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

Area of Science:

• Gerontology
• Biomedical Engineering
• Public Health

Background:

• Accidental falls in older adults are a significant global public health issue, leading to increased morbidity and mortality.
• Effective fall risk assessment is crucial for timely intervention and prevention strategies.

Purpose of the Study:

• To evaluate the effectiveness of supervised machine learning models in classifying fall risk using posturography data.
• To contribute data-driven insights for improving predictive modeling in fall risk assessment.

Main Methods:

• Utilized a public dataset of 147 adults (18-85 years) including posturography, sociodemographic, and clinical information.
• Applied and compared various supervised machine learning models for fall risk classification.
• Focused on Random Forest and XGBoost algorithms for their performance evaluation.

Main Results:

• Random Forest and XGBoost models demonstrated strong classification performance for fall risk.
• Random Forest achieved superior accuracy (0.84 ± 0.04), F1 score (0.86 ± 0.03), and AUC (0.93 ± 0.03).

Conclusions:

• Machine learning, particularly Random Forest, offers a robust and accurate method for classifying fall risk based on posturography.
• These models can serve as effective tools for early detection and targeted prevention of falls in older adults, aiding clinical decision-making.