Optimizing Healthcare Billing Processes Through Machine Learning Using Diagnosis-Related Groups

This paper explores the integration of Machine Learning (ML) techniques with Diagnosis-Related Groups (DRGs) in healthcare to optimize the resource allocation of DRG coders to improve billing processes. By using ML algorithms and DRGs, healthcare providers can categorize patients based on clinical characteristics. The study focuses on classifying hospital bills according to their complexity levels, allowing domain experts to review bills that align with their expertise. Two approaches are examined: an Automated Machine Learning (AutoML) tool called TPOT and the XGBoost algorithm. Experimental results demonstrate the potential of these methods to slightly improve classification accuracy. With XGBoost we obtained an average accuracy of 0.55 over all classes and with TPOT an accuracy of 0.56.