PLOS digital health最新文献

Decision-making in perinatal management of extremely preterm infants is challenging. Mortality prediction tools may support decision-making. We used population-based routinely entered electronic patient record data from 25,902 infants born between 23+0-27+6 weeks' gestation and admitted to 185 English and Welsh neonatal units from 2010-2020 to develop and internally validate an online tool to predict mortality before neonatal discharge. Comparing nine machine learning approaches, we developed an explainable tool based on stepwise backward logistic regression (https://premoutcome.shinyapps.io/Death/). The tool demonstrated good discrimination (area under the receiver operating characteristics curve (95% confidence interval) of 0.746 (0.729-0.762)) and calibration with superior net benefit across probability thresholds of 10%-70%. Our tool also demonstrated superior calibration and utility performance than previously published models. Acceptable performance was demonstrated in a multinational, external validation cohort of preterm infants. This tool may be useful to support high-risk perinatal decision-making following further evaluation.

Large language models (LLMs) have demonstrated impressive capabilities in medical domains, yet their ability to handle the specialized reasoning patterns required in clinical neurology warrants systematic evaluation. Neurological assessment presents distinctive challenges that combine anatomical localization, temporal pattern recognition, and nuanced symptom interpretation-cognitive processes that are specifically tested in board certification examinations. We developed a comprehensive benchmark comprising 305 questions from Israeli Board Certification Exams in Neurology and classified each along three dimensions of complexity: factual knowledge depth, clinical concept integration, and reasoning complexity. We evaluated ten LLMs of varying architectures and specializations using this benchmark, testing base models, retrieval-augmented generation (RAG) enhancement, and a novel multi-agent system. Our analysis revealed significant performance variation across models and methodologies. The OpenAI-o1 model achieved the highest base performance (90.9% accuracy), while specialized medical models performed surprisingly poorly (52.9% for Meditron-70B). RAG enhancement provided variable benefits across models; substantial improvements for mid-tier models like GPT-4o (80.5% to 87.3%) and smaller models, but limited effectiveness on the highest complexity questions regardless of model size. In contrast, our multi-agent framework-which decomposes neurological reasoning into specialized cognitive functions including question analysis, knowledge retrieval, answer synthesis, and validation-achieved dramatic improvements, especially for mid-range models. The LLaMA 3.3-70B-based agentic system reached 89.2% accuracy compared to 69.5% for its base model, with particularly substantial gains on level 3 complexity questions across all dimensions. External validation on MedQA revealed dataset-specific RAG effects: while RAG improved board certification performance, it showed minimal benefit on MedQA questions (LLaMA 3.3-70B: + 1.4% vs + 3.9% on board exams), reflecting alignment between our specialized neurology textbook and board examination content rather than the broader medical knowledge required for MedQA. Most notably, the multi-agent approach transformed inconsistent subspecialty performance into remarkably uniform excellence, effectively addressing the neurological reasoning challenges that persisted even with RAG enhancement. We further validated our approach using an independent dataset comprising 155 neurological cases extracted from MedQA. The results confirm that structured multi-agent approaches designed to emulate specialized cognitive processes significantly enhance complex medical reasoning offering promising directions for AI assistance in challenging clinical contexts.

The term "artificial" implies an inherent dichotomy from the natural or organic. However, AI, as we know it, is a product of organic ingenuity-designed, implemented, and iteratively improved by human cognition. The very principles that underpin AI systems, from neural networks to decision-making algorithms, are inspired by the organic intelligence embedded in human neurobiology and evolutionary processes. The path from "organic" to "artificial" intelligence in digital health is neither mystical nor merely a matter of parameter count-it is fundamentally about organization and adaption. Thus, the boundaries between "artificial" and "organic" are far less distinct than the nomenclature suggests.

