PLOS digital health最新文献

{"title":"Identification of physiological adverse events using continuous vital signs monitoring during paediatric critical care transport: A novel data-driven approach.","authors":"Milan Kapur, Kezhi Li, Alexander Brown, Zhiqiang Huo, Philip Knight, Gwyneth Davies, Padmanabhan Ramnarayan","doi":"10.1371/journal.pdig.0000822","DOIUrl":"10.1371/journal.pdig.0000822","url":null,"abstract":"<p><p>Interhospital transport of critically unwell children exacerbates physiological stress, increasing the risk of deterioration during transport. Due to the nature of illness and interventions occurring in this cohort, defining \"normal\" vital sign ranges is impossible, which can make identifying deterioration events difficult. A novel data-driven approach was developed to identify adverse respiratory and cardiovascular events in critically ill children during interhospital transport. In this retrospective cohort study of 1,519 transports (July 2016 to May 2021), vital signs were recorded at one-second intervals and then analysed using an adaptation of Bollinger Bands, a technique borrowed from financial market analysis. This method dynamically established each patient's stable ranges for heart rate, blood pressure, oxygen saturation, and other respiratory parameters, and flagged adverse events when multiple parameters simultaneously fell outside their expected ranges. Adverse respiratory events were identified when oxygen saturation deviated below a dynamically defined threshold alongside at least one additional respiratory parameter. Cardiovascular events were defined by concurrent deviations in blood pressure and heart rate. Overall, 15.6 percent of transports had one or more adverse respiratory events, and 21.5 percent had at least one adverse cardiovascular event. To validate these labels, the number of adverse events and the cumulative duration of vital sign instability during transport were compared against clinical markers of deterioration. Each additional respiratory event was associated with increased odds of receiving respiratory support during transport and higher 30-day mortality, while each additional cardiovascular event was associated with increased odds of receiving vasoactive support during transport. Our method detects respiratory and cardiovascular adverse events during transport. The approach is readily adaptable to other high-resolution intensive care datasets, for both retrospective labelling as well as automated, real-time identification of adverse events in the clinical setting, offering a foundation for improved monitoring and early intervention in critically ill patients.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0000822"},"PeriodicalIF":7.7,"publicationDate":"2025-09-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12463272/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145152280","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":"mHealth technologies in research studying cardiovascular health in cancer: A systematic review.","authors":"Roberto M Benzo, Anvitha Gogineni, Macy K Tetrick, Rujul Singh, Peter Washington, Soledad Fernandez, Electra D Paskett, Frank J Penedo, Sanam Ghazi, Alex Osei, Steven K Clinton, Jessica Krok-Schoen, Sarah Weyrauch, Daniel Addison","doi":"10.1371/journal.pdig.0001027","DOIUrl":"10.1371/journal.pdig.0001027","url":null,"abstract":"<p><p>Cancer survivors face an increased risk of cardiovascular disease (CVD) due to treatment-related toxicity, lifestyle factors, and comorbidities. Addressing CV health is crucial for improving quality of life and long-term outcomes. The American Heart Association's Life's Essential 8 framework highlights modifiable determinants of CV health, emphasizing early detection and monitoring. Mobile health (mHealth) technologies, such as wearables and smartphone apps, offer continuous tracking, yet their applications in cancer survivorship remain unclear. This review systematically characterizes the types of mHealth technologies used to monitor CV health in cancer survivors, focusing on the specific data collected (major adverse CV events, CV risk factors, and surrogate endpoints) and the use of active versus passive collection methods. A systematic search of PubMed, Scopus, Embase, and Web of Science identified studies published between January 1, 2016, and June 13, 2024. Eligible studies included observational and interventional designs assessing at least one CV outcome using mHealth. Data were extracted on design, technology type, and outcomes. Risk of bias was evaluated using the Cochrane RoB-2 and ROBINS-I tools. Fourteen studies (13 interventional, one observational) met criteria. Physical activity was the most monitored risk factor, followed by HR. The most common technologies were mobile apps and commercial wearables. Passive methods typically captured PA and HR, while active methods captured PA, symptom tracking, and diet. A key finding was the lack of integration with electronic medical records, highlighting a gap in clinical implementation. mHealth provides scalable tools to track CV health indicators in cancer survivors. Findings highlight the potential to support practice by enabling remote oversight of risk-reducing behaviors and guiding lifestyle interventions. However, we also identified gaps, including the underutilization of biomarkers (e.g., HRV) and the lack of integration with electronic records. Future research must address these gaps to translate real-time data into clinical insights and optimize survivorship care.