JAMIA Open最新文献

{"title":"Data-reflector: an agentic exploratory data analysis platform for researchers.","authors":"Moein Sabounchi, Nimay Hazare, Chris Capone, Mateen Jangda, Wonsuk Oh, Pushkala Jayaraman, Shashank Gupta, Ashwin Sawant, Zachary Kuschner, Prathamesh Parchure, Prem Timsina, Ben Kaplan, Aditi Vakil, Roopa Kohli-Seth, Patricia Kovatch, Alexis Zebrowski, Benjamin Glicksberg, Pulkit Agrawal, Bruce Darrow, Lisa Stump, Alexander Charney, Girish N Nadkarni, Ankit Sakhuja","doi":"10.1093/jamiaopen/ooag194","DOIUrl":"https://doi.org/10.1093/jamiaopen/ooag194","url":null,"abstract":"<p><strong>Objectives: </strong>Exploratory data analysis (EDA) is foundational to clinical research yet inaccessible to clinician-scientists without programming training. Existing large language model (LLM) tools improve accessibility but struggle with hallucinations. We developed and evaluated data reflector (DR), a hybrid agentic platform pairing natural language interaction with deterministic statistical computation.</p><p><strong>Materials and methods: </strong>Data reflector restricts the LLM to intent interpretation, tool selection, and filter parsing while routing all numerical computation through predefined deterministic functions executed locally. We piloted DR with 6 participants of diverse technical backgrounds across 6 clinical (MIMIC-IV, eICU Collaborative Research Database, National Health and Nutrition Examination Survey) and nonclinical (energy, COVID-19, urban mobility) datasets, assessing usability (system usability scale, SUS), task completion time, analytical output correctness against participant-generated manual reference results, hallucination prevention, and independent reproducibility by a second operator. A head-to-head comparison with ChatGPT 5.3 chatbot was also conducted.</p><p><strong>Results: </strong>Data reflector achieved a mean SUS of 94.58, with reduced task completion time across all evaluable participants. Statistical outputs and generated distribution curves demonstrated complete agreement with participant-generated manual reference results across all evaluable outputs. Independent reexecution by a different operator using independently constructed prompts also demonstrated complete agreement with the original DR outputs across all evaluated tasks. Hallucination-prevention safeguards correctly declined out-of-scope and impossible queries. ChatGPT produced incomplete, internally inconsistent outputs on identical tasks where DR returned complete, accurate results.</p><p><strong>Discussion: </strong>Data reflector's architectural separation of LLM reasoning from numerical computation removes dependence on prompt phrasing, session history, and stochastic LLM sampling while preserving conversational accessibility, addressing reproducibility limitations of general-purpose LLM tools.</p><p><strong>Conclusion: </strong>Data reflector offers a practical pattern for reproducible, accessible artificial intelligence-assisted EDA across heterogeneous data sources in modern clinical and translational research.</p>","PeriodicalId":36278,"journal":{"name":"JAMIA Open","volume":"9 5","pages":"ooag194"},"PeriodicalIF":4.7,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13630455/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149115788","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":"Use of chat GPT for sentiment and readability analysis of nutrition articles in legacy media.","authors":"Ann Gaba, Safa Amir, Haley Liebman, Nevin Cohen, Steven Cordova, Sergio Costa, Ashish Joshi","doi":"10.1093/jamiaopen/ooag188","DOIUrl":"https://doi.org/10.1093/jamiaopen/ooag188","url":null,"abstract":"<p><strong>Objectives: </strong>Artificial intelligence (AI) can be used to measure sentiment and reading level, facilitating its use to improve nutrition communication, though AI has not yet been tested for such analysis of nutrition articles. The present study investigated this by looking at reading level and sentiment in magazines aimed at various demographics as determined with ChatGPT.</p><p><strong>Materials and methods: </strong>Sentiment analysis was conducted on a sample of articles about nutrition from legacy media collected over 1 year. Reading level of the articles was determined using the Simple Measure of Gobbledygook (SMOG) formula. Nutrition content sub-themes were examined for reading level and sentiment.</p><p><strong>Results: </strong>Average sentiment ratings remained close to neutral (mean = 0.26 ± 2.05), with no consistent changes over time. Articles written at higher reading levels showed more positive sentiment values than those at lower reading levels. Of the twelve nutrition sub-themes were identified, the most prevalent were food/diet recommendations for health (23.8%), articles about celebrity diets (15.6%), cooking and culinary articles (14.1%), and food insecurity (10.7%). Comparison of sub-samples of sentiment ratings or Simple Measure of Gobbledygook (SMOG) scores done by a human coder versus ChatGPT were not different (<i>P</i> = NS).