Effect of a Real-Time Risk Score on 30-day Readmission Reduction in Singapore

Area of Science:

• Health Informatics
• Machine Learning in Healthcare
• Predictive Analytics

Background:

• Hospital readmissions pose a significant challenge to healthcare systems, impacting patient outcomes and resource utilization.
• Accurate prediction of readmission risk is crucial for implementing timely interventions and improving patient management.
• Electronic Medical Records (EMR) contain rich patient-specific data that can be leveraged for predictive modeling.

Purpose of the Study:

• To develop and implement a real-time risk score for predicting 30-day patient readmissions using EMR data in Singapore.
• To enable prospective identification of high-risk patients for targeted interventions.
• To integrate a predictive risk score into the EMR system for seamless clinical workflow.

Main Methods:

• Retrospective extraction of EMR data from 25,472 patients discharged from a medicine department.
• Development and internal validation of machine-learning models to estimate 30-day readmission probability.
• Real-time risk score generation within the EMR system, flagging high-risk patients to care providers.

Main Results:

• Machine-learning models demonstrated good predictive performance with an area under the receiver operating characteristic curve ranging from 0.77 to 0.81.
• Proactive identification and management of high-risk patients led to a significant reduction in the 30-day readmission rate.
• The 30-day readmission rate decreased from 11.7% in 2017 to 10.1% in 2019 (p < 0.01) after risk adjustment.

Conclusions:

• Machine-learning models integrated into EMR systems can provide real-time readmission risk predictions.
• This predictive capability enhances clinical decision-making and facilitates proactive patient care.
• Deploying such models can lead to improved patient outcomes and more efficient healthcare delivery.