Enhancing Fall Prediction Models for the Elderly: Analyzing the Impact of Using Steady Gait Data from Motion Sensors

Falls are a significant cause of severe health problems among the elderly, often resulting in bedridden conditions. However, assessing the risk of falls through tests is impractical due to the associated human cost. In this paper, we construct a fall prediction model using data collected from 6-axis motion sensors attached to the subjects' feet. We specifically examine the impact of using data from steady walking on the models' prediction accuracy. The results showed that the model using only the steady walking part achieved higher accuracy.