Healthcare AI Is Booming. So Why Are Providers Still Losing Billions?

Ramya Ganti is the founder and CEO of Oprox, a VC-backed AI-native revenue intelligence platform for healthcare providers.

getty

​The healthcare AI market is booming. Across the industry, sophisticated platforms now automate prior authorizations, scrub claims before submission and appeal denials faster than any human team could. The technology works. The results are measurable. And yet the administrative crisis in American healthcare is getting worse, not better.

The reason is not a lack of innovation. It is a matter of direction. The industry has been automating the wrong thing.

The Reactive Trap

Here is how the traditional revenue cycle works in most healthcare organizations today: A provider delivers care. A claim gets submitted. The payer denies it. The provider's team scrambles to appeal, rework and resubmit. Repeat, indefinitely.

The numbers behind this loop are staggering. According to Optum's "2024 Revenue Cycle Denials Index", an analysis of more than 124 million hospital claims across 1,400 U.S. hospitals, the average claim denial rate has climbed to 12%, up from 9% in 2016, with 84% of those denials categorized as potentially avoidable. Meanwhile, the administrative burden on physicians has reached a breaking point. A 2024 American Medical Association survey of 1,000 practicing physicians found that the average practice completes 40 prior authorization requests per physician per week, consuming 13 hours of physician and staff time. Ninety-five percent of those physicians reported that prior authorization delays patient care, and 94% said it contributes to burnout.

The first wave of healthcare AI addressed this by automating the reaction. Smarter appeals. Faster resubmissions. AI that flags likely denials after the fact. These are genuine improvements, and the industry deserves credit for building them. But they are still fundamentally reactive. They are optimizing a broken loop rather than addressing the causes of failure.

Why Reactive Automation Has A Ceiling

The problem with reactive automation becomes clear when you understand what is happening on the payer side. Payers are now deploying AI systems that can review and deny claims in seconds, processing denials at scale and speed that manual provider workflows cannot approach. Every time providers get faster at reacting, payers get faster at denying. The arms race has no finish line.

There is also a compounding human cost. The administrative burden does not just drain revenue. It drains people. AMA research found that physicians spend nearly two hours on administrative tasks for every hour of direct patient care. Prior authorizations alone consume an average of 24 minutes per request according to industry data, and more than one in four physicians say prior authorization has led to a serious adverse event for a patient in their care, while 78% report that delays cause patients to abandon care altogether, according to Waystar.

Faster appeals do not fix that. They just make the hamster wheel spin a little quicker.

The Shift That Actually Changes The Equation

The organizations starting to pull ahead are not the ones with the best denial management. They are the ones that have stopped managing denials and started preventing them. By applying AI-driven validation at the point of entry rather than at the point of rejection, organizations can transition from a reactive defense to a proactive offense.

This is a fundamentally different operating model. Instead of asking, "How do we recover revenue after it is lost?" proactive automation asks, "How do we ensure revenue is never at risk in the first place?" That means identifying authorization requirements before a service is rendered, flagging documentation gaps before a claim is submitted and surfacing revenue opportunities that the organization does not even know it is missing.

This evolution represents a fundamental shift in how healthcare organizations approach revenue cycle management, moving from reactive problem-solving to proactive optimization that anticipates challenges before they impact financial performance.

I have seen this firsthand working with behavioral health organizations. The ones still running reactive systems spend enormous energy chasing payments they should have collected weeks earlier. The ones that have moved to proactive intelligence know what they are owed before they even submit, and they structure their workflows around protecting that revenue from the start.​

What Proactive Automation Actually Looks Like

The distinction between reactive and proactive is not just philosophical. It shows up in concrete operational differences.

A reactive system tells you a claim was denied and helps you appeal it. A proactive system tells you three days before a patient appointment that the planned service will require authorization, surfaces the relevant payer criteria automatically and flags whether the documentation on file is sufficient to guarantee approval.

A reactive system recovers revenue you lost. A proactive system identifies revenue you did not know you were leaving on the table, including undercoded services, missed billing opportunities and contractual underpayments that would never surface in a standard denial workflow.

With predictive analytics, organizations can be proactive rather than reactive, which not only increases revenue but also reduces the stress on staff and improves relationships with patients and insurers.

The Road Ahead

Reactive automation was a necessary first step, and the companies that built it deserve credit for moving the industry forward. But the next decade of healthcare AI will be defined by organizations that stop optimizing the reaction and start eliminating the need for it.

The question for every healthcare leader right now is not whether their organization is using AI. It is whether their AI is playing defense or offense. The gap between those two answers is measured in revenue, clinician burnout and patient outcomes.

The technology to move from reactive to proactive exists today. The organizations that deploy it first will not just run more efficiently. They will deliver better care. We have spent years making broken systems faster. The next leap is building ones that don’t break in the first place.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?