Video-based detection of hemiparetic weakness side in post-stroke patient

One of the key symptoms of stroke is hemiparesis, a weakness with loss of movement in one or more sides of the body. Early detection of stroke-related weakness can improve patient outcomes and minimize the risk of long-term disability. This study presents a novel deep learning approach for efficiently and objectively detecting and quantifying stroke-related weakness. This approach addresses the limitations of current technologies, which can be invasive and time-consuming. The study describes a model that processes video data to classify left or right-sided stroke-induced motor weakness. The model extracts the movement of key human body joints and applies time series analysis to classify the stroke-affected side of the body. The model was developed using data collected from 23 post-stroke patients in a clinical setting over 48 hours.