Model For Assessing Fraudulent Medical Claims: An Application Of Machine Learning Algorithms In Health Care

Abstract

The insurance industry is experiencing challenges as claim verification has remained to be a manual process due to its nature that requires human observation. Health insurance cost has endlessly been increasing with time. In Kenya, the majority of citizens cannot afford healthcare services as most are not insured. The insurance cost is so high with low per capita income. The high cost is due to administrative cost that results from inefficient operational processes that are prone to errors and fraudulent claims. We propose a machine learning model that can be extended to a system which will create claim assessment automation that will greatly reduce the administrative resources required to process medical claims. The model will learn from previous data patterns and predict the correctness or review of the claim thus reducing errors that are prone to manual processes. Using machine learning techniques for classification like Logistic Regression and SVM we are able to classify a correct claim or a claim that will be called for review by the auditors. We will show the significance of our model and how it achieves better precision and accuracy than the manual process.