Development and implementation of explainable AI-based machine learning models for predicting hospital stay and treatment costs in cardiovascular patients

Comprehensive Summary

The aim of this study was to create a machine learning model that can effectively predict length of hospital stay and treatment costs for cardiovascular patients who have gone through heart procedures, including angioplasty and coronary artery bypass surgery. Using data from 7685 patients in a hospital in Tehran who went through a cardiovascular procedure, several machine learning models were developed which predicted length of hospital stay, out of pocket cost for patient, and total cost. The model XGBoost proved to be the most accurate predictor out of all the models generated. An analysis done on this model showed that factors of intervention type, age, number of medical conditions present in patients had the largest effects. The model developed and implemented for clinical use, will be helpful in hospital resource planning. The model was created to be a website where healthcare workers can implement the patients clinical and demographic information and the predicted length of hospital stay and cost will be generated.

Outcomes and Implications

The implementation of this model in hospitals will allow hospitals to have a better idea of inpatient length of stay for cardiovascular procedures, contributing to a more efficient resource planning. However, the authors note that the model should be tested and evaluated before being used in other locations in order for the program to be accurate in terms of costs.