Machine learning (ML) models are increasingly applied to predict body mass index (BMI) and related outcomes, yet their fairness across socioeconomic and caste groups remains uncertain, particularly in contexts of structural inequality. Using nationally representative data from more than 55,000 adults aged 45 years and older in the Longitudinal Ageing Study in India (LASI), we evaluated the accuracy and fairness of multiple ML algorithms-including Random Forest, XGBoost, Gradient Boosting, LightGBM, Deep Neural Networks, and Deep Cross Networks-alongside logistic regression for predicting underweight, overweight, and central adiposity. Models were trained on 80% of the data and tested on 20%, with performance assessed using AUROC, accuracy, sensitivity, specificity, and precision. Fairness was evaluated through subgroup analyses across socioeconomic and caste groups and equity-based metrics such as Equalized Odds and Demographic Parity. Feature importance was examined using SHAP values, and bias-mitigation methods were implemented at pre-processing, in-processing, and post-processing stages. Tree-based models, particularly LightGBM and Gradient Boosting, achieved the highest AUROC values (0.79-0.84). Incorporating socioeconomic and health-related variables improved prediction, but fairness gaps persisted: performance declined for scheduled tribes and lower socioeconomic groups. SHAP analyses identified grip strength, gender, and residence as key drivers of prediction differences. Among mitigation strategies, Reject Option Classification and Equalized Odds Post-processing moderately reduced subgroup disparities but sometimes decreased overall performance, whereas other approaches yielded minimal gains. ML models can effectively predict obesity and adiposity risk in India, but addressing bias is essential for equitable application. Continued refinement of fairness-aware ML methods is needed to support inclusive and effective public-health decision-making.

Online/digital health literacy is important for individuals to evaluate the influence of such input in their care and consent for treatment. The purpose of this systematic review is to examine the digital health literacy level among adults in studies that used the eHealth Literacy Scale (eHEALS) as a measure of digital health literacy. The authors searched Google Scholar, PubMed, Scopus, and Web of Science for evidence following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Statement, 2020 (PRISMA). Included were articles in which the researchers evaluated the level of digital health literacy using eHEALS, were peer reviewed, written in English or in which English translation was provided, and were published between 2020-2025. There were 200 articles initially identified in the search, 180 were excluded resulting in a sample of 20 publications. EHEALS scores, with possibilities from 8-40, had a weighted mean of 24.3 (95%CI: 17.1-31.6). The lowest mean score was 12.57; and the highest mean score was 35.1. The highest eHEALS score was from a qualitative interview study. Nine other studies reported overall means ≥ 30. There were three with eHEALS scores below 20. Globally, there is a wide range of reported digital health literacy levels. It is critical that the public gains skill and confidence in digital health literacy for healthcare decisions. The results of this study provide evidence of a large range of digital health literacy.

In the healthcare domain, images play a pivotal role in clinical diagnoses, treatment planning, surgical procedures, and epidemiological insights. Nevertheless, challenges such as limited experience among healthcare professionals, risk of misdiagnosis and subjective interpretation, and factors like stress and fatigue may jeopardize the precision with which patients are assessed. In this regard, professionals in the field of Dentistry face analogous challenges given that distinguishing anatomical structures in dental imaging requires expert interpretation and precise analysis. Convolutional Neural Networks (CNNs) offer promising opportunities to analyze images during patient care and can enhance diagnostic accuracy and clinical decision-making, benefiting both patients and healthcare providers. Here, we aimed to develop a specialized analyzer for digital dental radiography, that focuses on numbering teeth and detecting tooth cavities. The system is designed to achieve high precision, recall, accuracy, specificity, and F1-score, to ensure that diagnosis is reliable and accurate. In this study, we specifically explore Inception-v3 and InceptionResNet-v2 to discern cavitated teeth and tooth positions in dental panoramic radiographic images (PANs). On the basis of 935 PANs sourced from routine patient care, annotated by dentists at the Faculty of Dentistry of Ribeirão Preto in Brazil, our approach achieved precision of 0.98, recall of 0.98, accuracy of 0.998, specificity of 0.999 and F1-score of 0.98 for tooth numbering. Concerning identification of cavitated teeth, our approach reached precision of 0.96, recall of 0.91, accuracy of 0.94, specificity of 0.96 and F1-score of 0.94. By addressing the critical challenges and reaching high performance, our study serves as a benchmark that relates innovative research and real-world applications, fostering advancements in dental diagnosis. The performance reported herein demonstrates that our initiatives can modulate image analysis tasks and select a more suitable CNN for the job.

