Using machine-learning methods to support health-care professionals in making admission decisions

Area of Science:

• Health Informatics
• Machine Learning in Healthcare
• Predictive Analytics

Background:

• Tertiary hospitals experience long waiting lines, necessitating advance screening for patient hospitalization.
• Current patient admission screening relies on healthcare professionals ranking patients using registration information.

Purpose of the Study:

• To develop a machine learning (ML) approach for patient screening.
• To create screening rules using historical data and expert experience for automated patient prioritization.

Main Methods:

• Employed five ML methods: logistic regression (LR), random forest (RF), gradient-boosting decision tree (GBDT), extreme gradient boosting (XGBoost), and an ensemble model.
• These models were used to sequence and predict elective patient admissions.

Main Results:

• All five ML models demonstrated strong prioritization performance with high predictive values.
• XGBoost achieved the highest predictive performance, indicated by the area under the receiver operating characteristic curve (AUC) of 0.901.

Conclusions:

• Machine learning techniques offer significant value in automating the patient screening process.
• The developed ML model aids healthcare professionals in automatically evaluating less complex cases by identifying critical admission factors.