This study compared the predictive ability of four machine learning models in identifying the risk of health services use in a community sample of persons with heart failure. Results suggest the use of machine learning models with a large-scale administrative dataset had little benefit in this sample.
