Abstract / Summary
Abstract Background . Artificial intelligence (AI) and machine learning (ML) systems are increasingly proposed for elderly care settings such as nursing homes, assisted living, and home-based care. Existing reporting guidelines for AI studies have primarily been developed for medical and diagnostic applications, and their adequacy for elderly care contexts remains unclear. This rapid scoping review aimed to characterize current reporting and evaluation practices of AI and ML studies intended for elderly care settings. Methods . PubMed and IEEE Xplore were searched on 2026-02-18 for peer-reviewed studies published from 2020 onwards reporting empirical AI or ML systems intended to support tasks in elderly care contexts. Purely diagnostic or prognostic medical AI studies were excluded. A random sample of 200 retrieved records was screened by two reviewers as part of a rapid scoping review approach. Data were extracted using a standardized form assessing reporting of contextual characteristics, data and population information, evaluation practices, bias, uncertainty, limitations, and deployment-related aspects. Results were synthesized descriptively. Results. A total of 69 studies were included. Most focused on early-stage model development and internal technical validation, while external validation and evaluation in real-world care settings were uncommon. Use cases were dominated by monitoring and safety-oriented applications across home-based and institutional care settings. Reporting completeness varied substantially across domains. While intended system roles and data sources were commonly reported, reporting of missing data handling, external validation, uncertainty quantification, calibration, and subgroup analyses was limited. Intended users, decision-makers, and deployment contexts were frequently only partially specified. Discussion. Current reporting practices for AI and ML systems in elderly care may focus more on technical model development rather than system-level considerations given the complexity of care environments. The identified gaps suggest that existing medical AI reporting guidelines may not fully address the contextual and operational requirements of elderly care applications.