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0001027"},"PeriodicalIF":7.7,"publicationDate":"2025-09-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12463205/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145152306","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":"Artificial intelligence applications in refractive error management: A systematic review and meta-analysis.","authors":"Josephine Ampong, Sylvia Agyekum, Werner Eisenbarth, Albert Kwadjo Amoah Andoh, Isaiah Osei Duah Junior, Gabriel Amankwah, Gabriel Kwaku Agbeshie, Eldrick Adu Acquah, Clement Afari, Emmanuel Assan, Saphiel Osei Poku, Karen Ama Sam, Josephine Ampomah Boateng, Kwadwo Owusu Akuffo","doi":"10.1371/journal.pdig.0000904","DOIUrl":"10.1371/journal.pdig.0000904","url":null,"abstract":"<p><p>Artificial intelligence (AI) has transformed healthcare, and is becoming increasingly useful in eye care. We conducted a systematic review and meta-analysis of the use of AI in the diagnosis, detection, prediction, progression, and treatment of refractive errors (REs). The study adhered to the PRISMA checklist to ensure transparent reporting. The following databases were searched from inception to January 2025, with an English language restriction: PubMed, Web of Science, Embase, Scopus, Cochrane Library and Google Scholar. Two independent reviewers performed study screening, data extraction, and quality assessment, with a third author resolving discrepancies. All original studies on the use of AI techniques in RE were identified and the effectiveness of these techniques was compared. A critical appraisal was conducted using the QUADAS-2 risk-of-bias tool. A meta-analysis was performed using R software (version 4.5.0). Of 6,288 records retrieved, 45 met eligibility for systematic review, with 19 included in meta-analysis. Among these 45 studies, 55.5% (25/45) applied deep learning (DL) approaches, while 44.4% (20/45) employed machine learning (ML) techniques. The pooled sensitivity, specificity, diagnostic odds ratio (DOR), and summary of receiver operating characteristic (SROC) for detection and/or diagnosis studies were 0.94 (95%CI, 0.90-0.97), 0.96 (95%CI, 0.92-0.98), 382.56 (95% CI 111.91 -1307.77) and 0.98 (95%CI, 0.91-0.97), respectively. For prediction of REs, the pooled sensitivity, specificity, DOR, and SROC were 0.87 (95%CI, 0.73-0.94), 0.96 (95%CI, 0.90-0.980), 159.94 (95% CI, 40.17-636.85) and 0.96 (95%CI, 0.85-0.95), respectively. Among studies focused on progression, performance metrics ranged from AUC = 0.845-0.99, R² = 0.613-0.964, and MAE = 0.119D-0.49D. In treatment studies, performance varied more widely, with AUC values between 0.60-0.94 and MAE from 0.17D-0.54D. Collectively, AI technologies, particularly DL and ML, achieved high diagnostic and predictive accuracy in RE management. Future research should focus on developing generalizable models trained on diverse datasets to ensure broad clinical relevance.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0000904"},"PeriodicalIF":7.7,"publicationDate":"2025-09-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12463214/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145152259","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":"Individualized machine-learning-based clinical assessment recommendation system.","authors":"Devin Setiawan, Yumiko Wiranto, Jeffrey M Girard, Amber Watts, Arian Ashourvan","doi":"10.1371/journal.pdig.0001022","DOIUrl":"10.1371/journal.pdig.0001022","url":null,"abstract":"<p><p>Traditional clinical assessments often lack individualization, relying on standardized procedures that may not accommodate the diverse needs of patients, especially in early stages where personalized diagnosis could offer significant benefits. We aim to provide a machine-learning framework that addresses the individualized feature addition problem and enhances diagnostic accuracy for clinical assessments.Individualized Clinical Assessment Recommendation System (iCARE) employs locally weighted logistic regression and Shapley Additive Explanations (SHAP) value analysis to tailor feature selection to individual patient characteristics. Evaluations were conducted on synthetic and real-world datasets, including early-stage diabetes risk prediction and heart failure clinical records from the UCI Machine Learning Repository. We compared the performance of iCARE with a Global approach using statistical analysis on accuracy and area under the ROC curve (AUC) to select the best additional features. The iCARE framework enhances predictive