Incorporating Random Forests and Explainable AI for Upper Motor Function Assessment with the Wolf Motor Function Test

Rehabilitation therapies play a crucial role in restoring upper motor functions among individuals with neurological impairments. One widely accepted assessment tool for evaluating motor functions in individuals with neurological disorders, such as stroke, is the Wolf Motor Function Test (WMFT). This study highlights the potential of eXplainable Artificial Intelligence (XAI) techniques in enhancing upper motor functions rehabilitation using the WMFT. By integrating explainability into AI-driven rehabilitation systems, not only does it improve clinical decision-making, but it also empowers patients in their recovery journey. This integration contributes to the advancement of personalized and effective rehabilitation strategies for individuals with neurological impairments. The proposed system is both simple and effective, as it combines real- time sensor data collected during WMFT exercises with Random Forests classification and the Anchor XAI algorithm. This integration allows for the provision of personalized and adaptive rehabilitation strategies. To evaluate the accuracy of the developed model, a 7-fold cross-validation was conducted, resulting in an impressive accuracy rate of 92.80%. Additionally, the precision of the XAI model (Anchor) ranged between 0.94 and 0.97, further demonstrating its reliability.