Facial expression analysis using artificial intelligence (AI) represents an emerging approach for assessing mental health, particularly in neurocognitive disorders. This study encompassed observational investigations that assessed facial expressions in individuals aged 60 years and above. A comprehensive literature search was carried out across PubMed, Scopus, EMBASE, and Web of Science. Risk of bias and study quality were assessed using the QUADAS-2 and CLAIM tools. Descriptive analysis and meta-analysis of proportions were performed using STATA version 19. The pooled effect size (ES) was calculated using a random-effects model (DerSimonian–Laird method), and results were presented with corresponding 95% confidence intervals (CI). Six studies were analyzed, comprising a total of 433 participants aged over 60 years, representing diverse AI applications in the detection of neurocognitive disorders. The disorders evaluated included mild cognitive impairment (MCI) (37.4%), dementia (29.3%), and Alzheimer’s disease (AD) (33.3%). Most studies (83.3%) used video-based facial recordings analyzed through deep learning algorithms and emotion recognition models. The pooled meta-analysis demonstrated that AI-based facial recognition algorithms achieved a high overall detection accuracy in older adults (ES = 0.84; 95% CI: 0.77–0.91), with the best performance observed in Alzheimer’s disease (ES = 0.93; 95% CI: 0.89–0.97). AI-based facial analysis demonstrates high, robust, and non-invasive accuracy for the early and differential detection of neurocognitive disorders, including MCI, dementia-related conditions, and AD, in older adults.
Runzer-Colmenares et al. (Wed,) studied this question.