JMIR Aging最新文献

{"title":"Using Machine Learning to Predict Cognitive Decline in Older Adults From the Chinese Longitudinal Healthy Longevity Survey: Model Development and Validation Study.","authors":"Hao Ren, Yiying Zheng, Changjin Li, Fengshi Jing, Qiting Wang, Zeyu Luo, Dongxiao Li, Deyi Liang, Weiming Tang, Li Liu, Weibin Cheng","doi":"10.2196/67437","DOIUrl":"https://doi.org/10.2196/67437","url":null,"abstract":"<p><strong>Background: </strong>Cognitive impairment, indicative of Alzheimer disease and other forms of dementia, significantly deteriorates the quality of life of older adult populations and imposes considerable burdens on families and health care systems worldwide. The early identification of individuals at risk for cognitive impairment through a convenient and rapid method is crucial for the timely implementation of interventions.</p><p><strong>Objective: </strong>The objective of this study was to explore the application of machine learning (ML) to integrate blood biomarkers, life behaviors, and disease history to predict the decline in cognitive function.</p><p><strong>Methods: </strong>This approach uses data from the Chinese Longitudinal Healthy Longevity Survey. A total of 2688 participants aged 65 years or older from the 2008-2009, 2011-2012, and 2014 Chinese Longitudinal Healthy Longevity Survey waves were included, with cognitive impairment defined as a Mini-Mental State Examination (MMSE) score below 18. The dataset was divided into a training set (n=1331), an internal test set (n=333), and a prospective validation set (n=1024). Participants with a baseline MMSE score of less than 18 were excluded from the cohort to ensure a more accurate assessment of cognitive function. We developed ML models that integrate demographic information, health behaviors, disease history, and blood biomarkers to predict cognitive function at the 3-year follow-up point, specifically identifying individuals who are at risk of experiencing significant declines in cognitive function by that time. Specifically, the models aimed to identify individuals who would experience a significant decline in their MMSE scores (less than 18) by the end of the follow-up period. The performance of these models was evaluated using metrics including accuracy, sensitivity, and the area under the receiver operating characteristic curve.</p><p><strong>Results: </strong>All ML models outperformed the MMSE alone. The balanced random forest achieved the highest accuracy (88.5% in the internal test set and 88.7% in the prospective validation set), albeit with a lower sensitivity, while logistic regression recorded the highest sensitivity. SHAP (Shapley Additive Explanations) analysis identified instrumental activities of daily living, age, and baseline MMSE scores as the most influential predictors for cognitive impairment.</p><p><strong>Conclusions: </strong>The incorporation of blood biomarkers, along with demographic, life behavior, and disease history into ML models offers a convenient, rapid, and accurate approach for the early identification of older adult individuals at risk of cognitive impairment. This method presents a valuable tool for health care professionals to facilitate timely interventions and underscores the importance of integrating diverse data types in predictive health models.</p>","PeriodicalId":36245,"journal":{"name":"JMIR Aging","volume":"8 ","pages":"e67437"},"PeriodicalIF":5.0,"publicationDate":"2025-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12058036/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144042420","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":"Perceptions of the Use of Mobile Apps to Assess Sleep-Dependent Memory in Older Adults With Subjective and Objective Cognitive Impairment: Focus Group Approach.","authors":"Aaron Lam, Simone Simonetti, Angela D'Rozario, David Ireland, DanaKai Bradford, Jurgen Fripp, Sharon L Naismith","doi":"10.2196/68147","DOIUrl":"https://doi.org/10.2196/68147","url":null,"abstract":"<p><strong>Background: </strong>Sleep-dependent memory (SDM) is the phenomenon where newly obtained memory traces are consolidated from short-term memory stores to long-term memory, underpinning memory for daily life. Administering SDM tasks presents considerable challenges, particularly for older adults with memory concerns, due to the need for sleep laboratories and research staff being present to administer the task. In response, we have developed a prototype mobile app aimed at automating the data collection process.</p><p><strong>Objective: </strong>This study investigates the perspectives of older adults, with subjective or objective cognitive impairment, regarding barriers and facilitators to using a new mobile app for at-home assessment of SDM.</p><p><strong>Methods: </strong>In total, 11 participants aged 50 years and older were recruited from the Healthy Brain Ageing memory clinic, a specialized research memory clinic that focuses on the assessment and early intervention of cognitive decline. Two focus groups were conducted and thematically analyzed using NVivo (version 13; Lumivero).