PLOS digital health最新文献

{"title":"Predicting patient enrollment in a telephone-based principal care management service using topic modeling.","authors":"Annisa Marlin Masbar Rus, Julie S Ivy, Min Chi, Mitchell Plyler, Elaine Wells-Gray, Maria E Mayorga","doi":"10.1371/journal.pdig.0000992","DOIUrl":"10.1371/journal.pdig.0000992","url":null,"abstract":"<p><p>Diabetic Retinopathy (DR) is a complication related to diabetes that can lead to vision impairment. To assist DR patients, a care management company provides a telephone-based principal care management (PCM) service, which includes care coaching and other services to reduce barriers to care for patients with DR. Despite its benefits, enrollment in the program is suboptimal. This study developed predictive models using call transcripts to investigate factors associated with patient enrollment in the PCM service. We analyzed transcripts of calls made during the enrollment process (prior to enrollment) and feature-engineered the call metadata (i.e., transcript length, number of calls, time between calls, customer and agent sentiment). In addition, we extracted topics discussed in the transcripts using Structural Topic Modeling (STM) and converted them into vector representations. Utilizing call metadata alongside topics, we developed three classification models (call metadata, topic-based, and topic+metadata) to predict patient enrollment, with the latter demonstrating superior performance. The topic+metadata classification model outperformed the other two models in distinguishing between patient enrollment and non-enrollment, with AUC values ranging from 0.81 to 0.99 across models using 3 to 15-topics. The findings suggest that proactively offering to schedule an appointment after the program benefits explanation leads to a higher odds of enrollment. When the scheduling portion of the conversation is not considered, agents should cover all parts of the script over multiple calls. Additionally, agents who explain the program and maintain longer intervals between calls have higher odds of patient enrollment, suggesting that there is value in allowing patients adequate time to reflect between calls. These findings offer valuable insights for agents to evaluate their strategies in patient enrollment. As the first point of contact, enrollment agents play a crucial role in determining whether patients can benefit from care coordination and management programs.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0000992"},"PeriodicalIF":7.7,"publicationDate":"2025-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12445525/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145088413","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Digital access, transportation, and women's empowerment in breast cancer screening uptake among Cambodian women: Analysis of the Cambodia demographic and health survey 2021-2022.","authors":"Samnang Um, Channnarong Phan, Daraden Vang, Tharuom Ny, Sothy Heng","doi":"10.1371/journal.pdig.0000976","DOIUrl":"10.1371/journal.pdig.0000976","url":null,"abstract":"<p><p>Breast cancer incidence is increasing globally, and it is the third leading cause of morbidity and mortality among women in Cambodia. This study explores how access to digital tools, media exposure, transportation, travel time to health facilities, and autonomy in health decisions relate to breast cancer screening among Cambodian women aged 15-49. The study used nationally representative, cross-sectional data from the Cambodia Demographic and Health Survey (CDHS) 2021-2022. After excluding 204 women who were unaware of breast or cervical cancer screening, the final weighted sample comprised 19,292 participants. The outcome was whether a woman had ever received a breast examination from a healthcare provider, encompassing clinical breast examinations (CBEs) and imaging techniques, such as mammograms. Multivariable logistic regression, adjusted for demographic and socioeconomic characteristics, was used. Only 10.9% (95% CI: 9.7%-11.6%) of women had undergone a breast exam. Exposure to multiple forms of media was associated with a higher odds of screening (AOR = 1.47; 95% CI: 1.13-1.91). Phone ownership-both non-smartphone (AOR = 1.35; 95% CI: 1.03-1.78) or smartphone (AOR = 1.37; 95% CI: 1.03-1.82)-was also positively associated. In contrast, longer travel times of over 30 minutes (AOR = 0.55; 95% CI: 0.39-0.78) and a lack of autonomy in healthcare decisions (AOR = 0.70; 95% CI: 0.52-0.94) were associated with reduced screening. Wealthier women had greater odds of being screened (AOR = 1.86; 95% CI: 1.40-2.48). These findings highlight the need for health initiatives that use digital communication to reach and emphasize the importance of improving transportation, and support women's decision-making to increase screening rates in Cambodia.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0000976"},"PeriodicalIF":7.7,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12443266/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145082647","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Design approaches for developing quality checklists in healthcare organizations: A scoping review.","authors":"Elizabeth Kwong, Amy Cole, Elizabeth Byrd, Dorothy Sippo, Fei Yu, Karthik Adapa, Christopher M Shea, Carlton Moore, Shiva Das, Lukasz Mazur","doi":"10.1371/journal.pdig.0001015","DOIUrl":"10.1371/journal.pdig.0001015","url":null,"abstract":"<p><p>Quality checklists have demonstrated benefits in healthcare and other high-reliability organizations, but there remains a gap in the understanding of design approaches and levels of stakeholder engagement in the development of these quality checklists. This scoping review aims to synthesize the current knowledge base regarding the use of various design approaches for