Estimation of Functional Independence Measure Using Frequency Domain Feature

Strokes are the second leading cause necessitating long-term care in Japan. Stoke patients require long-term rehabilitation. Functional independence measure (FIM) is a measure of daily activities of stroke patients and is a reliable physical therapy assessment that can be used in a multidisciplinary setting. However, the assessment of FIM scores is typically done only by experienced nurses and difficult perform accurately at home. In this study, we developed a system to estimate the FIM score easily at home using gyro sensors that measure angular velocity from the target bodily movements. Two types of target bodily movements can occur: “rising from a bed” and “sit-to-stand (STS) movement”. To easily estimate the FIM motor score from daily movements, frequency domain features were calculated from the acquired angular velocity time-series data. Then, the FIM motor score can be estimated by performing a Gaussian process regression analysis with the features as explanatory variables. The FIM motor score was objectively assessed by a nurse. Finally, we performed a series of experiments to evaluate the proposed method. The mean absolute error (MAE) between the FIM motor score estimated by the proposed method and that assessed by the nurse was calculated. The lowest resulting MAE was 8.82.