Fall incidence prediction system for elderly people based on IoT and classification techniques

Abstract

Health monitoring systems based on the internet of things (IoT) improve patient well-being and reduce mortality risks. Machine learning techniques are most helpful in early fall prediction and detection. In this paper, fall prediction analysis and decision-making are done with existing benchmark clinical records. Classification techniques are incorporated to track the consistency and precision of data acquired by the IoT-based remote health monitoring for elderly people, especially those who are living alone. This work undertakes two approaches to early predicting a patient’s acute illness. The first approach has analyzed the existing benchmark patient activity data with different features. This approach builds the classification model for fall incidence with the help of machine learning models. In second approach, we collect real-time sensor data such as blood pressure and heart rate from IoT sensor gadgets which are transmitted to the prediction model for early prediction. Experimental results prove that the random forest (RF) classifiers and XGBoost provides the maximum accuracy.