Medical Reimbursement Prediction Using Artificial Intelligence

US Healthcare has experienced a steady increase in hospital expenditure due to an increase in overall healthcare costs to provide quality care to patients. For healthcare providers to continue to operate efficiently, it is therefore important to strengthen their operational workflows to monitor the revenue collected in return for the care provided. Healthcare providers spend huge amounts on the treatment of patients based on the treatment complexity and resource intensiveness. However, getting reimbursed for the amount spent on patients often take months and may end up in partial or no payment from the insurance companies (also known as payers). Healthcare providers also allocate resources to Account Receivables (AR) department for follow-up on the long queue of claim reimbursements from insurance companies which is a time-consuming process. This follow-up queue is oftentimes prioritized based on the billed amount. In most cases, the reimbursement is usually far lower than the billed amount which makes the follow-up process inefficient. Hence, a predictive reimbursement solution can prevent healthcare providers from facing unavoidable financial losses. In this study, we propose a novel reimbursement prediction tool that utilizes a boosting-based Machine Learning (ML) framework and explainable Artificial Intelligence (AI) to fetch the important features leading to lower reimbursement. The reimbursement prediction would aid administrative staff to make appropriate edits by investigating the important features and reasons behind lower reimbursement of the claim lines before submitting them to the payers. Reimbursement prediction could also be useful in prioritizing the claims having higher reimbursement for accounts receivable follow-up also keeping into consideration the response time of the payers. The reimbursement model showed good performance with an Adjusted R-Squared value of 0.91 and 0.8 for professional and institutional claims, respectively. Based on our projections and feedback from the Subject Matter Experts (SMEs), the model has the potential to improve the reimbursement amount for healthcare providers, thereby helping them operate efficiently.