Towards Automated Fall Risk Classification in Older Adults Using Supervised Machine Learning

The rising incidence of accidental falls in older adults, along with the associated health impacts such as morbidity and mortality, has turned this into a worldwide public health concern. This article presents the preliminary results of a study aimed at contributing valuable information to ongoing efforts that use supervised machine learning to classify fall risk based on posturography data. A public dataset containing posturography, sociodemographic, and clinical data from 147 adults (18-85 years) was used. Various machine learning models were used to classify the risk of falls. Random Forest and XGBoost models exhibited good classification performance, with Random Forest consistently outperforming in accuracy (0.84 ± 0.04), F1 score (0.86 ± 0.03), and area under the ROC curve (0.93 ± 0.03). Our study contributes to the growing body of evidence supporting the adoption of machine learning methods for predictive modeling to assess the risk of falls, offering an effective tool for early detection and underscoring their potential to revolutionize fall risk assessment in public health.Clinical relevance- This study demonstrates that machine learning models, particularly Random Forest and XGBoost, can effectively classify fall risk based on posturography data. By providing accurate and objective fall risk assessments, these models could aid clinicians in early detection and targeted prevention strategies, ultimately reducing fall-related injuries among older adults.