accuracy and AUC metrics when additional features exhibit distinct predictive capabilities, as evidenced by synthetic datasets 1-3 and the early diabetes dataset. Specifically, in synthetic dataset 1, iCARE achieved an accuracy of 0.999 and an AUC of 1.000, outperforming the Global approach with an accuracy of 0.689 and an AUC of 0.639. In the early diabetes and heart disease dataset, iCARE shows improvements of 6-12% in accuracy and AUC across different numbers of initial features over other feature selection methods. Conversely, in synthetic datasets 4-5 and the heart failure dataset, where features lack discernible predictive distinctions, iCARE shows no significant advantage over global approaches on accuracy and AUC metrics. iCARE provides personalized feature recommendations that enhance diagnostic accuracy in scenarios where individualized approaches are critical, improving the precision and effectiveness of medical diagnoses.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0001022"},"PeriodicalIF":7.7,"publicationDate":"2025-09-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12463258/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145152289","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 health interventions for colorectal cancer screening uptake: A scoping review.","authors":"Sujin Kim, Andrew J Whipkey, Jihye Bae, Avinash Bhakta","doi":"10.1371/journal.pdig.0001028","DOIUrl":"10.1371/journal.pdig.0001028","url":null,"abstract":"<p><p>Digital health interventions (DHIs) are increasingly employed to improve colorectal cancer (CRC) screening uptake, yet comprehensive syntheses of their effectiveness across diverse contexts remain scarce. This scoping review examines how individual, contextual, technological, and timing-related factors shape CRC screening outcomes in DHI-based trials. Following PRISMA-ScR guidelines, we conducted a systematic search of PubMed, Google Scholar, and ClinicalTrials.gov from March 1 to April 20, 2024, identifying 4,523 records through databases and an additional 2,039 through backward citation tracking. After deduplication and screening, 51 studies were included and charted using the PICOT (Population, Intervention, Comparison, Outcome, and Timing) framework. Included studies spanned urban health systems, rural community clinics, and Federally Qualified Health Centers in the United States, Europe, Asia, and Australia, with intervention durations ranging from six weeks to ten years. Keyword co-occurrence mapping revealed four thematic domains: (1) patient-centered technology and adherence, (2) behavioral design and personalization, (3) clinical workflow and provider interaction, and (4) equity, disparities, and community engagement. Findings showed that tailored telephone outreach, mailed fecal immunochemical testing combined with navigation support, EMR-based automated reminders, and mobile applications delivering personalized education increased screening rates by 20.9% to 37.7% compared with conventional approaches. Hybrid models combining digital tools with human facilitation were particularly effective for underserved populations, including racial and ethnic minorities, rural communities, and individuals with limited health literacy. However, research gaps persist for younger adults at risk for early-onset CRC and for understanding the long-term sustainability and cost-effectiveness of digital interventions. Temporal aspects such as intervention timing, frequency, and duration were identified as important factors but were inconsistently reported. Future research should address digital health literacy, implementation barriers, and long-term follow-up to support sustained CRC screening adherence through user-centered, scalable, and culturally responsive digital solutions.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0001028"},"PeriodicalIF":7.7,"publicationDate":"2025-09-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12463253/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145152283","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":"Detection of pneumonia in children through chest radiographs using artificial intelligence in a low-resource setting: A pilot study.","authors":"Taofeeq Oluwatosin Togunwa, Abdulhammed Opeyemi Babatunde, Oluwatosin Ebunoluwa Fatade, Richard Olatunji, Godwin Ogbole, Adegoke Falade","doi":"10.1371/journal.pdig.0000713","DOIUrl":"10.1371/journal.pdig.0000713","url":null,"abstract":"<p><p>Pneumonia is a leading cause of death among children under 5 years in low-and-middle-income-countries (LMICs), causing an estimated 700,000 deaths annually. This burden is compounded by limited diagnostic imaging expertise. Artificial intelligence (AI) has potential to improve pneumonia diagnosis from chest radiographs (CXRs) through enhanced accuracy and faster diagnostic time. However, most AI models lack validation on prospective clinical data from LMICs, limiting their real-world applicability. This study aims to develop and validate an AI model for childhood pneumonia detection