Optimizing prior authorization with AI-driven efficiency

The AI-powered arms race in prior authorization

As healthcare payers and providers seek to optimize their workflows, AI has emerged as the critical technology driving innovation in prior authorization.

"AI is the lever that both sides will use in the prior authorization arms race," Friese insisted. “It’s the perfect use case for this type of technology.” He envisions a near future where AI is fully integrated into prior authorization processes across the industry, with non-adopters at risk of being left behind.

The Centers for Medicare & Medicaid Services Final Rule on prior authorization, set to take effect next year, underscores the urgency for payers and providers to modernize their workflows. As regulatory changes, financial considerations and public perception converge, AI adoption is rapidly becoming a necessity rather than an option.

Recent studies emphasize the impact of prior authorization delays on patient care. For instance, a survey by the American Medical Association (AMA) found that over 90% of physicians reported that prior authorization delays access to necessary care, and nearly 24% reported that these delays have led to serious adverse events, including hospitalization, permanent impairment or death. Such statistics highlight the urgent need for more efficient prior authorization processes.

Addressing provider hesitation

Despite AI's potential, many providers remain hesitant to embrace automated prior authorization solutions. Friese attributes this reluctance to a lack of understanding about AI's capabilities rather than concerns about the technology itself.

"Once providers have a clear understanding of the impacts of the technology, their hesitation is usually not about using AI itself but rather about broader aspects of technology vendor implementation," he explained.

Common roadblocks include IT resource availability, implementation challenges and change management complexities. To mitigate these concerns, companies could partner with organizations that can help streamline implementation, reduce IT burdens and provide dedicated change management resources. By addressing these logistical hurdles, AI developers can encourage broader adoption and accelerate the transition to AI-driven prior authorization systems.

Enhancing patient outcomes

One of the most pressing concerns in prior authorization is its impact on patient care. Because prior authorization-related delays can lead to negative health outcomes, AI can offer a powerful solution by expediting the process and ensuring timely access to necessary treatments, Friese noted.

"AI can help speed up the entire process for both providers and payers, which has a direct impact on care and outcomes," Friese emphasized.

For providers, AI automates the most time-consuming steps, such as compiling the appropriate clinical documentation required for authorization.

"Bundling the right clinical data is a heavy lift and time-consuming for anyone, but it’s a task AI can accomplish quickly and efficiently," Friese said.

On the payer side, AI enhances review efficiency by helping decision-makers quickly determine whether a request aligns with coverage guidelines. However, concerns remain regarding AI-driven denials.

According to a 2024 Senate committee report, health insurers' use of AI tools has led to higher rates of care denials, sometimes 16 times higher than typical. This raises ethical concerns about potential inappropriate denials of necessary care. Statistics like these highlight the importance of implementing AI with oversight mechanisms to prevent unjustified denials and ensure fairness in decision-making.

Ensuring ethical AI governance

A significant ethical concern in AI-driven prior authorization is the potential for automation to introduce bias or unfair denials. Friese firmly advocates for a governance model in which AI can only approve requests but never deny them outright.

"AI should have the ability to say 'Yes' and to auto-approve an authorization, but it should never say 'No.' There will always be unique cases that require additional human review," he proposed.

To ensure this principle is upheld, Friese proposes an industry-wide scoring methodology in which only high-confidence approvals, such as cases scoring above 90 on a 100-point scale, are automated. Cases falling below this threshold would be flagged for mandatory human review. Such a system would ensure fairness, transparency and accountability while maintaining efficiency, Friese confirmed.

Balancing automation and human oversight

While AI significantly accelerates prior authorization, it must be balanced with human oversight to maintain accuracy and trust. Friese described a phased implementation approach in which AI initially operates alongside human reviewers.

"When first integrating AI into a workflow, teams collaborate closely with stakeholders to confirm that automation aligns with their needs," Friese explained. "At the outset, human reviewers remain involved, and the system refines its processes based on manual adjustments made to prior authorizations."

As providers and payers gain confidence in AI's accuracy, the system transitions to a more autonomous model, with fully touchless authorizations occurring only when stakeholders are comfortable with the technology's performance.

Bridging the information gap

One of the primary tensions in prior authorization stems from misaligned expectations between payers and providers. Providers often struggle to determine the exact documentation needed, while payers are inundated with irrelevant information, leading to inefficiencies on both sides.

"Both sides are feeling the extreme pain that this process can cause in its current, manual state," Friese emphasized.

AI helps bridge this gap by streamlining data submission so that providers send only the necessary information, enabling payers to process requests more efficiently. By removing ambiguity and reducing frustration, AI promotes better collaboration between stakeholders.

The future of prior authorization

Looking ahead, Friese envisions a future where 90% of all prior authorizations are touchless and processed from submission to approval without human intervention.

"However, there will always be complex, unique cases that will require a human review, which might end up being around 10% of cases," he acknowledged.

Beyond automation, Friese highlighted the importance of patient transparency: "Currently, [patients] do not have any visibility into the prior authorization process, but technology can play a role in providing those updates." By integrating AI-driven communication tools, patients could receive real-time updates on their prior authorization status, enhancing trust and reducing uncertainty.

As AI continues to reshape the prior authorization process, healthcare organizations that embrace this technology will continue to gain a competitive edge in efficiency, cost savings and patient satisfaction. However, successful implementation requires thoughtful governance, seamless provider–payer collaboration and a commitment to balancing automation with human oversight.

Alivia Kaylor is a scientist and the senior site editor of Pharma Life Sciences.