Healthcare Price Finder: A Web-Based Application for Estimating Healthcare Medical Cost by Modeling Medicare Data With Machine Learning

Abstract

The pricing intransparency in the healthcare system of the United States has been a long-standing problem. This thesis describes the Healthcare Price Finder project, a web-based solution that helps users to estimate how much the medical procedures cost. Using the Medicare Provider Utilization and Payment Dataset between 2012 and 2017, an XGBoost regression machine learning model was trained to predict the cost of medical procedures The model enables the Healthcare Price Finder to provide medical cost estimates even when past pricing data is not available. A Healthcare Price Finder web application was also developed using JavaScript framework and libraries including React.js and Node.js, and provides an easy-to-use interface for average users to get cost estimates for medical procedures and look up the past prices from the Medicare Provider and Payment dataset. The current version of the Healthcare Price Finder tool was developed with California data, and can be scaled up to provide medical procedure charge amount estimates and past price lookups for the entire United States.