A systematic review shows that using inertial measurement units with AI to recognize functional activity in stroke patients is technically feasible. The review included 17 studies involving 224 stroke patients; 12 studies used inertial measurement units, and 11 were conducted in clinical settings. Convolutional neural networks and Random Forest were the most commonly used methods. The review concludes that this approach can support objective assessment and continuous monitoring, but real-world validation, long-term follow-up, and studies involving acute-phase and severe stroke populations remain insufficient.
Source:BIOENGINEER.ORG · Read Original · 前沿热点
Leave a Reply