Machine Learning for Prediction of Clinical Appointment No-Shows

Abstract

A no-show occurs when patient misses his appointment for visiting doctor in an outpatient clinic. No-shows result in inefficiencies in scheduling, capacity wastage and discontinuity in care. The study aims to develop and compare different models for predicting appointment no-shows in a hospital. The no-show estimation was made using five algorithms including Logistic Regression, Decision Tree Classifier, Random Forest, Linear Support Vector Machine and Gradient Boosting. The performance of each model is measured in terms of accuracy, specificity, precision, recall and F measure. The receiver operating characteristic curve and the precision-recall curve are obtained as further performance indicators. The result shows gradient boosting is more evident in giving consistent performance. The categorical variables used for prediction are gender, mapped age, appointment type, previous no-shows, number of previous no-shows, appointment weekday, waiting interval days, scholarship, hypertension, diabetes, alcoholism, handicap and SMS received.