developing quality checklists in healthcare. Secondary objectives are to explore theoretical frameworks, design principles, stakeholder involvement and engagement, and characteristics of the design methods used for developing quality checklists. The review followed the Preferred Reporting Items for Systematic Reviews 2020 checklist. Seven databases (PubMed, APA PsycInfo, CINAHL, Embase, Scopus, ACM Digital Library, and IEEE Xplore) were searched for studies using a comprehensive search strategy developed in collaboration with a health sciences librarian. Search terms included \"checklist\" and \"user-centered design\" and their related terms. The IAP2 Spectrum of Participation Framework was used to categorize studies by level of stakeholder engagement during data extraction. Twenty-nine studies met the inclusion criteria for this review. Twenty-three distinct design methods were identified that were predominantly non-collaborative in nature (e.g., interviews, surveys, and other methods that involved only one researcher and one participant at a given time). Analysis of the levels of stakeholder engagement revealed a gap in studies that empowered their stakeholders in the quality checklist design process. Highly effective, clear, and standardized methodologies are needed for the design of quality checklists. Future work needs to explore how stakeholders can be empowered in the design process, and how different levels of stakeholder engagement might impact implementation outcomes.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0001015"},"PeriodicalIF":7.7,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12440183/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145076788","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Explainable cluster-based learning for prediction of postprandial glycemic events and insulin dose optimization in type 1 diabetes.","authors":"Najib Ur Rehman, Ivan Contreras, Aleix Beneyto, Josep Vehi","doi":"10.1371/journal.pdig.0000996","DOIUrl":"10.1371/journal.pdig.0000996","url":null,"abstract":"<p><p>Effective management of postprandial glycemic excursions in type 1 diabetes requires accurate prediction of adverse events and personalized insulin adjustments informed by interpretable models. This study presents an explainable dual-prediction framework that simultaneously forecasts postprandial hypoglycemia and hyperglycemia within a 4-hour window using cluster-personalized ensemble models. Glycemic profiles were identified through a hybrid unsupervised approach combining self-organizing maps and k-means clustering, enabling the training of specialized random forest classifiers. The system outperformed baseline models on both real-world and simulated datasets, achieving high performance (AUC = 0.84 and 0.93; MCC = 0.47 and 0.73 for hypo- and hyperglycemia, respectively). Model interpretability was addressed using global (SHAP) and local (LIME) explanations, while interaction analysis revealed the non-linear effects of carbohydrate intake and insulin bolus combinations. An insulin adjustment module further refined pre-meal bolus recommendations based on predicted risk. Simulated evaluations confirmed improved postprandial time-in-range and reduced hypoglycemia without excessive hyperglycemia. These results underscore the potential of profile-driven and explainable machine learning approaches to support safer, individualized diabetes care.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0000996"},"PeriodicalIF":7.7,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12440209/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145076844","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Classifying simulated gait impairments using privacy-preserving explainable artificial intelligence and mobile phone videos.","authors":"Lauhitya Reddy, Ketan Anand, Shoibolina Kaushik, Corey Rodrigo, J Lucas McKay, Trisha M Kesar, Hyeokhyen Kwon","doi":"10.1371/journal.pdig.0001004","DOIUrl":"10.1371/journal.pdig.0001004","url":null,"abstract":"<p><p>Accurate diagnosis of gait impairments is often hindered by subjective or costly assessment methods, with current solutions relying on either expensive multi-camera equipment or subjective clinical observation. There is a critical need for accessible, objective tools that can aid in gait assessment while preserving patient privacy. In this work, we present a mobile phone-based, privacy-preserving artificial intelligence (AI) system for classifying gait impairments that leverages a novel dataset of 743 videos capturing seven distinct gait types. The dataset consists of frontal and sagittal views of clinicians simulating normal gait and six types of pathological gait (circumduction, Trendelenburg, antalgic, crouch, Parkinsonian, and vaulting), recorded using standard mobile phone cameras. Our system achieved 86.5% accuracy using combined frontal and sagittal views, with sagittal views generally outperforming frontal views except for specific gait types like circumduction. Model feature importance analysis revealed that frequency-domain features and entropy measures were critical for classification performance. Specifically, lower limb keypoints proved most important for classification, aligning with clinical understanding of gait assessment. These findings demonstrate that mobile phone-based systems can effectively classify diverse gait types while preserving privacy through on-device processing. The high accuracy achieved using simulated gait data suggests their potential for rapid prototyping of gait analysis systems, though clinical validation with patient data remains necessary. This work represents a significant step toward accessible, objective gait assessment tools for clinical, community, and tele-rehabilitation settings.