The US Centers for Medicare & Medicaid Services (CMS) will need to retain a ‘human in the loop’ as it looks to implement AI in its approval frameworks, according to an industry expert.
The CMS is rolling out an AI-led initiative to modernise how Medicare approves certain procedures before they are performed, an approval process called prior authorisation.
Go deeper with GlobalData
Discover B2B Marketing That Performs
Combine business intelligence and editorial excellence to reach engaged professionals across 36 leading media platforms.
CMS’s Wasteful and Inappropriate Service Reduction (WISeR) pilot is set to run for six years from 1 January 2026 across six US states, including Texas, Arizona, and Washington.
Given that waste in healthcare contributes to up to 25% of healthcare spending, WISeR’s focus will be on expediting and improving the review process for abuse and fraud-prone healthcare services, including steroid injections, spine surgeries, and skin and tissue substitutes.
Speaking with Medical Device Network, Dr Ahzam Afzal, CEO of value-based care company Puzzle Healthcare, highlighted that patients can experience lengthy waits for prior authorisations under Medicare’s existing approval model.
Afzal said: “With a model like WISeR, it’s likely that patients will get faster answers due to more consistent application of the criteria and less administrative back and forth for clinicians.
US Tariffs are shifting - will you react or anticipate?
Don’t let policy changes catch you off guard. Stay proactive with real-time data and expert analysis.
By GlobalData“If it’s done properly, this initiative will reduce delays and give clinicians faster yes or no answers across the board.”
While the benefits of the initiative appear well rationalised, Afzal foresees some potential downsides, such as moves by provider groups and technology companies to build out AI-centric prior authorisation solutions.
With this prospect, he highlights that there are concerns from patients and payors that WISeR could result in AI making determinations “across the board”.
Afzal continued: “While AI will help to streamline the prior authorisation process in many ways, it will be critical that coverage determinations remain clinician-led and patient-specific.
“Therefore, as the CMS’s pilot rolls out, it must ensure there remains a human in the loop alongside an implementation that appropriately captures the upside of AI’s application for prior authorisation.”
Afzal’s view is that as the WISeR pilot progresses, it will be important to have a named reviewer sign off for any denial or curtailment of care, to give patients and payors a good sense that there is a clinician involved in the process.
He commented: “I think one of the things that more groups, providers and others are going to be looking for is that level of transparency across the board.
“They’re going to want the models to ‘show their work’ and provide a plain language rationale that gives clinicians and patients the full reason for any authorisation’s denial and the data that informed it.”
By the end of the initiative’s pilot phase on 31 December 2031, Afzal’s hope is that an external audit of the AI model’s performance, potential biases, and security will be undertaken. He states that such measure should help tune the CMS’s model appropriately and iron out any potential decisioning defects to ensure it is fit for purpose.
Afzal concluded: “There’s a lot of efficiencies that AI can help build, but the CMS needs to ensure that automation doesn’t fully outrun the human oversight element.”
