The Future Is Autonomous. The Problems Aren’t.
Dr. Nick van Terheyden aka Dr. Nick
Host of News You Can Use
LinkedIn: Nick van Terheyden, MD
X: @drnic1
Why do we attend conferences? There are all sorts of reasons to go to conferences and they exist for almost every kind of activity you can imagine, from personal interest and activities to business and the many associations. Certainly, post-pandemic, it became an opportunity to redress the isolation I know I felt during the isolation and lockdowns many of us endured.
For me, that remains and the networking opportunities are always a huge plus, but varied depending on the conference, the number attending, and underlying familiarity with the space and even the conference
But primarily, I see them as learning experiences and increasingly try to arrive and check the bias I can easily identify in myself. That was my approach to the recent HFMA conference (agenda). And HFMA did not disappoint. The most valuable conferences aren’t the ones where everyone agrees. They’re the ones where the contradictions become impossible to ignore. This year’s HFMA meeting was full of those moments, from transparency rules that nobody actually uses, to AI solutions that seem designed to replace the people who built them, to denial processes that, as I have stated before, increasingly resemble a technological arms race (see below).
Transparency Nobody Reads
One of my favorite questions when I meet people is what their main problems are… what keeps you up at night. Mostly, there were no surprises here with denials and billing problems featuring top of mind, and cybersecurity coming in close behind. But in an early conversation with a leader in a large health system in charge of Revenue Integrity, the response took me by surprise
“Transparency file and getting that produced.”
This activity has turned into a major time sink, and as I found out later, an uncompensated activity that has burdened health systems with more costs with no clear benefits.
Lauree Handlon, Vice President, Reimbursement Policy, and Anna Barnes, VP Product Management from The Craneware Group, in their session “Beyond the Numbers: Making Sense of Price Transparency Data in a Negotiation-Driven World,” all made much more sense. Hospitals are required to produce a big machine-readable JSON file that it would appear almost nobody reads, and certainly not the original target audience of patients, as it requires specialist tools and software. They captured the healthcare transparency currently. Organizations are spending enormous resources generating files to comply with regulations while openly acknowledging that few people derive meaningful value from them. Early compliance was poor, but the warning notices and fines arrived, and suddenly everyone became interested in transparency. Funny how that works.
Hospitals publish enormous machine-readable files, but with little standardization, simple differences make for a massive data set that one organization calls a payer “BCBS” and another uses “Blue Cross Blue Shield.” And that’s before the different contract structures, naming conventions, and formats.
So good luck to any patient able to view these files, trying to get a sense of what their bill might be for a specific procedure.
But as is often the case, there are unintended consequences, and of special notice for the hospital and health systems required to produce these files. Payers are now mining these public files to negotiate lower reimbursement rates. Ruh row.
Transparency may finally be working, but perhaps not for the audience originally intended.
My takeaways
- This created a whole new business line and companies that have focused on ingesting this data to sell it back to those same organizations, payers and even patients. To be clear, this feels like an added cost to the system not a saving, at least not any ending up in the patient’s (empty) financial bucket
- Healthcare organizations should start paying close attention to the transparency data they’re publishing, not just to comply but to focus on the content and try to reconcile against others in their marketspace, because payers certainly are. I know, more cost, but failing to do so will hurt the bottom line in the future if it has not already
Denials: Healthcare’s Most Expensive Hobby
One of the stand-out sessions came from Intermountain Health’s denials management team. Their message was both impressive and depressing.
Impressive because of the sophistication of their work.
Depressing because of the amount of work required to overcome payer behavior.
And denial rates continue to run at least from 11 – 15% (as reported in this presentation) and the cost of recovering denied claims is rising, with each “re-touch” of a claim costing around $25. Think about that. A claim may bounce through multiple reviews, appeals, corrections, and resubmissions before ultimately being paid. Healthcare organizations are spending enormous sums simply trying to get reimbursed for care that was already delivered. But you knew that already. One statistic nearly caused collective whiplash in the audience:
93% of claims from a single payer were denied.
And in a conversation I had later at the conference with one individual, reported that in some cases it can rise to 100%, which were subsequently overturned on appeal. At some point, we should stop calling that utilization management and start calling it what it is: administrative warfare.
The question that stuck with me was asked by the presenters:
“Do you have people or do you use technology?”
The answer increasingly appears to be:
Both.
For now.
But the trajectory is obvious. Payers are deploying automation. Providers are deploying automation. Appeals are being automated. Reviews are being automated. Documentation is being automated. This increasingly feels like a battle between machines while clinicians and patients wait to see who wins.
