 The motor function assessment of post-stroke patients is important for their recovery. However, conventional methods are subjective and rely heavily on the experience of therapists. This paper proposes a novel approach to automate the assessment process by analyzing the patient's performance during three common activities of daily living, ADL's tooth brushing, face washing, and drinking. The authors systematically searched for the best parameters for the data segmentation, feature extraction, feature optimization, and classifier selection. They found that the best results could be achieved with three sensor modules, simple square integral, slope sign change, and modified mean absolute value one and two features, principle component analysis for feature optimization, and logistic regression classifier. This approach can accurately identify the stages of stroke patients using only these parameters, which is more cost effective than other approaches. Furthermore, this study is the first to explore the use of sensor configuration and parameter optimization in the BRS classification framework. This article was authored by Long Meng, Sinyu Jiang, Hai Baqin, and others. We are article.tv, links in the description below.