Research on upper limb rehabilitation assessment model based on belief rule base.

Abstract

Rehabilitation assessments hold an irreplaceable role in the field of rehabilitative therapy. However, due to the subjectivity of traditional physicians and the variability of patient conditions, this leads to a lack of detailed grading and inaccurate assessment results. To address this issue, we developed an upper limb rehabilitation evaluation model. This model integrates muscle strength assessment methods and the Belief Rule Base (BRB), along with qualitative knowledge such as clinical rehabilitation theories and expert experiences. It also utilizes training data from actual patients, collected by an upper limb rehabilitation robot. We then optimized the BRB model's evaluation accuracy using the Fmincon algorithm and compared its result with commonly used methods such as the Back Propagation (BP) neural network and Support Vector Machine (SVM). This comparison validated the effectiveness and advancement of our BRB approach. This work has laid both a theoretical and practical groundwork for developing a clinical decision support system based on the BRB for upper limb rehabilitation evaluations.