Hospitals and health care providers tend to get involved in exaggerated and fraudulent medical claims initiated by national insurance schemes. The present study applies data mining techniques to detect fraudulent or abusive reporting by healthcare providers using their invoices for diabetic outpatient services. This research is pursued in the context of Taiwan's National Health Insurance system. We compare the identification accuracy of three algorithms: logistic regression, neural network, and classification trees. While all three are quite accurate, the classification tree model performs the best with an overall correct identification rate of 99%. It is followed by the neural network (96%) and the logistic regression model (92%).
People rely on government-managed health insurance systems, private health insurance systems, or both to share the expensive healthcare costs. With such an intensive need for health insurances, however, health care service providers' fraudulent and abusive behavior has become a serious problem. In this research, we propose a data-mining framework that utilizes the concept of clinical pathways to facilitate automatic and systematic construction of an adaptable and extensible detection model. The proposed approaches have been evaluated objectively by a real-world data set gathered from the National Health Insurance (NHI) program in Taiwan. The empirical experiments show that our detection model is efficient and capable of identifying some fraudulent and abusive cases that are not detected by a manually constructed detection model.
In 1988, Congress passed the Clinical Laboratory Improvement Amendment (CLIA), thereby extending coverage of the Clinical Laboratory Improvement Act of 1967 to include quality standards for all laboratory-based testing. The CLIA was enacted to ensure the accuracy, reliability, and timeliness of patient test results, regardless of the location where the tests were performed. This article assessed trends in the enforcement policy of the CLIA through an examination of the Laboratory Registry, an annual publication of those individuals or entities that have had sanctions imposed on them by the Centers for Medicare and Medicaid Services. We reviewed the CLIA, including its oversight, regulations that were promulgated based on it, and its enforcement procedures. We obtained the Laboratory Registries for 1993-2001. Sanctions were categorized into groups per the enforcement regulations (42 C.F.R. section sign 493.2 2000). The data indicated an increasing use of more lenient sanctions from 1997 to 2001, and a gradual increase in fraudulent activity for that same period. One possible explanation for this finding is that implementation of compliance plans by participating clinical laboratories had a mitigating effect on enforcement policy. Compliance plan guidance from the OIG provides an opportunity for laboratory service providers to be proactive in their attempts to decrease errors, and thus improve accuracy and reliability by documenting laboratory policies, procedures, and objectives.
This paper identifies which types of patients and hospitals have abusive Medicare billings that are responsive to law enforcement. For a 20% random sample of elderly Medicare beneficiaries hospitalized from 1994 to 1998 with one or more of six illnesses that are prone to abuse, we obtain longitudinal claims data linked with social security death records, hospital characteristics, and state/year-level anti-fraud enforcement efforts. We show that increased enforcement leads certain types of types of patients and hospitals to have lower billings, without adverse consequences for patients' health outcomes.
Hospitals and health care providers tend to get involved in exaggerated and fraudulent medical claims initiated by national insurance schemes. The present study applies data mining techniques to detect fraudulent or abusive reporting by healthcare providers using their invoices for diabetic outpatient services. This research is pursued in the context of Taiwan's National Health Insurance system. We compare the identification accuracy of three algorithms: logistic regression, neural network, and classification trees. While all three are quite accurate, the classification tree model performs the best with an overall correct identification rate of 99%. It is followed by the neural network (96%) and the logistic regression model (92%).