</p><p><strong>Results: </strong>On average, participants were aged 68.5 (SD 5.1) years, and 4/11 were male. Eight participants had subjective cognitive impairment, and 3 participants had mild cognitive (objective) impairment. Two main themes emerged from the focus groups, shedding light on participants' use of mobile phones and the challenges and facilitators associated with transitioning from traditional laboratory-based assessments to home assessments. These challenges include maintaining accurate data, engaging with humans versus robots, and ensuring accessibility and task compliance. Additionally, potential solutions to these challenges were identified.</p><p><strong>Conclusions: </strong>Our findings underscore the importance of app flexibility in accommodating diverse user needs and preferences as well as in overcoming barriers. While some individuals required high-level assistance, others expressed the ability to navigate the app independently or with minimal support. In conclusion, older adults provided valuable insights into the app modifications, user needs, and accessibility requirements enabling home-based SDM assessment.</p>","PeriodicalId":36245,"journal":{"name":"JMIR Aging","volume":"8 ","pages":"e68147"},"PeriodicalIF":5.0,"publicationDate":"2025-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12052296/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144004613","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":"Relationship Between Within-Session Digital Motor Skill Acquisition and Alzheimer Disease Risk Factors Among the MindCrowd Cohort: Cross-Sectional Descriptive Study.","authors":"Andrew Hooyman, Matt J Huentelman, Matt De Both, Lee Ryan, Kevin Duff, Sydney Y Schaefer","doi":"10.2196/67298","DOIUrl":"https://doi.org/10.2196/67298","url":null,"abstract":"<p><strong>Background: </strong>Previous research has shown that in-lab motor skill acquisition (supervised by an experimenter) is sensitive to biomarkers of Alzheimer disease (AD). However, remote unsupervised screening of AD risk through a skill-based task via the web has the potential to sample a wider and more diverse pool of individuals at scale.</p><p><strong>Objective: </strong>The purpose of this study was to examine a web-based motor skill game (\"Super G\") and its sensitivity to risk factors of AD (eg, age, sex, APOE ε4 carrier status, and verbal learning deficits).</p><p><strong>Methods: </strong>Emails were sent to 662 previous MindCrowd participants who had agreed to be contacted for future research and have their APOE ε4 carrier status recorded and those who were at least 45 years of age or older. Participants who chose to participate were redirected to the Super G site where they completed the Super G task using their personal computer remotely and unsupervised. Once completed, different Super G variables were derived. Linear and logistic multivariable regression was used to examine the relationship between available AD risk factors (age, sex, APOE ε4 carrier status, and verbal learning) and distinct Super G performance metrics.</p><p><strong>Results: </strong>Fifty-four participants (~8% response rate) from the MindCrowd web-based cohort (mean age of 62.39 years; 39 females; and 23 APOE ε4 carriers) completed 75 trials of Super G. Results show that Super G performance was significantly associated with each of the targeted risk factors. Specifically, slower Super G response time was associated with being an APOE ε4 carrier (odds ratio 0.12, 95% CI 0.02-0.44; P=.006), greater Super G time in target (TinT) was associated with being male (odds ratio 32.03, 95% CI 3.74-1192,61; P=.01), and lower Super G TinT was associated with greater age (β -3.97, 95% CI -6.64 to -1.30; P=.005). Furthermore, a sex-by-TinT interaction demonstrated a differential relationship between Super G TinT and verbal learning depending on sex (βmale:TinT 6.77, 95% CI 0.34-13.19; P=.04).</p><p><strong>Conclusions: </strong>This experiment demonstrated that this web-based game, Super G, has the potential to be a skill-based digital biomarker for screening of AD risk on a large scale with relatively limited resources.</p>","PeriodicalId":36245,"journal":{"name":"JMIR Aging","volume":"8 ","pages":"e67298"},"PeriodicalIF":5.0,"publicationDate":"2025-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12045524/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144051731","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":"Association Between Sleep Duration and Cognitive Frailty in Older Chinese Adults: Prospective Cohort Study.","authors":"Ruixue Cai, Jianqian Chao, Chenlu Gao, Lei Gao, Kun Hu, Peng Li","doi":"10.2196/65183","DOIUrl":"10.2196/65183","url":null,"abstract":"<p><strong>Background: </strong>Disturbed sleep patterns are common among older adults and may contribute to cognitive and physical declines. However, evidence for the relationship between sleep duration and cognitive frailty, a concept combining physical frailty and cognitive impairment in older adults, is lacking.</p><p><strong>Objective: </strong>This study aimed to examine the associations of sleep duration and its changes with cognitive frailty.</p><p><strong>Methods: </strong>We analyzed data from the 2008-2018 waves of the Chinese Longitudinal Healthy Longevity Survey. Cognitive frailty was rendered based on the modified Fried frailty phenotype and Mini-Mental State Examination. Sleep duration was categorized as short (<6 h), moderate (6-9 h), and long (>9 h). We examined the association of sleep duration with cognitive frailty status at baseline using logistic regressions and with the future incidence of cognitive frailty using Cox proportional hazards models. Restricted cubic splines were used to explore potential nonlinear associations.