Despite the millions of individuals struggling with persistent symptoms, Long COVID has remained difficult to diagnose due to limited objective biomarkers, often leading to underdiagnosis or even misdiagnosis. To bridge this gap, we investigated the potential of utilizing wearable sensor data to aid in the diagnosis of Long COVID. We analyzed longitudinal heart rate (HR) data from 126 individuals with acute SARS-CoV-2 infections to develop machine learning models capable of predicting Long COVID status using derived HR features, symptom features, or a combination of both feature sets. The HR features were derived across six analytical categories, including time-domain, Poincaré nonlinear, raw signal, Kullback-Leibler (KL) divergence, variational mode decomposition (VMD), and the Shannon energy envelope (SEE), enabling the capture of heart rate dynamics over various temporal scales and the quantification of day-to-day shifts in HR distributions. The symptom features used in the final models included chest pain, vomiting, excessive sweating, memory loss, brain fog, heart palpitations, and loss of smell. The combined HR- and symptom-feature model demonstrated robust predictive performance, achieving an area under the Receiver Operating Characteristic curve (ROC-AUC) of 95.1% and an area under the Precision-Recall curve (PR-AUC) of 85.9%. These values represent a significant improvement of approximately 5% in both the ROC-AUC and PR-AUC over the symptoms-only model. At the population level, this improvement in discrimination could lead to clinically meaningful reductions in misclassification and improved patient outcomes, achieved through a non-invasive diagnostic tool. These findings suggest that wearable HR data could be used to derive an objective biomarker for Long COVID, thereby enhancing diagnostic precision.

Induction of labor (IOL) is a common yet complex clinical procedure associated with varying risks, including cesarean section (CS). Accurate prediction models may help support more informed, personalized decision-making. This study aimed to develop and validate an explainable machine learning prediction model for CS following IOL. We used population-based administrative perinatal datasets from two Australian states (New South Wales (NSW) and Queensland) covering all births between 2016 and 2019 for model development. Temporal validation was conducted using 2020 births from NSW, and geographical validation using 2016-2018 births from Victoria. We included women with singleton, cephalic, term, live births who attempted IOL and had no prior CS. Seven models (logistic regression, random forest, gradient boosting, LightGBM, XGBoost, CatBoost, and AdaBoost) were developed with hyperparameter tuning and feature selection. Performance was assessed using the area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve, calibration plot (overall and across sociodemographic subgroups), decision curve analysis, Brier Score, and model parsimony. SHAP (SHapley Additive exPlanations) values were used to explain predictor contributions. A total of 180,700 women were included in model development (mean age 31 ± 5 years; CS = 20.8%). The optimal model, developed using XGBoost with ten predictors, achieved AUROCs of 0.76 (95% CI: 0.75-0.77) and 0.75 (95% CI: 0.74-0.76) in temporal (n = 14,527; CS = 22.5%) and geographical (n = 14,755; CS = 19.0%) validations, respectively. The most influential predictors were nulliparity, pre-pregnancy body mass index, and maternal age, while diabetes and hypertension (pre-existing or pregnancy-related) contributed least. Women with higher predicted CS probabilities had increased inpatient costs and maternal morbidity, regardless of actual mode of birth. The final model is accessible via an interactive web application (https://csai-8ccf2690242c.herokuapp.com/). This model demonstrates strong predictive performance using routinely collected maternal factors. Further co-design and implementation research is needed before potential clinical adoption.

Chronic disease management is a burden for many patients. Digital health tools (DHTs) can leverage technology to rapidly develop and disseminate interventions to alleviate obstacles faced and promote self-care. Primary care physicians (PCPs) are most directly involved in the care of chronic disease patients; however, their perspective is often overlooked. To develop an effective DHT for chronic disease management, PCP attitudes are critical to ensure improved patient integration, adoption and care outcomes. The purpose of this rapid review is to explore and identify PCPs' perspectives and attitudes regarding DHTs for chronic disease management and generate major themes from our findings using key literature. The themes will be used to guide DHT creators, clinicians and policy makers on adoption and implementation considerations. We conducted a rapid review of primary qualitative research between 2000 and 2022. Two reviewers, independently, conducted study screening, selection, and data abstraction. The themes identified in the articles were extracted and presented narratively. The data was analyzed using NVIVO12 software. Braun and Clarke's deductive thematic analysis was used, and the themes identified were extracted and presented narratively. Nine qualitative research studies met the inclusion criteria. Themes were classified into two major categories: physician-patient relationship and physician-technology relationship. Within these, seven subcategories were identified: (1) Increased Physician Workload, (2) Data Capture & Data Quality, (3) Evidence-Based Care, (4) Education and Training, (5) Liability, (6) Patient Interactions, and (7) Patient Empowerment and Suitability. DHT creators/endorsers need to consider how DHTs affect the patient-physician relationship and the physician-technology relationship as this affects how PCPs perceive DHTs. PCPs' perspectives must be taken into consideration to promote self-care for patients living with chronic diseases.