This review examined how wearable sensors, computer vision, and artificial intelligence (AI) measure or infer psychological states in tennis and racket sports. Outcomes included competitive stress, anxiety, mental fatigue, cognitive or attentional states, recovery, and flow or “zone” experiences. Computer-vision studies were treated as behavioural proxies unless a psychological target was reported. Reporting followed PRISMA 2020. A Scopus search identified 253 records from 2015–2025; 43 texts were assessed, and 10 studies met the criteria. Eligible studies involved racket-sport athletes, psychological/affective constructs or behavioural proxies, and digital sensing or computational modalities. Approaches included physiological, EEG, inertial, computer-vision, and multimodal methods. Findings indicate overall feasibility, but performance was not comparable because constructs, labels, samples, settings, and validation procedures differed. No external-validation dataset was identified. Cohen’s kappa could not be reconstructed, and FICO was an internal appraisal rubric. Priorities include longitudinal field validation, standardized labeling, psychometric anchoring, explainable models, and privacy-preserving data governance.