FallDS-IoT: A Fall Detection System for Elderly Healthcare Based on IoT Data Analytics

Abstract

Fall represents a major health risk for the elderly people. If the situation is not alerted in time then this leads to loss of life or impairment in the elderly, which reduces the quality of life. In this paper, we solve this problem by introducing a Fall Detection System based on Internet of Things (FallDS-IoT) by designing a wearable system to detect the falls of elderly people. We use Accelerometer and Gyroscope sensors to get accurate results of fall detection. We classify the daily activities of elderly people into sleeping, sitting, walking and falling. We use two well-known machine learning algorithms, namely K-Nearest Neighbors (K-NN) algorithm and decision tree to deal with the above work. The resultant accuracies for our generated dataset were 98.75% and 90.59%, respectively. Therefore, we were able to conclude that K-NN gives more accuracy in detecting falls and this method is used for classification. whenever a fall happens, a message informing about the fall will be sent to a registered phone number through a Python module.