</p><p><strong>Results: </strong>Among 11,303 participants, 1298 (11.5%) had cognitive frailty at baseline. Compared to participants who had moderate sleep duration, the odds of having cognitive frailty were higher in those with long sleep duration (odds ratio 1.71, 95% CI 1.48-1.97; P<.001). A J-shaped association between sleep duration and cognitive frailty was also observed (P<.001). Additionally, during a mean follow-up of 6.7 (SD 2.6) years among 5201 participants who were not cognitively frail at baseline, 521 (10%) participants developed cognitive frailty. A higher risk of cognitive frailty was observed in participants with long sleep duration (hazard ratio 1.32, 95% CI 1.07-1.62; P=.008).</p><p><strong>Conclusions: </strong>Long sleep duration was associated with cognitive frailly in older Chinese adults. These findings provide insights into the relationship between sleep duration and cognitive frailty, with potential implications for public health policies and clinical practice.</p>","PeriodicalId":36245,"journal":{"name":"JMIR Aging","volume":"8 ","pages":"e65183"},"PeriodicalIF":4.8,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12043274/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144018003","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: Machine Learning Models for Frailty Classification of Older Adults in Northern Thailand: Model Development and Validation Study.","authors":"Natthanaphop Isaradech, Wachiranun Sirikul, Nida Buawangpong, Penprapa Siviroj, Amornphat Kitro","doi":"10.2196/75690","DOIUrl":"https://doi.org/10.2196/75690","url":null,"abstract":"","PeriodicalId":36245,"journal":{"name":"JMIR Aging","volume":"8 ","pages":"e75690"},"PeriodicalIF":5.0,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12040286/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143989500","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 Emergency Care With Digital Twins.","authors":"Haoran Li, Jingya Zhang, Ning Zhang, Bin Zhu","doi":"10.2196/71777","DOIUrl":"https://doi.org/10.2196/71777","url":null,"abstract":"<p><p>Digital twins-dynamic and real-time simulations of systems or environments-represent a paradigm shift in emergency medicine. We explore their applications across prehospital care, in-hospital management, and recovery. By integrating real-time data, wearable technology, and predictive analytics, digital twins hold the promise of optimizing resource allocation, advancing precision medicine, and tailoring rehabilitation strategies. Moreover, we discuss the challenges associated with their implementation, including data resolution, biological heterogeneity, and ethical considerations, emphasizing the need for actionable frameworks that balance innovation with data governance and public trust.</p>","PeriodicalId":36245,"journal":{"name":"JMIR Aging","volume":"8 ","pages":"e71777"},"PeriodicalIF":5.0,"publicationDate":"2025-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12053090/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144056856","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 of a Longitudinal Model for Disability Prediction in Older Adults in China: Analysis of CHARLS Data (2015-2020).","authors":"Jingjing Chu, Ying Li, Xinyi Wang, Qun Xu, Zherong Xu","doi":"10.2196/66723","DOIUrl":"https://doi.org/10.2196/66723","url":null,"abstract":"<p><strong>Background: </strong>Disability profoundly affects older adults' quality of life and imposes considerable burdens on health care systems in China's aging society. Timely predictive models are essential for early intervention.</p><p><strong>Objective: </strong>We aimed to build effective predictive models of disability for early intervention and management in older adults in China, integrating physical, cognitive, physiological, and psychological factors.</p><p><strong>Methods: </strong>Data from the China Health and Retirement Longitudinal Study (CHARLS), spanning from 2015 to 2020 and involving 2450 older individuals initially in good health, were analyzed. The dataset was randomly divided into a training set with 70% data and a testing set with 30% data. LASSO regression with 10-fold cross-validation identified key predictors, which were then used to develop an Extreme Gradient Boosting (XGBoost) model. Model performance was evaluated using receiever operating characteristic curves, calibration curves, and clinical decision and impact curves. Variable contributions were interpreted using SHapley Additive exPlanations (SHAP) values.</p><p><strong>Results: </strong>LASSO regression was used to evaluate 36 potential predictors, resulting in a model incorporating 9 key variables: age, hand grip strength, standing balance, the 5-repetition chair stand test (CS-5), pain, depression, cognition, respiratory function, and comorbidities. The XGBoost model demonstrated an area under the curve of 0.846 (95% CI 0.825-0.866) for the training set and 0.698 (95% CI 0.654-0.743) for the testing set. Calibration curves demonstrated reliable predictive accuracy, with mean absolute errors of 0.001 and 0.011 for the training and testing sets, respectively. Clinical decision and impact curves demonstrated significant utility across risk thresholds. SHAP analysis identified pain, respiratory function, and age as top predictors, highlighting their substantial roles in disability risk. Hand grip and the CS-5 also significantly influenced the model. A web-based application was developed for personalized risk assessment and decision-making.