</p><p><strong>Discussion: </strong>Reading levels are relatively high overall. These varied by source periodical and theme as would be expected with editors targeting varied readership groups. Analysis by nutrition themes showed differences in sentiment that were concurrent with the type of content analyzed.</p><p><strong>Conclusion: </strong>This study demonstrated that AI such as ChatGPT can be utilized as a for assessment of sentiment and reading levels in nutrition articles.</p>","PeriodicalId":36278,"journal":{"name":"JAMIA Open","volume":"9 5","pages":"ooag188"},"PeriodicalIF":4.7,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13629438/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149115885","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":"Harmonizing UK primary care prescription records for research: a case study in the UK biobank.","authors":"Cai R Ytsma, Ana Torralbo, Natalie K Fitzpatrick, Maik Pietzner, Daniela Nguyen, Ioannis Louloudis, Sabrina Ansarey, Spiros Denaxas","doi":"10.1093/jamiaopen/ooag190","DOIUrl":"https://doi.org/10.1093/jamiaopen/ooag190","url":null,"abstract":"<p><strong>Objective: </strong>We aimed to develop and validate an automated, scalable framework to harmonize fragmented UK primary care prescription records into a research-ready dataset by mapping diverse medical ontologies to a unified reference standard from the UK Biobank.</p><p><strong>Materials and methods: </strong>Preprocessing involved selecting a single drug code where multiple were recorded, cleaning codes to match reference presentations, expanding code granularity via drug descriptions, and updating outdated codes. Harmonization mapped British National Formulary (BNF) and Read2 codes to dm+d, the National Health Service standard vocabulary, with records then homogenized to the Virtual Medicinal Product (VMP) granularity. We validated our approach by profiling prescribing patterns across 312 diseases.</p><p><strong>Results: </strong>We preprocessed 57 659 844 records from 221 868 participants; we dropped 48 950 records for missing drug codes and 13% used multiple ontologies. Most records were BNF-encoded (76%), and 49% required granularity expansion via drug description. In total, 72% of records were harmonized to dm+d, of which 99.98% were converted to VMP. Across 312 diseases, we identified 23 352 disease-drug associations involving 237 medications (BNF subparagraphs) that survived statistical correction, most resembling drug-indication pairs.</p><p><strong>Discussion: </strong>Our framework addresses a longstanding barrier to large-scale pharmacoepidemiological research by reconciling heterogeneous coding systems into a single, consistent standard. The high conversion rate to VMP demonstrates near-complete coverage, while the recovery of plausible drug-indication associations supports validity. Limitations include residual uncodable records and dependence on dm+d completeness.</p><p><strong>Conclusion: </strong>Our approach transforms fragmented prescription records into a streamlined, enriched dataset suitable for large-scale discovery research.</p>","PeriodicalId":36278,"journal":{"name":"JAMIA Open","volume":"9 5","pages":"ooag190"},"PeriodicalIF":4.7,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13630483/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149115897","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":"Advancing multi-database cohort studies: lessons learned from a studyathon in a pharmaceutical company.","authors":"Elif Inan-Eroglu, Daloha Rodriguez-Molina, Khaled Abdelgawwad, Carina Dinkel-Keuthage, Montse Soriano Gabarró, Pareen Vora, Gianmario Candore, David Vizcaya, Katrin Manlik","doi":"10.1093/jamiaopen/ooag189","DOIUrl":"https://doi.org/10.1093/jamiaopen/ooag189","url":null,"abstract":"<p><strong>Objectives: </strong>To evaluate the feasibility and applicability of the studyathon format within a pharmaceutical company setting for generating reliable real-world evidence rapidly for a business-relevant research question.</p><p><strong>Materials and methods: </strong>Conducted at Bayer, the studyathon assessed recurrent events following a first-ever transient ischemic attack. It involved an 8-week preparatory phase, a 1-week execution phase, and a 3-week post-event phase, using multiple real-world data sources.</p><p><strong>Results: </strong>The studyathon concluded successfully with a conference abstract submission. Key learnings included the importance of defining precise research questions, organizational alignment, thorough preparation, and leveraging diverse expertise and skill sets.</p><p><strong>Discussion: </strong>The studyathon format proved suited for a pharmaceutical setting and revealed insights into its application. Our learnings highlight differences from multi-stakeholder community studyathons.</p><p><strong>Conclusion: </strong>The studyathon demonstrated its potential as a valuable research approach within a pharmaceutical company. It improved study timelines and cross-functional collaboration while preserving scientific rigor.