By Matthew Hawley, Executive Vice President of Payment Integrity, Cotiviti
LinkedIn: Matthew Hawley
LinkedIn: Cotiviti
Artificial intelligence (AI) has been touted as a “solution” to modernize many payers’ processes and functions, including their traditional payment integrity programs. Yet seeing AI as the answer to health plans’ administrative burdens is overly simplistic. Instead, payers would be better served viewing AI as a tool that they should use strategically to reap greater value from their payment integrity programs and improve the member and provider experience.
Optimized, effective payment integrity programs are essential for payers because they shield plans not only from fraud, waste, and abuse (FWA) but also from inappropriate payments caused by inefficiencies and inaccuracies during their prepay and postpay processes. By responsibly integrating AI into their payment integrity efforts, plans can attain business results that matter, including realizing up to 11% in medical cost savings. Such savings is critical to bend the cost curve as specialty drug costs, demand for behavioral health services, and other factors are expected to drive an 8% year-over-year medical cost trend in the group markets — and just slightly less in the individual market — in 2025.
Goals of using AI in payment integrity programs
AI can be a powerful tool for plans to use their limited resources more effectively while increasing claim payment accuracy. Using AI, plans can:
Improve prepay claim review: Payment policies are always evolving, creating opportunities for missed updates. Failing to update even a single policy could cost a health plan millions of dollars in just months. Leveraging tools like web crawlers combined with generative AI (GenAI) can enable plans to systematically gather and summarize large amounts of external information and help them identify policy modifications to improve payment accuracy and reduce inappropriate payments.
Harness advanced analytics to optimize diagnosis-related group (DRG) clinical review and improve provider satisfaction: Using machine learning, a plan can more accurately predict which claims are most likely to be over-coded, reducing provider abrasion associated with medical records requests and enabling nurse reviewers and clinical coders to focus on the claims and records that matter most. With natural language processing (NLP), plans can also reduce the time that nurse reviewers spend searching through lengthy medical records, easing their administrative burden. Working with a partner that can leverage data across the market further helps plans focus their attention on the most relevant claims and records, offering greater accuracy than if they only used their own data. After one year of using AI-enabled clinical review, one large regional plan grew its overpayment recoveries by nearly six times the previous amount.
Employ task-focused, narrow AI to develop better algorithms to detect fraud schemes: Fraud often goes undiscovered in the vast amount of health plan data. But task-focused, “narrow” AI can identify suspicious billing patterns and block inappropriate claims from being paid within the parameters of plans’ contractual prompt-pay requirements. Specifically, plans can use AI-driven predictive models that analyze vast amounts of data for billing patterns that fall outside of the norm, allowing them to focus their investigative efforts. Payers can also employ GenAI — which is poised to become a standard practice for businesses of all types by 2026 — to summarize medical records to streamline fraud detection.
Utilize large language models (LLMs) and analytics to improve coding accuracy and meet regulatory needs: With LLMs capable of natural language processing (NLP) tasks, AI tools can quickly capture a wide breadth of medical record data so it is ready for human reviewers, enabling them to be more efficient and effective in their roles. Meanwhile, advanced analytical models can also add value: They boost coding accuracy and optimize risk-adjustable revenues by identifying which members have the highest likelihood of undocumented conditions in their medical records, enabling plans to capture the full nature of anticipated care required.
Potential pitfalls for payers when using AI
Despite the great potential of AI, the technology has some limitations and drawbacks. Harnessing AI to bring greater effectiveness and efficiency to payment integrity programs requires resources that many payers don’t have. Beyond robust computing power, these resources include data scientists who can understand, build, and deploy AI tools, train models and establish proper governance over usage and data. AI can also be difficult to integrate with payers’ existing systems and entrenched workflows. Finally, the algorithms and data models that drive AI can be biased, which is why they should not be used to replace human decisions, especially clinical decisions.
With these concerns in mind, some health plan leaders may be tempted to take a “wait and see” approach to implementing AI. Yet given how quickly the technology is transforming healthcare, payers that fail to deploy this technology to support payment integrity risk falling behind their competitors. By working with an external partner that can deploy AI in a scalable way and aggregate data from multiple payers, health plans can harness the technology significantly more effectively than implementing it on their own.
Ensuring responsible use of AI in payment integrity
To overcome the technology’s potential pitfalls, plans must promote proper governance of AI. Payers should establish their own AI governance committee to ensure their organizations utilize AI as a tool, not as a decision-maker. They should also query their vendors to confirm that all their partners are following well-established standards for privacy and security, and are committed to using AI responsibly.
By embracing judicious use of AI within their payment integrity programs, health plans can rein in inappropriate medical costs and stay competitive. They can also elevate the capabilities of their payment integrity programs and achieve greater return from their efforts, all while helping to ensure the member and provider experience is not negatively impacted.