Ultimately, the future of denials management may become the first major healthcare conflict fought almost entirely by AI agents. I covered this in a presentation at NPAC 2025 with my friend and colleague Justin Krawitt MD, MHCDS, CHCQM, Sys. Medical Director, Dartmouth-Health: Artificial Intelligence: Your New Sidekick, Not Your Replacement!
And this graphic I think, captures the arms race we are in
Cybersecurity
Sigh. We know this is a problem, and anyone working full-time in a healthcare company and for any organization that has you using corporate email and laptops, you are familiar with the never-ending stream of training and awareness, but especially testing employees. The session “Cybersecurity in the Age of AI: Risks, Resilience and the Human Factor”, hosted by Brandon Burnett, Vice President, Revenue Cycle at Community Health System and Greg Surla, SVP, Chief Information Security Officer for FinThrive dived into the increasing risk posed by AI enabled bad actors and the many way sin which new technology has expanded the attack surface and speed we all face. I am reasonably certain that data poisoning was a new term for many and will no doubt be a new word added in the annual announcements if it hasn’t been already, like “Bomb Cyclone”!
But the relevant point that was not surprising to me was that people are the weakest link. No surprise here. I have talked about this and covered it extensively over the years and continue to do so:
Trust No One (2019) and this interview, “Raising Red Team Thinking in Healthcare” (and a link to my other cybersecurity-focused articles).
But what I heard seemed too simplistic – More mandatory training.
But in my head I am going through my experiences in prior companies where a new CISO arrives and implements this, failing to recognize that in some cases, some of the employees probably know more about some of the topics than you might expect. Mandatory HIPAA training immediately comes to mind. My experience, I suspect, is typical where the programmers have found ways to force people to sit and watch bland, irritating videos they created, thinking that video scenarios would be helpful and engaging. Maybe it is for some, but not everyone. For me, it was a royal waste of an hour or more while I was forced to watch videos for information I could have gleaned from written material much faster, followed by a knowledge test.
That’s back to front. Why not test people’s knowledge going in and then use the results to inform? Targeted reminder/refresh followed by a targeted retest
What I did hear and like as the concept of team meetings sharing some of the attack details and approaches, and while not mentioned, perhaps bringing this back to the individual, because cybersecurity, much like healthcare, is personal as well as a business activity. Very personal in my case, as I detailed my post-mortem (3 Minutes to Financial Ruin) when I was subjected to a relatively sophisticated SMISHing attack (aka SIM jacking). I encourage you to read all about it and try to secure your life against this as much as is possible
The most shocking moment came when one organization disclosed that employees who fail three phishing tests within an 18-month period are terminated. There was no mention of the denominator, so I have no idea how many phishing messages they sent out in that period. No context but a clear
Just three strikes and you’re out.
Apparently, cybersecurity now has its own version of baseball.
My Takeaway
The Beatings Will Continue Until Security Improves. But the smart companies will move away from this and build awareness in a positive fashion.
The Show Floor
I had lots of great discussions and interactions with folks on the floor. Catching up with old friends and colleagues, as well as meeting new people at the booths. There was a good balance of bringing attendees and people who work in the industry to the floor, interspersed with sessions taking place at other times.
Prior Authorization: The Nuclear Arms Race Nobody Wants
Once again, no surprise to find AI all over the show floor – in some cases, window dressing and “me too”, but the technology is impacting many areas and finding utility. It was hard to determine what some companies did based on their signage, and many were just outright confusing, at least to me. But Rhyme was a standout for me.
Their booth had a 3-word sign – that was it:
“Eliminating Prior Authorization.”
Not optimizing.
Not accelerating.
Not automating.
Eliminating.
And in a real approach to the fundamental problems we face in healthcare, they are as Brett Boswell described it to me
“Acting as a Marriage counselor.”
And working with payers alongside, who are even funding some of the efforts to introduce technology that removes the need for prior authorization. I really enjoyed my discussion with Brett, refreshing for its lack of focus on technology and AI, and more about solving the problem and insistence on getting everyone involved in the solution. For me, that represents something we don’t often see in healthcare:
A genuine attempt to solve the underlying problem rather than merely improving the efficiency of suffering through it. Instead of building a better missile, they’re trying to negotiate peace.
Healthcare could use more of that.
While on the topic of quotes and honesty, another standout for me was the quote from Teri Kindzia from WelcomeWare – a check-in kiosk solution, but not your typical dumb terminal affair
“Healthcare self-check-in is a tool, not a solution.”
Simple, accurate, and applicable far beyond self-check-in. Replace “self check-in” with AI.
Or ambient listening.
Or predictive analytics.
Or digital transformation.
The statement remains true. Technology is a tool that may help create a solution. Healthcare would save itself considerable frustration if we remembered that distinction more often. I know, not revolutionary, but it is not often you hear this in my experience acknowledged on a booth at a conference.