using Nigerian CXR data. In a multi-center cross-sectional study in Ibadan, Nigeria, CXRs were prospectively collected from University College Hospital (a tertiary hospital) and Rainbow-Scans (a private diagnostic center) radiology departments via cluster sampling (November 2023-August 2024). An AI model was developed on open-source paediatric CXR dataset from the USA, to classify the local prospective CXRs as either normal or pneumonia. Two blinded radiologists provided consensus classification as the reference standard. The model's accuracy, precision, recall, F1-score, and area-under-the-curve (AUC) were evaluated. The AI model was developed on 5,232 open-source paediatric CXRs, divided into training (1,349 normal, 3,883 pneumonia) and internal test (234 normal, 390 pneumonia) sets, and externally tested on 190 radiologist-labeled Nigerian CXRs (93 normal, 97 pneumonia). The model achieved 86% accuracy, 0.83 precision, 0.98 recall, 0.79 F1-score, and 0.93 AUC on the internal test, and 58% accuracy, 0.62 precision, 0.48 recall, 0.68 F1-score, and 0.65 AUC on the external test. This study illustrates AI's potential for childhood pneumonia diagnosis but reveals challenges when applied across diverse healthcare environments, as revealed by discrepancies between internal and external evaluations. This performance gap likely stems from differences in imaging protocols/equipment between LMICs and high-income settings. Hence, public health priority should be developing robust, locally relevant datasets in Africa to facilitate sustainable and independent AI development within African healthcare.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0000713"},"PeriodicalIF":7.7,"publicationDate":"2025-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12459801/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145139766","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":"Ranking and selecting hospital information systems: A multi-criteria decision making approach using TOPSIS in Hamadan, Iran.","authors":"Hamid Bouraghi, Ali Mohammadpour, Ahmad Maamaki, Parviz Karamiyan","doi":"10.1371/journal.pdig.0001016","DOIUrl":"10.1371/journal.pdig.0001016","url":null,"abstract":"<p><p>This study evaluated and ranked Hospital Information Systems (HIS) in teaching hospitals affiliated with Hamadan University of Medical Sciences, Iran. A cross-sectional design was employed, evaluating three HIS systems (PARDAZESHGARAN, SAYAN, and RAYAVARAN). Eight key criteria-technical quality, software quality, support quality, workflow support quality, output quality, cost, user satisfaction, and interdepartmental communication quality-were considered. Criterion weights were determined through expert consultation. The Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) method was used to rank the HIS systems. Results indicated that cost and output quality were assigned the highest weights. The PARDAZESHGARAN system was ranked first, followed by SAYAN and RAYAVARAN. While this study provides insights into HIS evaluation in this context, limitations related to the sample size should be considered.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0001016"},"PeriodicalIF":7.7,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12456787/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145133132","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":"Exploring the clinical utility of rhythmic digital markers for schizophrenia.","authors":"Axel Constant, Vincent Paquin, Robert A Ackerman, Colin A Depp, Raeanne C Moore, Philip D Harvey, Amy E Pinkham","doi":"10.1371/journal.pdig.0001010","DOIUrl":"10.1371/journal.pdig.0001010","url":null,"abstract":"<p><p>This study investigates the clinical utility of rhythmic digital markers (RDMs) in schizophrenia. RDMs are digital markers capturing behavioral rhythms over different timescales - within 24 hours span (ultradian), at a span of 24 hours (circadian), or over cycles of more than 24 hours (infradian). While previous research has explored digital markers for schizophrenia, the focus has primarily been on sensor data variability rather than rhythmic patterns. This study introduces two RDMs: an entropy RDM, which quantifies uncertainty in activity distribution over the infradian cycles, and a dynamic RDM, which is derived from models of transitions in entropy and psychotic symptom intensity using Markov chain analysis. Data were ecological momentary assessments (EMAs) of 39 activities collected from 390 individuals diagnosed with schizophrenia (N = 153) or bipolar disorder (N = 192) and controls (N = 45). We assessed associations between RDMs and symptom severity and whether participants could be differentiated based on these RDMs. We found that participants with schizophrenia significantly differed on dynamic RDMs, suggesting a potential diagnostic utility. However, dynamic RDMs were not associated with symptom severity, and entropy RDM had no significant clinical correlate. Our findings contribute to the growing evidence on digital markers in psychiatry and highlight the potential of rhythmic digital markers (RDMs) in characterizing digital phenotypes for schizophrenia.