</p><p><strong>Conclusions: </strong>A reliable predictive model for 5-year disability risk in Chinese older adults was developed and validated. This model enables the identification of high-risk individuals, supports early interventions, and optimizes resource allocation. Future efforts will focus on updating the model with new CHARLS data and validating it with external datasets.</p>","PeriodicalId":36245,"journal":{"name":"JMIR Aging","volume":"8 ","pages":"e66723"},"PeriodicalIF":5.0,"publicationDate":"2025-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12021300/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144022550","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":"Unveiling the Frailty Spatial Patterns Among Chilean Older Persons by Exploring Sociodemographic and Urbanistic Influences Based on Geographic Information Systems: Cross-Sectional Study.","authors":"Yony Ormazábal, Diego Arauna, Juan Carlos Cantillana, Iván Palomo, Eduardo Fuentes, Carlos Mena","doi":"10.2196/64254","DOIUrl":"https://doi.org/10.2196/64254","url":null,"abstract":"<p><strong>Background: </strong>Frailty syndrome increases the vulnerability of older adults. The growing proportion of older adults highlights the need to better understand the factors contributing to the prevalence of frailty. Current evidence suggests that geomatic tools integrating geolocation can provide valuable information for implementing preventive measures by enhancing the urban physical environment.</p><p><strong>Objective: </strong>The aim of this study was to analyze the relationship between various elements of the urban physical environment and the level of frailty syndrome in older Chilean people.</p><p><strong>Methods: </strong>A cohort of 251 adults aged 65 years or older from Talca City, Chile, underwent comprehensive medical assessments and were geographically mapped within a Geographic Information Systems database. Frailty was determined using the Fried frailty criteria. The spatial analysis of the frailty was conducted in conjunction with layers depicting urban physical facilities within the city, including vegetables and fruit shops, senior centers or communities, pharmacies, emergency health centers, main squares and parks, family or community health centers, and sports facilities such as stadiums.</p><p><strong>Results: </strong>The studied cohort was composed of 187 women and 64 men, with no significant differences in age and BMI between genders. Frailty prevalence varied significantly across clusters, with Cluster 3 showing the highest prevalence (14/47, P=.01). Frail individuals resided significantly closer to emergency health centers (960 [SE 904] m vs 1352 [SE 936] m, P=.04), main squares/parks (1550 [SE 130] m vs. 2048 [SE 105] m, P=.03), and sports fields (3040 [SE 236] m vs 4457 [SE 322]m, P=.04) compared with nonfrail individuals. There were no significant differences in urban quality index across frailty groups, but frail individuals lived in areas with higher population density (0.013 [SE 0.001] vs 0.01 [SE 0.0007], P=.03).</p><p><strong>Conclusions: </strong>Frail individuals exhibit geospatial patterns suggesting intentional proximity to health facilities, sports venues, and urban facilities, revealing associations with adaptive responses to frailty and socioeconomic factors. This highlights the crucial intersection of urban environments and frailty, which is important for geriatric medicine and public health initiatives.</p>","PeriodicalId":36245,"journal":{"name":"JMIR Aging","volume":"8 ","pages":"e64254"},"PeriodicalIF":5.0,"publicationDate":"2025-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12021301/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143989504","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-Driven Biological Age Prediction Model Using Comprehensive Health Checkup Data: Development and Validation Study.","authors":"Chang-Uk Jeong, Jacob S Leiby, Dokyoon Kim, Eun Kyung Choe","doi":"10.2196/64473","DOIUrl":"https://doi.org/10.2196/64473","url":null,"abstract":"<p><strong>Background: </strong>The global increase in life expectancy has not shown a similar rise in healthy life expectancy. Accurate assessment of biological aging is crucial for mitigating diseases and socioeconomic burdens associated with aging. Current biological age prediction models are limited by their reliance on conventional statistical methods and constrained clinical information.</p><p><strong>Objective: </strong>This study aimed to develop and validate an aging clock model using artificial intelligence, based on comprehensive health check-up data, to predict biological age and assess its clinical relevance.</p><p><strong>Methods: </strong>We used data from Koreans who underwent health checkups at the Seoul National University Hospital Gangnam Center as well as from the Korean Genome and Epidemiology Study. Our model incorporated 27 clinical factors and employed machine learning algorithms, including linear regression, least absolute shrinkage and selection operator, ridge regression, elastic net, random forest, support vector machine, gradient boosting, and K-nearest neighbors. Model performance was evaluated using adjusted R2 and the mean squared error (MSE) values. Shapley Additive exPlanation (SHAP) analysis was conducted to interpret the model's predictions.