If applied correctly, AI could save the healthcare system billions of dollars per year.
That optimistic claim comes from Doug Williams, COO at HMS, provider of cost reduction analytics and engagement solutions. In a recent column at HITConsultant, Williams points out that estimates for healthcare “waste,” by which he primarily means unnecessary or excessive medical testing and treatment, run up to $200 billion per year.
But new AI promises to go a long way toward fixing that.
The problem, he says, is not spikes in the use of specific tests so much as “more subtle over-utilization that can be from provider behaviors or from individual patients.” Waste of a subtle nature often goes undetected because of how payment integrity programs are currently run.
“Today,” he says, “most payment integrity programs include some automation but also a significant amount of manual work to verify the payment, identify potential fraud or misuse, and acknowledge patterns or anomalies. The complexity and variety of claims apart from medical, such as imaging, pharmacy and behavioral, increases the challenge.”
AI, however, may be precisely the necessary prescription to tackle that complexity, if not always an outright cure.
That’s because “AI can detect patterns and anomalies in seconds instead of days or weeks to help plans reduce current and potential future waste, fraud, and abuse of the system. By effectively utilizing AI, health plans can decrease unnecessary spending, but also educate providers so all parties can collaborate on high-quality, evidence-based care alternatives.”
More specifically, Williams says, “AI technology that supports pre-pay reviewer staff . . . can rapidly identify a potentially improper payment before a claim is paid due to the technology’s pattern recognition that learns the more it is utilized. By reviewing claims this way, health plans can avoid payment errors related to Diagnosis Related Group (DRG) coding and validation, readmissions, level of care, place of service, and more, which saves time and burden trying to recoup payment.”
Moreover, he argues, when combined with natural language processing (NLP) and other technologies, AI can improve efficiencies and reduce the number of false positives, reducing unnecessary work for payers and providers.
But AI’s biggest contribution on the overutilization front may be its capacity “to help providers detect it before the service is performed or claim is submitted. During the appeal or overpayment recovery process, health plans can educate providers about the errors, non-compliance or why the test or service was rejected or overpaid using data pulled from analytics as evidence.”
In short, says Williams, when a test or service is ordered, providers with access to AI can “analyze the vast amount of data available” to determine if the care will be considered medically necessary while also noting the factors that may cause a claim denial.
And that kind of efficiency, he says, can save healthcare stakeholders, and society at large, a lot of money.