The Internet of Medical Things (IoMT) is the new digital healthcare application paradigm that offers many healthcare services to users. IoMT-based emerging healthcare applications such as cyborgs, the combination of advanced artificial intelligence (AI) robots, and doctors performing surgical operations remotely from hospitals to patients in their homes. For instance, robot-based knee replacement procedures, and thigh medical care real-time performance monitoring systems are cyborg applications. The paper introduces the multi-agent federated reinforcement learning policy (MFRLP) indicated in mobile and fog agents based on the socket remote procedure call (RPC) paradigm. The goal is to design a consumer-centric cyborg-efficient training testing system that executes the overall application mechanism with minimum delays in the IoMT system. The study develops the RPC based on reinforcement learning and federated learning that adopts dynamic changes in the environment for cyborg applications. As a result, MFRLP minimized the training and testing in the mobile and fog environments by 50%, local processing time by 40%, and processing time by 50% compared to existing machine learning (ML) methods for cyborg applications. The code is publicly available at https://github.com/prayagtiwari/CIoMT IEEE