After a heart attack or a stroke, the patient needs rehabilitation; nevertheless, obviously, conventional approaches are costly, time-consuming, and need a highly qualified staff, which excludes the majority of patients. As part of the proposed solution, this research incorporates Rehabilitation Internet-of-Things (RIoT) that uses Mediapipe for hand gesture detection and voice to guide the exercises. The culmination of the system is to offer availability of computer vision coupled with speech recognition to evaluate the performance during the exercise and to report the extent of rehabilitation within the shortest time. In particular, these movements include flexion, extension of fingers, pinch using the thumb index finger, and opening/closing of the hand and full hand movement that helps in determining the degree of motion for performing movements during the rehabilitation exercises. The RIoT system acts as a voice-activated, on-the-body graphical display that helps the partly mobile users as they obtain real-time feedback from their hand gestures. The sensitivity of the deep learning-based gesture recognition and the speech synthesized is then tested and practiced on recovering patients before testing on the system platform. Thus, the system, in the framework of utilizing assistive automation for rehabilitation, releases the necessity to use human observers while still keeping the overall control by doctors or other healthcare managers. and enables the access to the high-quality rehabilitation therapy for patients, contributes to the decreased healthcare expenditures, and improve the outcomes of the overall patient rehabilitation.