Predicting Falls in Older Adults Aged 65 and up Based on Fall Risk Dataset

Document Type

Conference Presentation

Department

Health, Exercise, and Sport Sciences Department

Conference Title

SAI Intelligent Systems Conference

Location

Amsterdam, Netherlands

Date of Presentation

9-1-2021

DOI

10.1007/978-3-030-82199-9_41

Volume

3

First Page

602

Last Page

616

Abstract

Falls are one of the leading causes of death for adults aged 65 and up. Due to the increase of falls in older adults and the severity of the resulting injuries, medical costs spent on treating affected patients are high; therefore, it is imperative to derive models that can predict falls in older adults and identify contributing factors that can be used to determine preventive methods and reduce costs. In this study, a test suite of four supervised machine learning models was created to predict falls using a fall risk dataset consisting of 593 patients and 764 internal and external features. After the training and testing phases were completed, a ten-fold cross-validation was performed to confirm the results which demonstrated accuracies of 80% and above for making predictions using the testing dataset. The results demonstrated that the model with the highest accuracy is the one that uses the support vector machine. It predicted 87% of falls correctly and had an area under the curve of 0.86. Additionally, the most significant feature in the dataset was the total number of medications which was selected along with 29 other features that were primarily prescription medications. Based on these findings, we assert that patients who take prescription medications that may result in cognitive impairment are at high risk for falls.

Link to Full Text

Share

COinS