竹村 匡正

Abstract Background Although artificial intelligence (AI) has been widely applied to electronic health record (EHR) data in hospital settings, its use in long-term care (LTC) facilities remains unexplored. Limited information technology infrastructure and unique challenges in LTC settings require a comprehensive examination of AI’s potential to enhance care quality and operational efficiency. This scoping review aimed to identify and map current AI applications utilizing EHR data in LTC facilities to inform future research and practice. Methods The Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for scoping reviews was used as a guideline to structure the review topics. The protocol of this scoping review was registered on the Open Science Framework. The inclusion criteria were EHR (participants), AI (concept), and LTC facilities (context), with no date restrictions, but limited to original articles published in English. A systematic search was performed on MEDLINE, CINAHL, Cochrane Library databases, and SCOPUS by information specialists. Results The search included original articles published in English up to October 10, 2024. Ten studies met the inclusion criteria. Most were retrospective studies ( n  = 7), alongside two longitudinal and one development study. AI methods included natural language processing, machine learning algorithms (e.g., XGBoost, Random Forest), Bayesian networks, and rule-based reasoning. Applications focused on predicting outcomes such as dementia-related symptoms, malnutrition, pressure injuries, delirium, falls, frailty, urinary tract infections, mortality, and nurse staffing optimization. Data sources included large-scale EHR datasets from nursing homes, assisted living facilities, and senior care communities, often containing over 1,000 resident records or 30,000 clinical notes. Conclusion AI applications in LTC settings show promise for enhancing predictive accuracy and care quality. However, current evidence is limited, and further research is needed to improve model generalizability, validate results prospectively, and integrate AI into routine practice in diverse aged care environments.