Hand and finger exercises only work if patients do them, and repetition makes many people stop. Exergames hide the exercise inside the game: the player moves their hands to control the screen. The open question is which gestures a person can comfortably make.
A team from Okayama University, Universitas Brawijaya and partner institutions built a small exergame to answer that. The system uses a web camera and the Mediapipe library to recognise hand shapes in real time. Five gestures map to five controls: up, down, left, right and release.
The team began with 32 simple one hand gestures. Sixteen students made each one and rated how easy it felt on a five point scale. Ten survived. The weakest was gesture 27 at 1.7 out of 5, a shape that forces the middle and little fingers together.
Because the maze needs five controls, the team split the top ten into two groups. The best used gestures 1, 2, 4, 12 and 29, the second best 3, 7, 11, 21 and 8. Any gap in performance then comes from the gestures themselves.
A separate test with 30 students measured how often the camera read each gesture correctly. The best group was recognised more reliably. The misses had practical causes: dim lighting, low camera resolution, overlapping fingers, and hands held too far away.
The study has limits. Most participants were young students rather than rehabilitation patients, so the ratings could shift with age or with an injured hand. Even so, the pattern is clear: easy gestures produced better play and better usability scores without changing the game underneath.
Husna et al. (2025). An Investigation of Hand Gestures for Controlling Video Games in a Rehabilitation Exergame System. Computers.
