Advancements in computing, networking, and sensor technologies have led to unprecedented developments in medical equipment and devices. An example of such implantable and wearable medical devices is the brain neurostimulation device, utilized for diagnostic, monitoring, and therapeutic functions in epilepsy. This article presents a runtime model based on reinforcement learning for validating the performance of this medical device, particularly under critical conditions. The proposed approach models the operational structure of the device using brain sensors within Petri nets and employs physiological body sensors as learning data to enable adaptive and refined decision-making during runtime. Given the fuzzy and uncertain nature of the body, sensor values and operational rules are simulated in a fuzzy manner, incorporated into the runtime conditions of the Petri net. In this research, to verify the validity of the execution time, the verification of the time of execution of the behaviour and decisions produced by the device with a formal characteristics in each decision cycle is determined so that in case of consistency, error is identified during execution and control policy through the reinforcement learning mechanism and physiological feedback of the decision. Performance validation of proposed framework is presented and evaluated by running the execution time in the form of different operational scenarios and is considered as an effective step in the design of the next generation of adaptive and intelligent neurostimulation devices by achiving the 100% recall rate and 100% correction coverage.