</p>","PeriodicalId":36278,"journal":{"name":"JAMIA Open","volume":"9 5","pages":"ooag189"},"PeriodicalIF":4.7,"publicationDate":"2026-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13625018/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149115794","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 validity and performance of a modifiable prediction model to predict new long-term opioid use in prescription drug monitoring programs.","authors":"Iraklis Erik Tseregounis, Eugene Shin, Stephen G Henry, Chris Delcher","doi":"10.1093/jamiaopen/ooag192","DOIUrl":"https://doi.org/10.1093/jamiaopen/ooag192","url":null,"abstract":"<p><strong>Objective: </strong>Prescription drug monitoring programs (PDMPs) offer a promising data source for the development of clinical decision support (CDS) systems to guide opioid prescribing, but most PDMP agencies lack capacity for in-house model development-instead relying on third-party tools. This study aims to externally validate a PDMP-based model, developed using 2018 California PDMP records, to predict new long-term use among Kentucky opioid-naïve patients and compare its performance to Kentucky-specific machine learning (ML) models.</p><p><strong>Materials and methods: </strong>We identified opioid-naïve Kentucky residents who filled an index pill opioid prescription between 2021 and 2023 (development cohort: 2021-2022; validation cohort: 2023). The California model was updated via logistic calibration and evaluated against 5 newly developed Kentucky ML models (least absolute shrinkage and selection operator [LASSO], elastic net, deep neural network, random forest, XGBoost) using discrimination and calibration metrics.</p><p><strong>Results: </strong>Among 985 317 opioid-naïve patients, 33 298 (3.4%) transitioned to new long-term opioid use. The updated California model achieved excellent discrimination (area under the receiver operating characteristic curve [AUROC]: 0.919 [0.915-0.923]) and was calibrated up to the highest risk levels (calibration intercept: 0.010, [-0.021, 0.041]; slope: 0.997, [0.979, 1.014]), with negative predictive values exceeding 0.99. Performance was comparable to Kentucky-specific ML models, including the highest performer XGBoost (AUROC: 0.930 [0.926-0.933]). All models struggled with low positive predictive values (0.15-0.19).</p><p><strong>Conclusion: </strong>A California PDMP-derived prediction model transported successfully to Kentucky with minimal updating. Its strong comparative performance provides a feasible foundation for PDMP agencies to develop transparent, in-house CDS systems with potential clinical implications for safer and more individualized opioid prescribing.</p>","PeriodicalId":36278,"journal":{"name":"JAMIA Open","volume":"9 5","pages":"ooag192"},"PeriodicalIF":4.7,"publicationDate":"2026-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13625009/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149115899","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":"Rethinking input complexity in transformer-based clinical prediction: implications for feature dimensionality and sequence length in longitudinal electronic health record data.","authors":"Wansu Chen, Botao Zhou, Robert S Zeiger, William Crawford, Michael Schatz, Eric Puttock, Stanley Xu, Matthew T Slaughter, Fagen Xie","doi":"10.1093/jamiaopen/ooag193","DOIUrl":"https://doi.org/10.1093/jamiaopen/ooag193","url":null,"abstract":"<p><strong>Objectives: </strong>Transformer-based models for clinical prediction using longitudinal electronic health record (EHR) data are often developed with large feature sets and long patient histories under the assumption that more data improves performance. However, high-dimensional inputs and long sequences increase computational burden, potentially limiting scalability. We evaluated how feature dimensionality and sequence length affect predictive performance, calibration, risk stratification, and computational efficiency in EHR prediction.</p><p><strong>Materials and methods: </strong>Using longitudinal EHR data from adults with mild asthma in an integrated healthcare system, we evaluated input representation design for predicting acute asthma exacerbation. Feature dimensionality was reduced using Integrated Gradients attribution scores, univariate performance-based selection, and clinically guided selection strategies. Sequence length was varied using percentile-based truncation of patient histories. Performance was assessed using discrimination, calibration, high-risk classification, threshold-based event capture, and computational efficiency across regions.</p><p><strong>Results: </strong>Models using fewer features achieved discrimination comparable to the 80-feature reference model, with area under the receiver operating characteristic curve values ranging from 0.843 to 0.864 vs 0.870 for the full model. Moderate sequence-length truncation reduced training time by more than 70% with minimal loss in discrimination. Although reduced-dimensional models showed attenuation of predicted risk at the upper tail, they identified similar high-risk populations and captured comparable proportions of asthma exacerbation events at clinically relevant thresholds.