Enhancing Health Equity by Predicting Missed Appointments in Health Care: Machine Learning Study.

Area of Science:

• Health Informatics
• Machine Learning in Healthcare
• Predictive Analytics

Background:

• Patient no-shows (Did Not Show/Attend/Fail to Attend) waste healthcare resources and negatively impact patient health.
• Effective prediction of patient no-shows is crucial for optimizing outpatient appointment scheduling and resource allocation.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting the likelihood of patients missing hospital outpatient appointments.
• To assess the performance of logistic regression, random forest, and Extreme Gradient Boosting (XGBoost) models.

Main Methods:

• Utilized 5 years of outpatient records (1,080,566 visits) from the MidCentral District Health Board (MDHB).
• Developed and compared three machine learning models: logistic regression, random forest, and XGBoost.
• Employed 10-fold cross-validation and hyperparameter tuning for model optimization and evaluation using accuracy, sensitivity, specificity, and AUROC.

Main Results:

• The XGBoost model demonstrated superior performance with an AUC of 0.92, sensitivity of 0.83, and specificity of 0.85.
• Key predictors for no-shows included patient's DNS history, age, ethnicity, and appointment lead time.
• Machine learning models trained on extensive datasets can effectively predict patient no-shows.

Conclusions:

• This study presents a novel application of machine learning for Did Not Show (DNS) management in New Zealand's healthcare system.
• The developed models serve as a proof of concept for benchmarking DNS predictions within the MDHB and other health boards.
• Further qualitative research is recommended to understand root causes of DNS and develop targeted interventions for improved resource utilization and health equity.