Machine Learning Fights FWA in Healthcare
Aug 14, 2018
There are not many of us who would ignore fraudulent charges on a credit card bill, wouldn’t consider eliminating the wasteful gym membership that never gets used, or would not try to alter behavior to not abuse indulgences, whether that is shopping, golfing, or that daily cup of specialty coffee. Why should it be any different for healthcare organizations? State health and human services agencies, insurers, and managed care organizations can and should take advantage of advances in technology, including program integrity solutions that apply advanced analytic techniques, such as machine learning algorithms (MLAs) to identify fraud, waste, and abuse that is going undetected today.
To achieve long-term success, program integrity solutions must be built upon a flexible and scalable framework that can adapt and grow to accommodate evolving needs. The solution needs to be deployed rapidly to demonstrate a return on investment. If a program integrity solution is slow to deploy, that solution is not going to be flexible enough to stay ahead of sophisticated fraudsters – and they are sophisticated. Much fraud is being committed by networks of individuals and criminal syndicates, whose day job is defrauding the government.
Today’s fraud cannot be stopped with business rules and traditional methods alone. Advanced analytic techniques must be used to identify fraudulent claims and encounters and should be used to stay ahead of malicious attacks by detecting and responding to known, unknown, and evolving threats. The solution must go beyond simple business rules to include methodologies across the entire analytic continuum. “Out of the box” pre-constructed algorithms are a necessary component to any fraud detection system. These algorithms are in place to deliver high-priority targeted leads. These algorithms should utilize advanced analytic techniques, such as unsupervised cluster models, neural networks, and supervised logistic regressions to identify and uncover hidden patterns of sophisticated fraud.
While a sophisticated modeling system is a powerful tool to combat the growing wave of fraud, waste, and abuse, complex models by themselves are not sufficient for long-term protection. Fraudsters are well-funded and incentivized to evolve their techniques. Even the most comprehensive modeling system will become less relevant over time. To keep pace with fraudsters, MLAs must be used, they inherently make the system more accurate. MLAS allow a system to adapt to changing trends and schemes and improve its accuracy and performance. Using feedback and self-learning capabilities these solutions can detect the newest schemes. Combining a robust modeling system with MLAs will provide a comprehensive long-term solution.
Fraud, waste, and abuse impact all organizations; public and private, large and small. As long as there are opportunities to cheat and take unfair advantage of a situation or program, there will be bad actors devoted to the craft. Public benefit programs such as Medicare and Medicaid are easy targets for these predators. Fraud, waste, and abuse can be difficult to identify and can place an undue burden on agencies, investigators, and law-abiding constituents. Innovative analytics-based approach, as described above, help states combat these problems in a few keys ways. A modern program integrity solution should:
- Identify fraud before the claim is paid
- Be invisible to providers and reduce the burden placed on legitimate providers
- Identify schemes and correlate complex events to stop fraud
- Apply machine learning algorithms that update continuously to detect the latest schemes
- Provide intuitive reports, making it easy to visualize data and make results understandable and actionable
- Analyze every claim or encounter
If an existing solution does not have these characteristics it is time to consider a modern, intelligent program integrity solution. After all, proposed changes to the ACA and Medicaid eligibility will have a drastic effect on a state’s ability to keep or pay for Medicaid and Medicare expansion. One way to make an impact is to make the system as efficient, innovative, and responsive as possible.
Category: Business Intelligence, Government, Predictive Analytics, Blog, Enterprise Technology, Healthcare, Data Governance, Business