Validating a Machine Learning Model Using Multi-Sensor Data to Predict Fall Risk in Stroke Patients

What this Trial Is About

Researchers are evaluating a machine learning model designed to predict fall risk among stroke patients using multi-sensor electromyography EMG signals. This prospective, multicenter, open-label confirmatory trial aims to validate the safety and accuracy of this model, which categorizes patients into high-risk or low-risk groups for falls. The study intends to improve current fall risk assessments that mainly rely on physical exams and patient history. During the study, 80 stroke patients and 10 healthy adults will have EMG sensors placed on key lower limb muscles while performing standardized movements. The collected muscle activity data will be analyzed by the machine learning model and compared to traditional assessment tools like the Berg Balance Scale. The primary focus is to measure the models sensitivity and specificity in predicting fall risk. Participants will undergo a single visit where EMG data is collected, and their muscle activity is recorded. Researchers will follow up to confirm actual fall incidents for validation. The study measures include the sensitivity and specificity of the machine learning model at the time of this visit. Healthy volunteers will also participate to establish baseline EMG readings. The study is expected to end in April 2026.

Predicting Fall Risk in Stroke Patients Using a Machine Learning Model and Multi-Sensor Data