Emotion Detection Systems in Healthcare: A Review of Artificial Intelligence for Elderly Mental Health Monitoring

Abstract

The rapidly increasing global population of older adults presents unprecedented challenges in addressing depression, anxiety, and social isolation, while traditional mental health services face limitations in accessibility, continuity of care, and resource availability, highlighting the need for scalable solutions. This systematic narrative review examines artificial intelligence-based emotion detection systems for elderly mental health monitoring, synthesizing evidence on four major technological approaches: conversational agents, multimodal emotion recognition systems, wearable physiological sensing technologies, and artificial intelligence-driven clinical decision support systems. Meta-analytic evidence from 12 randomized controlled trials (n=1,847) demonstrates that conversational agents produce moderate-to-large reductions in depressive symptoms (standardized mean difference = -0.58, 95% CI: -0.82 to -0.34, p<0.001) among elderly populations. Despite promising outcomes, significant challenges remain, including ethi cal concerns about ensuring equitable access across socioeconomic settings, mitigating algorithmic bias from non-representative training data, and addressing transparency in decision-making processes, as well as technical challenges involving integration of complex systems into clinical workflows and accurate interpretation of physiological signals as emotional indicators. Substantial evidence gaps persist, particularly the lack of longitudinal effectiveness studies and pragmatic trials in low- and middle income settings, making it essential to address these methodological and translational challenges for safe, effective, and equitable implementation in elderly mental health care.