AI-enabled wearable monitoring of athletes' psychological states: a systematic literature review

Abstract

The meeting point of artificial intelligence and wearable sensing has quietly reshaped precision sport psychology. Athletes can now be monitored continuously, with cognitive and emotional states tracked at the individual level, something traditional self-report methods never quite captured. Yet the evidence on how AI actually turns physiological signals into psychological insight is scattered across sport science, engineering, and digital health. This systematic review follows the PRISMA 2020 framework to pull that evidence together. A structured Scopus search returned 112 records published between 2020 and 2025. After duplicates were removed and screening ran in two stages, 74 studies fell away at the title-and-abstract stage and 28 more after full-text assessment. That left 10 studies for qualitative synthesis, with 40 contextual records supporting the analysis. Eligible studies had to involve athlete populations, wearable physiological sensing, AI or machine-learning techniques, and psychological, cognitive, or affective outcomes. Three themes stood out. Electroencephalography combined with deep learning dominates cognitive-state classification. Multimodal integration of cardiac, electrodermal, muscular, and respiratory signals supports emotion and stress recognition. And telling psychological stress apart from physical stress remains the core methodological hurdle. What the evidence shows is that personalized, real-time psychological feedback is technically within reach, but it is still held back by small samples, laboratory settings, and limited external validation. Future work should push toward ecologically valid field studies, standardized labeling of psychological states, explainable AI models, and stronger data privacy for athletes. That is the path to real deployment in competitive sport.