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0001004"},"PeriodicalIF":7.7,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12440163/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145076787","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Digital healthcare interventions to support parents with acutely ill children at home: A systematic review.","authors":"Matthew C Carey, Jane Peters, Anna Chick, Bernie Carter, Lucy Bray, Damian Roland, Sarah Neill","doi":"10.1371/journal.pdig.0000998","DOIUrl":"10.1371/journal.pdig.0000998","url":null,"abstract":"<p><p>Short lived acute illness in children is common, yet their parents often feel uncertain about recognising signs symptoms of acute illness and knowing when to seek medical intervention. This has led to seeking unscheduled or delayed support. Digital and mobile technologies are being used to support individuals with healthcare needs, known as digital health interventions. Parents have access to digital health interventions that provide information regarding children's health, yet there is limited exploration of how these are used to support decision-making when caring for acutely ill children. This systematic review was undertaken to explore digital interventions to support parents with acutely ill children at home. Studies were identified by following the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines. A search of five databases (MEDLINE, CINAHL, Embase, PsycNET, and Web of Knowledge) was conducted using search terms (Medical Subject Headings and keywords) relating to digital interventions, children, acute illness, and health information. Forty-eight papers were screened; seven were included in the review and critically appraised using the Mixed Method Appraisal Tool. In total, 3,558 parents were included. Meta-analysis was not possible due to heterogeneity of papers; thus, narrative synthesis was used to synthesize results and explore relationships between studies. The following aspects were documented: types and characteristics of interventions; how interventions were developed; accessibility, usability and acceptability; measures of impact upon parental knowledge, confidence; and satisfaction with the intervention and usefulness. Limited evidence exists on the availability, impact and efficacy of digital interventions supporting parents caring for acutely ill children at home. Barriers exist regarding accessibility, health literacy and there is limited representation of the diverse needs of parents from different countries, cultures and populations beyond mothers. Further research is needed to co-design and evaluate digital interventions designed with, and for, these parents.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0000998"},"PeriodicalIF":7.7,"publicationDate":"2025-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12435699/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145071186","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Consent-driven, semi-automated data collection during birth and newborn resuscitation: Insights from the NewbornTime study.","authors":"Sara Brunner, Anders Johannessen, Jorge García-Torres, Ferhat Özgur Catak, Øyvind Meinich-Bache, Siren Rettedal, Kjersti Engan","doi":"10.1371/journal.pdig.0000730","DOIUrl":"10.1371/journal.pdig.0000730","url":null,"abstract":"<p><p>Accurate observations at birth and during newborn resuscitation are fundamental for quality improvement initiatives and research. However, manual data collection methods often lack consistency and objectivity, are not scalable, and may raise privacy concerns. The NewbornTime project aims to develop an AI system that generates accurate timelines from birth and newborn resuscitation events by automated video recording and processing, providing a source of objective and consistent data. This work aims to describe the implementation of the data collection system that is necessary to support the project's purpose. Videos were recorded using thermal sensors in labor rooms and thermal and visible light cameras in resuscitation rooms. Consent from mothers was obtained before birth, and healthcare providers were given the option to delete videos by opting out on a case-by-case basis. The video collection process was designed to minimize interference with ongoing treatment and not impose unnecessary burden on healthcare providers. At Stavanger University Hospital, 1012 thermal videos of birth and 274 visible light videos of newborn stabilization and resuscitation have been collected during the data collection period from November 2021 to June 2025. The utilization of automated data collection and AI video processing around birth may allow for consistent and enhanced documentation, quality improvement initiatives, and research opportunities on sequence, timing and duration of treatment activities during acute events, with less effort needed for capturing data and improved privacy for participants.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0000730"},"PeriodicalIF":7.7,"publicationDate":"2025-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12416717/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145024872","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Evaluating anti-LGBTQIA+ medical bias in large language models.","authors":"Crystal T Chang, Neha Srivathsa, Charbel Bou-Khalil, Akshay Swaminathan, Mitchell R Lunn, Kavita Mishra, Sanmi Koyejo, Roxana Daneshjou","doi":"10.1371/journal.pdig.0001001","DOIUrl":"10.1371/journal.pdig.0001001","url":null,"abstract":"<p><p>Large Language Models (LLMs) are increasingly deployed in clinical settings for tasks ranging from patient communication to decision support. While these models demonstrate race-based and binary gender biases, anti-LGBTQIA+ bias remains understudied despite documented healthcare disparities affecting these populations. In this work, we