Introduction: Advancements in artificial intelligence (AI) offer promise for addressing geriatric concerns such as falls, frailty, dementia, polypharmacy, and homecare provision, which significantly impact older adults (55+). This scoping review examined the current state of AI technology implementation in the care of older adults.
Methodology: Our search identified 1816 studies published between 2020 and 2023. Two reviewers screened titles and abstracts, followed by full-text review of 442 articles, with conflicts resolved by a third reviewer. Among the 169 articles meeting inclusion criteria, 97 corresponded to our research question of current and near-term clinical utility of AI across five key geriatric themes.
Results: AI technologies applied to geriatric concerns were categorized as: Prediction (51.5%), Classification (27.8%), Assistive Devices (11.3%), or Robot (9.3%). Predictive AI was seen in studies of falls (21.6%), dementia (15.5%), frailty (5.2%), homecare (4.1%), mild cognitive impairment (MCI) (3.1%) and delirium (2.1%). Classification using AI included two activities, detection of a condition or differentiation between individuals with or without a condition, and was used in dementia (17,5%), frailty (4.1%), polypharmacy (3.1%), and falls, delirium and MCI (1% respectively). Across all studies, only 9% reported an area-under-the-curve (AUC) value greater than 0.8, suggesting limited studies demonstrating clinical utility (Figure 1). Few studies were randomized-controlled trials (RCTs) (9.3%), Most studies focused on dementia (26.8%), or specifically Alzheimer’s disease (17,5%), falls (24.7%) and frailty (10.3%). No studies compared AI performance with non-AI technologies and 75.2% of studies did not report the ethnicity of participants. One study reported on Black participants and no studies included Indigenous participants.
Conclusion: Most AI studies focused on older adults remain theoretical, lacking near-term clinical utility. Future research should emphasize RCTs, AUC reporting, comparison against established non-AI clinical tools and AI use among participants of different ethnicities to support real-world application.
Authors:
Kelly Kay Ph.D, Justin Kim MD, Asghar Khan, Daniel Yacoub MD, Sai-Amrit Maharaj Ph.D student, Nihal Haque MD
Presented at T-CAIREM 2025