A Machine Learning Approach to Predict Same-Day Discharge after Angiography Procedures

Abstract

Discharging patients from the hospital to care at home on the same day has been demonstrated to be as safe as overnight stays. The correct selection of patients who may undergo same-day discharge (SDD) is a significant challenge due to the large volume of variables that must be considered. This paper proposes a machine-learning approach to predict SDD after angiography procedures. Data from 3227 patients scheduled for angiography procedures, including information about geographical location, demographic characteristics, pre-existing diseases, and clinical procedure type, were used to train and test machine learning algorithms, such as logistic regression (LR), K-Nearest Neighbor, Naive Bayes, Multilayer Perceptron, and Support Vector Machine (SVM). The target was to predict patients with or without same-day discharge after they received an angiography procedure. Performance metrics, such as accuracy, precision, recall, F1-score, learning time, and inference time, were computed for each algorithm. The algorithms with the best performance were LR and SVM, with an F1-score of 0.800 and 0.806, respectively. LR had better behavior than SVM concerning learning and inference time.