Creating the Right Cost Containment Toolkit: A Common Understanding and Collaboration across Technologies, Automation, and Domain Experts
By Saurav Subedi, Director of Data Science, Codoxo
With the rapid movement toward data-driven decision making, healthcare payers are looking to many tools, technologies, people and processes to inform best practices and opportunities for optimization. While building the perfect “toolkit”, healthcare payers must strike the right balance between their platforms, digital solutions, data and domain experts. They must all work together, complement each other and strive toward common goals.
One such example lives within the Special Investigations Unit (SIU) of a health plan. This unit is responsible for identifying and investigating insurance fraud, waste, and abuse (FWA). Its investigators scrutinize the claims and billing process to uncover the signs of overbilling, false coding, and other non-compliant or wasteful practices. Largely, this unit supports broader cost containment efforts and helps proactively mitigate risk. The financial losses due to FWA are estimated at up to 10% of total annual healthcare expenditures – or more than $300 billion each year – according to the National Health Care Anti-Fraud Association (NHCAA). The role of an SIU unit is critical to helping to combat rising healthcare costs. There are well established processes and practices built around traditional SIU toolkits including Excel spreadsheets and/or home-grown solutions paired with deep domain healthcare investigators. In recent years, we are seeing the adoption of digital solutions to automate manual processes and create efficiencies across operations. More importantly, application of artificial intelligence (AI) and machine learning in healthcare cost containment efforts has gained significant momentum, primarily because of advancement in tools, technology, algorithms, and availability of skilled human resources. This has changed the nature of the ‘toolkit’ being used pan-industry, and consequently, we are seeing higher dollar impact and faster speed-to-market.
Technology & Automation vs. Humans
Many wonder the impact technology and automation will have on human-driven roles. A health plan’s SIU unit is not any different. Will AI, machine learning and the like replace humans? Will there be fewer healthcare investigators needed? It is our experience that when done well, health plans use such technologies as only one of their tools in the kit. Successful implementations typically rely on the tripod balance of state-of-the-art machine learning methods and big data architecture in close collaboration with the subject matter experts. There exists a symbiotic relationship between the experts and these algorithms. Machine learning workforce leverages the expertise of the SMEs to properly navigate the legal and legacy complexities of the healthcare payment integrity space, incorporate their thoughts and processes into feature engineering, and establish business reasonability checks and validations. The experts, on the other hand, use these innovations to complement their efforts and help them do their job better, more efficiently. Traditionally, vast amounts of data were handled by humans. Today, we have the ability to empower these experts with solutions that ease the burden of labor-intensive tasks and help implement better practices. What this equates to is data-driven decisions made faster.
For instance, for identifying a typical FWA case, traditional approaches would include navigating through the universe of different codes, including International Classification of Diseases (ICD) codes, Current Procedure Terminology (CPT) codes, and Healthcare Common Procedure Coding System (HCPCS) codes, individually or through a known combination pattern. This is naturally a time-consuming process and would be less likely to detect novel schemes. On the other hand, an AI-based approach could analyze such multidimensional utilization profiles efficiently and effectively to identify novel schemes and complement the traditional rules-driven architecture.
Investigators vs. Data Scientists
While each have their unique role, we see healthcare payers challenged with creating collaboration across their investigators and data science teams. Healthcare investigators are known to have deep roots and experience in discovering, investigating and optimizing processes around fraud, waste, abuse, and non-compliance. Data Scientists are experts at understanding data and, importantly, surfacing meaningful insights from the data. Pairing these two skillsets together can create unsurpassed ability to manage costs and shed light on business problems. Where one skillset leaves off, the other picks up. Both automation and data science can give healthcare SIU’s the edge in battling fraud, waste and abuse.
For instance, the use of telemedicine proliferated since the onset of COVID-19 pandemic. We started observing increases in the utilization of telehealth codes in relevant specialties such as, psychiatry and sleep medicine. At the same time, we also observed significant increase in telehealth practices in other non-traditional specialties from a telemedicine perspective. Since it is an unknown territory, our clients are interested to gain more insights and potential impact on cost containment. This provided an interesting opportunity of collaboration between the SMEs and AI. While AI could identify anomalies and potential FWA based on significant changes in patterns and peer comparison through big data analytics for SIUs to launch a targeted investigation, it required a guidance and business reasonability checks from the SMEs to assist in the design of the algorithms and refine the results.
As we continue to see major evolution happen across the healthcare spectrum, we are empowered to take hold of innovation, embrace change, and develop a mindset ready to collaborate, innovate and succeed.
