As global aging accelerates, the prevalence of hand joint disorders among the elderly is increasing, leading to reduced functional independence and quality of life. Traditional physical rehabilitation methods, although clinically effective, often suffer from low adherence due to their repetitive and monotonous nature, especially among older adults. This study addresses these challenges by developing an interactive rehabilitation game system that integrates real-time hand gesture recognition. The proposed system, developed using the Unity engine and MediaPipe, captures seven distinct hand gestures in real time and provides immediate feedback through visual and auditory channels. The game includes gamification elements such as rhythmic interactions, scoring systems, and adaptive difficulty levels, all designed with senior-friendly user interfaces to improve accessibility and engagement. A two-week training program was conducted with 50 elderly participants aged 65 and above, many of whom had symptoms related to arthritis, joint stiffness, and reduced grip strength. Pre- and post-assessments of range of motion (ROM) and grip strength were conducted, alongside subjective evaluations of usability and user satisfaction. Results showed statistically significant improvements in both ROM (average increase of 13°) and grip strength (average increase of 3 kg), with high levels of engagement and positive feedback regarding the system’s interactivity and motivational design. These findings suggest that interactive games based on real-time gesture recognition offer substantial potential as an effective and engaging tool for elderly hand rehabilitation, particularly in home-based and self-guided contexts.