Neurological diseases are one of the major causes of death and disability in the world. One of the limiting factors in dealing with neurological diseases is the lack of specialized staff, which increases costs and reduces the quality of care. This research aims to study the degree to which AI can support the workload of physical therapists in the context of home rehabilitation. To this end, a fuzzy Decision-Support System capable of automatically suggesting and adapting personalized exercise routines has been developed. The system employs XAI techniques to increase the interpretability of the model, and facilitate the integration of the system into established human workflows. In addition to the development and preliminary testing of the initial prototype, the research has been focused on generating advanced rehabilitation metrics from patient pose data, such as compensatory patterns and fatigue. The major remaining milestone until the dissertation is the comparison of the fuzzy inference system, which is used to implement the prototype, with Large Language Models (LLMs).