</p><p><strong>Results: </strong>The Gradient Boosting model achieved the best performance with a mean (SE) MSE of 4.219 (0.14) and a mean (SE) R2 of 0.967 (0.001). SHAP analysis identified significant predictors of biological age, including kidney function markers, gender, glycated hemoglobin level, liver function markers, and anthropometric measurements. After adjusting for the chronological age, the predicted biological age showed strong associations with multiple clinical factors, such as metabolic status, body compositions, fatty liver, smoking status, and pulmonary function.</p><p><strong>Conclusions: </strong>Our aging clock model demonstrates a high predictive accuracy and clinical relevance, offering a valuable tool for personalized health monitoring and intervention. The model's applicability in routine health checkups could enhance health management and promote regular health evaluations.</p>","PeriodicalId":36245,"journal":{"name":"JMIR Aging","volume":"8 ","pages":"e64473"},"PeriodicalIF":5.0,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12007724/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144047076","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":"Identifying Deprescribing Opportunities With Large Language Models in Older Adults: Retrospective Cohort Study.","authors":"Vimig Socrates, Donald S Wright, Thomas Huang, Soraya Fereydooni, Christine Dien, Ling Chi, Jesse Albano, Brian Patterson, Naga Sasidhar Kanaparthy, Catherine X Wright, Andrew Loza, David Chartash, Mark Iscoe, Richard Andrew Taylor","doi":"10.2196/69504","DOIUrl":"https://doi.org/10.2196/69504","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;Polypharmacy, the concurrent use of multiple medications, is prevalent among older adults and associated with increased risks for adverse drug events including falls. Deprescribing, the systematic process of discontinuing potentially inappropriate medications, aims to mitigate these risks. However, the practical application of deprescribing criteria in emergency settings remains limited due to time constraints and criteria complexity.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Objective: &lt;/strong&gt;This study aims to evaluate the performance of a large language model (LLM)-based pipeline in identifying deprescribing opportunities for older emergency department (ED) patients with polypharmacy, using 3 different sets of criteria: Beers, Screening Tool of Older People's Prescriptions, and Geriatric Emergency Medication Safety Recommendations. The study further evaluates LLM confidence calibration and its ability to improve recommendation performance.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;We conducted a retrospective cohort study of older adults presenting to an ED in a large academic medical center in the Northeast United States from January 2022 to March 2022. A random sample of 100 patients (712 total oral medications) was selected for detailed analysis. The LLM pipeline consisted of two steps: (1) filtering high-yield deprescribing criteria based on patients' medication lists, and (2) applying these criteria using both structured and unstructured patient data to recommend deprescribing. Model performance was assessed by comparing model recommendations to those of trained medical students, with discrepancies adjudicated by board-certified ED physicians. Selective prediction, a method that allows a model to abstain from low-confidence predictions to improve overall reliability, was applied to assess the model's confidence and decision-making thresholds.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;The LLM was significantly more effective in identifying deprescribing criteria (positive predictive value: 0.83; negative predictive value: 0.93; McNemar test for paired proportions: χ&lt;sup&gt;2&lt;/sup&gt;&lt;sub&gt;1&lt;/sub&gt;=5.985; P=.02) relative to medical students, but showed limitations in making specific deprescribing recommendations (positive predictive value=0.47; negative predictive value=0.93). Adjudication revealed that while the model excelled at identifying when there was a deprescribing criterion related to one of the patient's medications, it often struggled with determining whether that criterion applied to the specific case due to complex inclusion and exclusion criteria (54.5% of errors) and ambiguous clinical contexts (eg, missing information; 39.3% of errors). Selective prediction only marginally improved LLM performance due to poorly calibrated confidence estimates.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;This study highlights the potential of LLMs to support deprescribing decisions in the ED by effectively filtering relevant criteria. However, challenges remain in applyin","PeriodicalId":36245,"journal":{"name":"JMIR Aging","volume":"8 ","pages":"e69504"},"PeriodicalIF":5.0,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12032504/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144051810","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}