Improving Hospital Outpatient Clinics Appointment Schedules by Prediction Models

Area of Science:

• * Health Informatics
• * Operations Research
• * Machine Learning

Background:

• * Patient no-shows and inefficient appointment scheduling decrease clinical efficiency and service quality.
• * Optimizing appointment scheduling is crucial for hospital outpatient clinics.
• * Existing models lack comprehensive features for accurate prediction.

Purpose of the Study:

• * To develop supervised machine learning models for predicting patient no-shows and appointment length (LOA).
• * To enhance the efficiency and quality of service in hospital outpatient clinics through improved scheduling.
• * To identify key features influencing patient no-shows and LOA.

Main Methods:

• * Retrospective study utilizing over 100,000 patient appointment records.
• * Development of supervised machine learning models using patient, physician, and appointment characteristics.
• * Feature engineering combining novel and established features for enhanced prediction accuracy.

Main Results:

• * Developed models achieved high performance: LOA model (6.92 MAE) and no-show model (92.1% F-score).
• * Novel features significantly improved model performance compared to previous research.
• * Integration with appointment systems resulted in an 80% reduction in patient waiting time and a 33% decrease in physician idle time.

Conclusions:

• * Machine learning models can effectively predict patient no-shows and LOA.
• * Novel features are critical for improving prediction accuracy in healthcare scheduling.
• * Optimized scheduling through predictive modeling substantially enhances operational efficiency and patient experience.