Applying artificial intelligence to predict falls for inpatient.

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Patient Safety Research

Background:

• Inpatient falls are a significant cause of patient injury and increased healthcare costs.
• Effective prediction of fall risk is crucial for implementing targeted prevention strategies.

Purpose of the Study:

• To identify key risk factors for inpatient falls using machine learning.
• To develop a predictive model for accurately assessing fall risk in hospitalized patients.

Main Methods:

• Retrospective analysis of 53,122 electronic health records (EHR) from 2015-2019.
• Application of eight artificial intelligence models, including gradient boosted trees (XGBoost), using RapidMiner Studio.
• Evaluation of model performance using sensitivity, specificity, and area under the ROC curve (AUC) with 5-fold cross-validation.

Main Results:

• The XGBoost model demonstrated superior performance, achieving a cross-validated accuracy of 95.11%, AUC of 0.990, and F1 score of 95.1%.
• The study identified critical factors contributing to inpatient fall risk through machine learning analysis.

Conclusions:

• Machine learning, specifically the XGBoost model, offers a highly predictive approach for detecting patients at risk of falling.
• Improved fall risk detection can enhance patient care quality and decrease nursing staff workload associated with fall assessments.