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0001010"},"PeriodicalIF":7.7,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12456810/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145133127","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":"Artificial intelligence-enhanced handheld breast ultrasound for screening: A systematic review of diagnostic test accuracy.","authors":"Arianna Bunnell, Dustin Valdez, Fredrik Strand, Yannik Glaser, Peter Sadowski, John A Shepherd","doi":"10.1371/journal.pdig.0001019","DOIUrl":"10.1371/journal.pdig.0001019","url":null,"abstract":"<p><p>Breast cancer screening programs using mammography have led to significant mortality reduction in high-income countries. However, many low- and middle-income countries lack resources for mammographic screening. Handheld breast ultrasound (BUS) is a low-cost alternative but requires substantial training. Artificial intelligence (AI) enabled BUS may aid in both the detection and classification of breast cancer, enabling screening use in low-resource contexts. The purpose of this systematic review is to investigate whether AI-enhanced BUS is sufficiently accurate to serve as the primary modality in screening, particularly in resource-limited environments. This review (CRD42023493053) is reported in accordance with the PRISMA guidelines. Evidence synthesis is reported in accordance with the SWiM (Synthesis Without Meta-analysis) guidelines. PubMed and Google Scholar were searched from January 1, 2016 to December 12, 2023. Studies are grouped according to AI task and assessed for quality. Of 763 candidate studies, 314 full texts were reviewed and 34 studies are included. The AI tasks of included studies are as follows: 1 frame selection, 6 lesion detection, 11 segmentation, and 16 classification. 79% of studies were at high or unclear risk of bias. Exemplary classification and segmentation AI systems perform with 0.976 AUROC and 0.838 Dice similarity coefficient. There has been encouraging development of AI for BUS. However, despite studies demonstrating high performance, substantial further research is required to validate reported performance in real-world screening programs. High-quality model validation on geographically external, screening datasets will be key to realizing the potential for AI-enhanced BUS in increasing screening access in resource-limited environments.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0001019"},"PeriodicalIF":7.7,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12453205/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145126734","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":"Integrating multiple data sources to predict all-cause readmission or mortality in patients with substance misuse.","authors":"Tim Gruenloh, Preeti Gupta, Askar Safipour Afshar, Madeline Oguss, Elizabeth Salisbury-Afshar, Marie Pisani, Ryan P Westergaard, Michael Spigner, Megan Gussick, Matthew Churpek, Majid Afshar, Anoop Mayampurath","doi":"10.1371/journal.pdig.0001008","DOIUrl":"10.1371/journal.pdig.0001008","url":null,"abstract":"<p><p>Patients with substance misuse who are admitted to the hospital are at heightened risk for adverse outcomes, such as readmission and death. This study aims to develop methods to identify at-risk patients to facilitate timely interventions that can improve outcomes and optimize healthcare resources. To accomplish this, we leveraged the Substance Misuse Data Commons to predict 30-day death or readmission from hospital discharge in patients with substance misuse. We explored several machine learning algorithms and approaches to integrate information from multiple data sources, such as structured features from a patient's electronic health record (EHR), unstructured clinical notes, socioeconomic data, and emergency medical services (EMS) data. Our gradient-boosted machine model, which combined structured EHR data, socioeconomic status, and EMS data, was the best-performing model (c-statistic 0.746 [95% CI: 0.732-0.759]), outperforming other machine learning methods and structured data source combinations. The addition of unstructured text did not improve performance, suggesting a need for further exploration of how to leverage unstructured data effectively. Feature importance plots highlighted the importance of prior hospital and EMS encounters and discharge disposition in predicting our primary outcome. In conclusion, we integrated multiple data sources that offer complementary information from data sources beyond the typically used EHRs for risk assessment in patients with substance misuse.</p>","PeriodicalId":74465,"journal":{"name":"PLOS digital health","volume":"4 9","pages":"e0001008"},"PeriodicalIF":7.7,"publicationDate":"2025-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12445462/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145088431","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}