Related Experiment Video
Updated: Jun 18, 2026
08:05
Design and Analysis for Fall Detection System Simplification
Published on: April 6, 2020
Admission clinical and functional assessments for predicting inpatient fall risk using machine learning models
Ya-Wen Lee1,2, Ying-Lin Hsu3, Shu-Mei Lai4
1Department of Administration, Lutung Christian Hospital of Changhua Christian Medical Foundation, Changhua, Taiwan.
Digital Health
|June 17, 2026
Summary
Machine learning models can identify patients at high risk for falls. While improving recall, data balancing techniques like SMOTENC present a precision-recall trade-off, requiring careful interpretation of results.
Area of Science:
- Healthcare Informatics
- Patient Safety Research
- Machine Learning in Medicine
Background:
- Falls pose a significant patient safety risk in hospital settings.
- Early identification of high-risk patients is crucial for fall prevention strategies.
Purpose of the Study:
- To develop and evaluate machine learning models for early identification of high-risk patients prone to falls.
- To assess the impact of clinical factors and data balancing techniques on model performance.
Main Methods:
- Retrospective analysis of 483,174 adult inpatient records.
- Application of multiple machine learning algorithms including Logistic Regression, SVM, Random Forest, XGBoost, and MLP.
- Evaluation of models with and without SMOTENC using five-fold cross-validation, focusing on AUC-ROC, recall, and F1-score.
Main Results:
- Key predictors for falls included history of falls, medication use, dizziness, sensory impairment, and reduced functional status.
- Models without SMOTENC showed moderate AUC-ROC (0.62-0.72) but very low recall for fall events.
- SMOTENC improved recall to 50-60% with stable AUC-ROC (0.67-0.70), though it impacted precision.
Conclusions:
- Machine learning models effectively identify fall risk factors, demonstrating moderate discrimination (AUC-ROC).
- Severe class imbalance in fall data significantly hinders the detection of fall events.
- SMOTENC enhances recall but necessitates careful consideration of the precision-recall trade-off and interpretation of metrics in imbalanced clinical datasets.