</p><p><strong>Conclusion: </strong>Transformer-based prediction models maintained strong performance across reduced feature sets. While dimensionality reduction modestly affected calibration at the highest risk levels, moderate sequence-length reduction substantially reduced computational burden with limited change in overall discrimination. These findings highlight tradeoffs between input complexity, predictive performance, and computational efficiency.</p>","PeriodicalId":36278,"journal":{"name":"JAMIA Open","volume":"9 5","pages":"ooag193"},"PeriodicalIF":4.7,"publicationDate":"2026-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13625019/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149115910","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":"An American Hospital Association-informed weighting framework for Oracle Health Real-World Data: validation against National Healthcare Cost and Utilization Project encounter databases.","authors":"Fares Qeadan, Benjamin Tingey, Mirjana Glisovic Bensa, Erin F Madden, Pooja Lagisetty, Philip J Kroth","doi":"10.1093/jamiaopen/ooag184","DOIUrl":"10.1093/jamiaopen/ooag184","url":null,"abstract":"<p><strong>Objective: </strong>To develop and validate a weighting framework to improve representativeness of Oracle Health Real-World Data (OHRWD).</p><p><strong>Materials and methods: </strong>We conducted cross-sectional analyses of OHRWD encounters in 2019 and 2022. The primary method (M1) applied design weights based on American Hospital Association (AHA) hospital encounter counts to balance OHRWD encounters across strata of US region, hospital system, bed size, and encounter type (inpatient, emergency department, ambulatory surgery). Comparative methods (M2-M5) used unified structural weights, encounter-type multipliers, iterative proportional fitting, and demographic post-stratification. Weighted OHRWD estimates were validated against the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS), Nationwide Emergency Department Sample (NEDS), and Nationwide Ambulatory Surgery Sample (NASS) for demographics, conditions (low back pain, opioid use disorder (OUD), hypertension, diabetes), and procedures (colonoscopy, appendectomy). Equivalence was assessed using two one-sided tests with ±20% margins; standardized differences were summarized using Cohen's d and h.</p><p><strong>Results: </strong>M1 produced close alignment between OHRWD and HCUP for age, sex, region, and most race/ethnicity groups, with negligible effect sizes. Low back pain, OUD, and appendectomy prevalences closely matched HCUP benchmarks, whereas hypertension and, in some settings, diabetes and Hispanic ethnicity showed larger deviations. Alternative methods (M2-M5) yielded mixed performance and did not consistently outperform M1.</p><p><strong>Discussion: </strong>Encounter-type-specific, structurally informed weighting improved alignment with HCUP national benchmarks, highlighting domains, such as hypertension and race/ethnicity estimates, that require additional calibration.</p><p><strong>Conclusion: </strong>AHA-based, encounter-type-stratified weighting enables OHRWD to better approximate encounter patterns and could support epidemiologic and health services research using large EHR datasets.</p>","PeriodicalId":36278,"journal":{"name":"JAMIA Open","volume":"9 5","pages":"ooag184"},"PeriodicalIF":4.7,"publicationDate":"2026-09-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13615648/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149064986","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":"Correction to: Use and characteristics of free/libre and open source electronic health records: an overview of reviews.","authors":"","doi":"10.1093/jamiaopen/ooag174","DOIUrl":"https://doi.org/10.1093/jamiaopen/ooag174","url":null,"abstract":"<p><p>[This corrects the article DOI: 10.1093/jamiaopen/ooag100.].</p>","PeriodicalId":36278,"journal":{"name":"JAMIA Open","volume":"9 4","pages":"ooag174"},"PeriodicalIF":4.7,"publicationDate":"2026-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13528623/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148866650","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":"Attitudes, perceptions, and UTAUT-based factors influencing the acceptance of medical artificial intelligence among Chinese oncology healthcare professionals: a national cross-sectional survey.","authors":"Zhaoyu Wang, Qianqian Dai, Maoshu Yang, Shiwu Shi, Jiaojiao Liao, Zhaoji Li, Xiaoqiu Dai, Liyuan Tao","doi":"10.1093/jamiaopen/ooag154","DOIUrl":"10.1093/jamiaopen/ooag154","url":null,"abstract":"<p><strong>Objectives: </strong>To conduct a nationwide survey among professionals working in oncology departments in China to investigate their attitudes, perceptions, and experiences regarding medical artificial intelligence (AI), and to explore and compare the factors influencing AI behavioral intention (BI; willingness to adopt AI) between physicians and nurses using the Unified Theory of Acceptance and Use of Technology (UTAUT).