evaluated the potential of LLMs to propagate anti-LGBTQIA+ medical bias and misinformation. We prompted 4 LLMs (Gemini 1.5 Flash, Claude 3 Haiku, GPT-4o, Stanford Medicine Secure GPT [GPT-4.0]) with 38 prompts consisting of explicit questions and synthetic clinical notes created by medically-trained reviewers and LGBTQIA+ health experts. The prompts consisted of pairs of prompts with and without LGBTQIA+ identity terms and explored clinical situations across two axes: (i) situations where historical bias has been observed versus not observed, and (ii) situations where LGBTQIA+ identity is relevant to clinical care versus not relevant. Medically-trained reviewers evaluated LLM responses for appropriateness (safety, privacy, hallucination/accuracy, and bias) and clinical utility. We found that all 4 LLMs generated inappropriate responses for prompts with and without LGBTQIA+ identity terms. The proportion of inappropriate responses ranged from 43-62% for prompts mentioning LGBTQIA+ identities versus 47-65% for those without. The most common reason for inappropriate classification tended to be hallucination/accuracy, followed by bias or safety. Qualitatively, we observed differential bias patterns, with LGBTQIA+ prompts eliciting more severe bias. Average clinical utility score for inappropriate responses was lower than for appropriate responses (2.6 versus 3.7 on a 5-point Likert scale). Future work should focus on tailoring output formats to stated use cases, decreasing sycophancy and reliance on extraneous information in the prompt, and improving accuracy and decreasing bias for LGBTQIA+ patients. We present our prompts and annotated responses as a benchmark for evaluation of future models. Content warning: This paper includes prompts and model-generated responses that may be offensive.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0001001"},"PeriodicalIF":7.7,"publicationDate":"2025-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12416741/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145024818","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Evaluating the predictive accuracy of cognitive screeners BAMCOG and MoCA in identifying postoperative delirium risk in aortic valve replacement patients: A cohort study.","authors":"Mariska E Te Pas, Sophie Adelaars, R Arthur Bouwman, Roy P C Kessels, Marcel G M Olde Rikkert, Daan van de Kerkhof, Erwin Oosterbos, Marc P Buise","doi":"10.1371/journal.pdig.0001005","DOIUrl":"10.1371/journal.pdig.0001005","url":null,"abstract":"<p><p>Postoperative delirium (POD) and postoperative encephalopathy (POE) are common complications in older adults undergoing aortic valve replacement (AVR), yet the predictive accuracy of cognitive screening tools remains uncertain. In this prospective cohort study, 50 patients aged 65 years and older scheduled for AVR between January and October 2022 underwent preoperative assessment with the Brain Aging Monitor Cognitive Assessment (BAMCOG) and Montreal Cognitive Assessment (MoCA). Postoperatively, POD was evaluated with the Delirium Observation Screening (DOS) scale and POE with electroencephalography (EEG). BAMCOG and MoCA showed poor accuracy in predicting POE, with AUROC values of 0.67 and 0.59 respectively, but BAMCOG demonstrated good accuracy for POD prediction (AUROC 0.85) compared with MoCA (AUROC 0.53). Higher BAMCOG scores were significantly associated with reduced POD incidence, with each 10% increase in score lowering the risk by 16%. These findings suggest that BAMCOG may be a valuable preoperative screening tool for POD, though larger studies are needed to confirm its clinical utility and establish optimal cutoff values.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0001005"},"PeriodicalIF":7.7,"publicationDate":"2025-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12416729/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145024858","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development and evaluation of a lightweight large language model chatbot for medication enquiry.","authors":"Kabilan Elangovan, Jasmine Chiat Ling Ong, Liyuan Jin, Benjamin Jun Jie Seng, Yu Heng Kwan, Lit Soo Ng, Ryan Jian Zhong, Justina Koi Li Ma, Yu He Ke, Nan Liu, Kathleen M Giacomini, Daniel Shu Wei Ting","doi":"10.1371/journal.pdig.0000961","DOIUrl":"10.1371/journal.pdig.0000961","url":null,"abstract":"<p><p>Large Language Models (LLMs) show promise in augmenting digital health applications. However, development and scaling of large models face computational constraints, data security concerns and limitations of internet accessibility in some regions. We developed and tested Med-Pal, a medical domain-specific LLM-chatbot fine-tuned with a fine-grained, expert curated medication-enquiry dataset consisting of 1,100 question and answer pairs. We trained and validated five light-weight, open-source LLMs of smaller parameter size (7 billion or less) on a validation dataset of 231 medication-related enquiries. We introduce SCORE, an LLM-specific evaluation criteria for clinical adjudication of LLM responses, performed by a multidisciplinary expert team. The best performing lighted-weight LLM was chosen as Med-Pal for further engineering with guard-railing against adversarial prompts. Med-Pal outperformed Biomistral and Meerkat, achieving 71.9% high-quality responses in a separate testing dataset. Med-Pal's light-weight architecture, clinical alignment and safety guardrails enable implementation under varied settings, including those with limited digital infrastructure.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0000961"},"PeriodicalIF":7.7,"publicationDate":"2025-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12410746/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145002182","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}