Comparison of Fall Risk Prediction of Patients using Tree Algorithms implemented on a wearable belt

Falls especially in older people can cause a potential serious damage. It is seen as the primary reason for further accidents, disability, and death in the target population. Henoe, this paper studies the comparison of two, fall risk prediction methods which can be used to build a device to help geriatric patients or hospital ward members to monitor their fall risk. The algorithms studied here are derivations of the Decision Tree algorithm i.e., Classification and Regression Trees algorithm and Random Forest algorithm, and the comparison between sensitivity, specificity and accuracy is carried out. The data is collected from volunteers in Bangalore, India and the steps in pre-processing, feature extraction, filtering and labelling are carried out. Finally, the model using both the algorithms are defined and results are compared. Also, the reason for the Random Forest algorithm to do better has been explored.