obstacles such as data integration, interoperability, clinical applicability, and ethical concerns, demanding careful consideration. This paper explores the key elements of a distributed and reliable framework tailored for deploying DTs in precision surgery, while also identifying areas for further exploration and addressing unresolved issues. The insights presented aim to catalyze targeted innovation and foster interdisciplinary partnerships to effectively harness this emerging technology for healthcare transformation.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"39 ","pages":"Article 100620"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145624937","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Integrating AI into healthcare systems: A multivocal literature review","authors":"Giulio Mallardi,&nbsp;Fabio Calefato,&nbsp;Luigi Quaranta,&nbsp;Filippo Lanubile","doi":"10.1016/j.smhl.2025.100631","DOIUrl":"10.1016/j.smhl.2025.100631","url":null,"abstract":"<div><div>The integration of artificial intelligence (AI) into healthcare systems promises to improve patient care, enhance operational efficiency, and facilitate personalized medicine. The goal of this paper is to provide a comprehensive review of the current challenges that hinder the seamless adoption of AI in healthcare. Additionally, the paper aims to delineate the best practices for achieving optimal integration of AI within the medical domain. To achieve these objectives, we employ a Multivocal Literature Review (MLR), a systematic literature review methodology that incorporates both peer-reviewed publications and non-peer-reviewed sources, including technical blog posts and white papers. Substantial evidence in the literature points to challenges related to data quality, model bias, interoperability, patient privacy, and the susceptibility of AI systems to adversarial attacks. Additionally, there is growing awareness of challenges such as the distributional shift between training and production data, as well as the critical need for continuous monitoring and retraining of AI models within dynamic clinical settings. Based on our review, we advocate for the adoption of best practices aimed at mitigating the identified challenges, including rigorous model evaluation, standardization of data practices, and promotion of interdisciplinary collaboration. Furthermore, we emphasize the need for responsible AI that aligns with principles of fairness, transparency, security, and reliability, underscoring the importance of multi-stakeholder engagement.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"39 ","pages":"Article 100631"},"PeriodicalIF":0.0,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790386","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Smart IoT-Cloud Healthcare System for Heart Disease Detection using Hierarchical Auto-Associative Polynomial Convolutional Neural Network","authors":"Preethi Sambandam Raju ,&nbsp;Dexter Woodward ,&nbsp;R. Giri Prasad ,&nbsp;Elangovan Muniyandy","doi":"10.1016/j.smhl.2025.100612","DOIUrl":"10.1016/j.smhl.2025.100612","url":null,"abstract":"<div><div>Real-time patient health monitoring and remote management of health seem to be getting closer to revolutionizing healthcare with the swift growth of IoT and cloud-based services. Considering heart disease as the biggest killer, timely and precise prediction systems are also needed to enhance patient outcomes. Current systems lack real-time data processing, security vulnerabilities, and challenges in optimizing prediction accuracy. This paper proposes a Smart IoT-Cloud Healthcare Monitoring System for the prediction of heart illness based on a Hierarchical Auto-Associative Polynomial Convolutional Neural Network (HAAP-CNN-TTAO) to address these inefficiencies. The system employs IoT sensors to collect heart-related data, It is pre-processed with Shape-Aware Mesh Normal Filtering to guarantee high-quality input and remove noise. Feature extraction is performed using a Modified ResNet-152 Network, followed by heart disease prediction with HAAP-CNN. To overcome the limitations of traditional HAAP-CNN in parameter optimization, the Triangulation Topology Aggregation Optimizer (TTAO) is incorporated, enhancing prediction accuracy and scalability. Additionally, the Lightweight Homomorphic Cryptographic Algorithm (LHCA) is employed to secure patient data during transmission. Experimental results validate the system's efficacy, achieving a remarkable 98.7 % accuracy, 98.1 % precision, and 97.5 % recall, while reducing computational time by 22 % compared to existing methods. The proposed system effectively addresses critical challenges in real-time health monitoring, providing a strong and expandable solution for early identification of heart disease, greatly enhancing patient care and results.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"38 ","pages":"Article 100612"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145060663","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}