Machine learning for predictive analytics

Advances in the field of machine learning are enabling the development of new predictive models in the field of cardiovascular medicine. In this chapter, we cover how predictive machine learning models are developed, their recent use-cases in cardiovascular medicine, and limitations for their expanded use in healthcare. Predictive modeling has long been used in medicine, but recently, machine learning methods offer the ability to learn patterns in large and heterogeneous health data for improved predictive modeling. In addition to describing the general methods by which machine learning models are developed, we share their recent applications in electrophysiology, interventional cardiology, heart failure, and preventive cardiovascular care. We conclude with a discussion about issues regarding data quality, generalizability, bias, and interpretability, and how these influence model development and clinical integration.