Predicting patient’s hospital charges using machine learning

Abstract

As the health care system moves toward value-based care, CMS (Clinical Management System) has designed a number of programs to improve the quality of patient care. One of these programs is called the Hospital Patient Admission Cost Analysis Program, which helps the patient and the hospital to diagnose the disease and estimate the cost of hospitalization. According to the World Health Organization (WHO), personal and medical costs have skyrocketed faster than the global economy. Major attributes which cause increase in expenditure include smoking, ageing and increased BMI (Body Mass Index). In this study, we aim to find a correlation between medical costs and various items using insurance data of different people with characteristics such as smoking, age, number of children, region and BMI. This study can also be used to demonstrate different models of regression that can be used to forecast insurance costs. Machine learning can significantly reduce human efforts because machine learning models can compute cost calculations in short time, for which human being takes long time to perform the same task.