</p><p><strong>Materials and methods: </strong>A nationwide cross-sectional survey was conducted among professionals in oncology departments across China. Sociodemographic characteristics, awareness, perceptions, and user experiences related to medical AI were collected. The UTAUT was specifically applied to the third section of the survey to evaluate participants' perceptions. Based on this framework, structural equation modeling (SEM) was performed to examine associations between BI and 6 latent constructs: performance expectancy (PE), effort expectancy (EE), social influence (SI), facilitating conditions (FC), perceived risk (PR), and trust (TR) based on the UTAUT. Subgroup SEM analyses were then conducted for physicians and nurses.</p><p><strong>Results: </strong>Six hundred ten valid responses were included in the final analysis, consisting of 188 (30.8%) physicians and 422 (69.2%) nurses. Only 17.7% reported having both heard of and used medical AI, while 67.7% had heard of it but never used it. SEM results showed that TR (confidence in system reliability; β = .668, <i>P</i> < .001) had the strongest effect on BI, followed by effort EE (perceived ease of use; β = .155, <i>P</i> = .026). Subgroup analyses revealed that for physicians, both EE (β = .257, <i>P</i> = .038) and TR (β = .486, <i>P</i> < .001) significantly influenced BI, while among nurses, only TR (β = .729, <i>P</i> < .001) showed a significant effect.</p><p><strong>Conclusions: </strong>There is a significant \"awareness-usage gap\" among Chinese oncology professionals, with high awareness but low practical integration of medical AI. This nationwide study indicates that TR and EE are associated with the intention to use AI. To bridge this gap, hospital management should implement differentiated training and local validation protocols that address the unique technical and ethical concerns of physicians and nurses, respectively.</p>","PeriodicalId":36278,"journal":{"name":"JAMIA Open","volume":"9 4","pages":"ooag154"},"PeriodicalIF":4.7,"publicationDate":"2026-08-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13505628/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148819677","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":"Adaptation of a fair individualized polysocial risk score for hospitalization risk prediction.","authors":"Qinglin Gou, Yu Hu, Xing He, Megan E Gregory, Jennifer H LeLaurin, Ramzi G Salloum, Jiang Bian, Jingchuan Guo, Yu Huang","doi":"10.1093/jamiaopen/ooag161","DOIUrl":"10.1093/jamiaopen/ooag161","url":null,"abstract":"<p><strong>Objectives: </strong>We adapted the individualized polysocial risk score (iPsRS), a machine learning model originally developed for patients with type 2 diabetes, to evaluate its generalizability in predicting 1-year hospitalization risk in a disease-agnostic adult cohort, with attention to fairness and explainability.</p><p><strong>Materials and methods: </strong>The study utilized de-identified electronic health record data from a retrospective cohort of 17 857 adult patients at the University of Florida Health. The original iPsRS framework was reimplemented using XGBoost and Logistic Regression models. These models were trained and fine-tuned using 13 individual-level social determinants of health (SDoH) predictors to predict all-cause hospitalization within 1 year. Models were fine-tuned at 5 levels (0%, 10%, 20%, 50%, and 70%) with 3 sampling strategies (none, oversampling, and undersampling). Model performance was primarily evaluated using the AUROC. Interpretability was assessed through SHapley Additive exPlanations and a causal structure learning algorithm, while fairness was analyzed by examining false negative rate disparities across age, sex, and racial/ethnic subgroups.</p><p><strong>Results: </strong>After fine-tuning, the adapted iPsRS demonstrated moderate predictive performance, achieving an AUROC of up to 0.671 with XGBoost and Normal sampling. SHapley Additive exPlanations analysis on the evaluation split showed age, food insecurity, marital status, and financial constraints emerged as consistently influential predictors. The causal structure analysis corroborated predictive findings, identifying age, race, and employment as proximal factors associated with hospitalization.</p><p><strong>Conclusion: </strong>The iPsRS was successfully adapted beyond diabetes, maintaining moderate predictive performance and meaningful risk stratification. Incorporating individual-level SDoH enables equity-aware prediction, supporting broader use of iPsRS in clinical care.</p>","PeriodicalId":36278,"journal":{"name":"JAMIA Open","volume":"9 4","pages":"ooag161"},"PeriodicalIF":4.7,"publicationDate":"2026-08-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13503007/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148814411","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}