A Machine Learning-Based Fall Risk Assessment Model for Inpatients.

Area of Science:

• Medical Informatics
• Machine Learning in Healthcare
• Patient Safety Research

Background:

• Inpatient falls are a significant cause of extended hospitalization and increased medical costs.
• Accurate fall prediction models are crucial for reducing patient injury and healthcare expenses.
• Existing fall risk assessments may not be optimally integrated into clinical workflows.

Purpose of the Study:

• To develop and evaluate data-driven inpatient fall prediction models using electronic medical records.
• To identify the most effective machine learning algorithms for predicting patient falls at different hospitalization stages.
• To provide a foundation for timely and individualized fall prevention interventions.

Main Methods:

• Utilized data mining techniques on electronic medical records from 15 inpatient wards.
• Developed classification models using supervised machine learning algorithms and ensemble techniques.
• Assessed prediction performance at four distinct time points during patient hospitalization.

Main Results:

• Bagging+RF classifiers demonstrated optimal prediction performance across all four assessment time points.
• The developed models effectively predicted inpatient falls based on electronic health record data.
• Machine learning models offer a promising approach for enhancing patient fall risk stratification.

Conclusions:

• The Bagging+RF classifier is a highly effective tool for inpatient fall prediction.
• Comprehensive fall risk assessment and individualized interventions are essential for reducing fall rates.
• Integrating predictive models into nursing workflows can improve patient safety and reduce healthcare costs.