Artificial intelligence (AI) has often been touted as the new frontier in healthcare: a solution to longstanding problems like administrative overload and burnout, helping providers diagnose diseases, document visits, and catch fraud.
And in many respects, that’s probably true. But recent reports have been questioning whether, despite all that, is the technology actually contributing to the healthcare affordability problem? And they’re all suggesting that the answer to that is: absolutely, yes.
PwC just predicted commercial healthcare cost trends to rise by 9 percent in 2027, the highest jump in almost two decades. When asked about why that may be, 70 percent of health plans surveyed by the firm identified AI-enabled tools as a top three cost inflator, as providers are utilizing the technology in documentation and coding tools that both save them time and allow them to record with greater specificity on claims.
This has led to more prior authorization submissions and more overall billing activity. And, according to PwC, to higher paid amounts per each claim.
Also on the provider side, HealthAffairs released an in-depth article examining the same findings of PwC. AI tools are simply allowing for earlier diagnoses and better treatment plans, meaning there’s simply more care happening overall. Diagnostic timelines that used to take days have been cut. Side effects of pharmaceuticals can be predicted without having to perform lengthy trials.
This activity expands the size of populations receiving treatment, including those who would have been diagnosed much later, or those whose conditions would have simply gone unmanaged. Obviously, this leads to increased overall costs through sheer utilization. As HealthAffairs put it: “The value is real. So is the inflation.”
But the flipside of that is that AI could be contributing to overdiagnosis and overtreatment. AI systems are admittedly very good at detecting more subtle patterns in imaging, labs, and patient records. This is often a good thing, but it can also hone in on patterns that never would have caused harm, meaning there are sometimes tests, treatments, and follow-ups that never needed to happen – all of which carry a cost.
This all, of course, leads to an interesting discussion on the way our healthcare’s current dominant payment model of fee-for-service will interact with AI. Without any reform whatsoever, it’s likely that AI’s ability to uncover and specify billing will simply lead to more claims – and more costs.
Payers are also utilizing AI technology in payment integrity practices that interact with the AI-enabled claims they are receiving from providers. The increase in care described previously leads to an overall increase in the number of interactions between providers and payers.
With AI-technology aiding both, it has been suggested that there’s a new pattern of back-and-forth going on over the same claims, where the money saved on clinical dollars is now simply switching to administrative ones.
We must ask ourselves about the balance between all the good AI technology can bring with the bad that could arise from its impact to costs. Certainly, AI has arrived at a time when healthcare spending was already sky-high; latest reports indicate it will soon reach $9 trillion annually by 2035 without intervention. AI can’t be blamed for that.
But it is optimizing what both providers and payers do – both clinically and administratively – and that’s leading to more care and more claims, processed more quickly. Without reform in our payment system, this could limit the transformative impact the technology could have in our industry.