Their application was compelling, and I came from a position of skepticism, given my awful experiences with these self-check-in kiosks, not just in healthcare but elsewhere. But this is not the dumb terminals or mobile devices placed in waiting areas configured to register you and, in some cases, administer clinical surveys; this is a one-button connect you with a real, live person who is plumbed into the systems and able to carry out registration and billing functions interactively.
This was one of many examples of outsourcing from the show floor (although, to be clear, you can still use your own staff, but make them more efficient and allow centralization or even remote work). In the case of UCB Intelligent Solutions, they outsource to Costa Rica, but all the usual countries were represented.
On a side note UCB had the most fun booth giveaway. A nice piece of jewelry that required participants to guess how many plastic crabs were in the jar:
The activities around this represented another small microcosm of our new world with AI. Some people guess, others like me, “cheating” using AI to take a picture and have it take the guess. But according to the booth staff, AI, when it produced anything (and in my case, I never got any suggestions, so I reverted to guesstimating) the answers were wrong, sometimes by a big factor!
US Healthcare Corrupts Good Intentions
I met with one company, “Nutriguard,” who were using image processing technology to address the issue of meals and nutrition in hospitals. If you have ever been a patient, and sadly also true for workers too, you will know that hospital meals can sometimes be a real hit or miss affair. Nutrition and its impact on health have long been a missed opportunity, as evidenced by the recent announcement by Secretary Kennedy to create nutrition standards in medical schools. Makes sense, but to be clear, given the gargantuan amount of content already crammed into medical education that has expanded every year since I went to medical school, something’s gonna have to give to fulfill that promise – just saying.
But this conversation represented yet another microcosm of problems in our healthcare system. They are a Singapore-based company, and it is in use in the Singapore health system. They have a product to take a picture of the meal and then assess it for nutritional value. The Singapore health system is using this to try to improve nutrition in the meals which has downstream benefits on pateints overall health and recovery. They even developed a post meal image processing to do a before and after assessment. Cool!
But in the US, the discussion quickly shifted toward identifying malnutrition after meals to improve coding and reimbursement.
The technology was identical.
The priorities were not.
Franz Kafka would have appreciated the irony.
Final Wrap Up
One of the troubling things I heard in discussions triggered by the details in price transparency and denials data was the wide swings in pricing and the frequency with which these changes. My mind quickly jumped to the use of Dynamic pricing in other industries. At this point, well-known in online shopping, especially Amazon, the airline industry, and taxis uber. What is perhaps not as widely known is the aberrant use cases that these companies have come up with. In the airline industry, they are known to review your device signature and change the price depending on what equipment and browser you use as a proxy for how much money you have. If you are perceived to be richer based on these data points, they will present higher pricing (and A/B test these differences to fine-tune). Not new behavior, just a digital version of what happens in real life. But in the case of Uber, they took it to a whole new level by obtaining users’ battery levels, and if you were unlucky enough to have a low battery, they would increase your pricing, expecting you to be more desperate to accept higher prices 😳❗️
Now imagine that in healthcare. I am not suggesting it is taking place, but rapid swings and variations have echoes of dynamic pricing. Like most cases, imo it might not start as out-and-out bad behavior, but through a series of smaller steps, stepping past what would be considered reasonable. I can’t help wondering whether one day healthcare pricing algorithms will be evaluating your deductible status, zip code, and financial profile before deciding what appears on your bill.
My Takeaway: Hopefully not. But healthcare has a habit of adopting innovations from other industries after first convincing itself it never would.
What HFMA 2026 revealed is a healthcare industry caught between two realities.
The first is the world we have today:
- Massive administrative burden
- Rising denial rates
- Compliance-driven transparency
- Fragmented workflows
- Endless manual intervention
The second is the world being built:
- AI-driven automation
- Agentic workflows
- Predictive operations
- Intelligent documentation
- Autonomous decision support
The challenge is that we’re currently operating both systems simultaneously. Meanwhile, every new technology promises simplification, but somehow healthcare keeps getting more complicated. Still, there were reasons for optimism. Organizations like Intermountain are proving that data can fight denials. But sharing is caring here, and while the presentation did that, we need a nationwide sharing of these insights between hospitals that overcomes the competitive barrier hurdle and tries to restore balance to the competing interests.
Innovators are exploring entirely new approaches, but it is going to take approaches like Rhyme questioning whether we are even doing the right thing at all, and should prior authorization exists at all. I think the surprise might be how much we agree on just this data point if we engage in some of that “Marriage Counseling”.
I am certain that the future may not arrive as quickly as the speakers promise and won’t be evenly distributed quickly either. But after attending HFMA, it’s clear that healthcare’s next major battles won’t be fought between providers and payers. They’ll be fought between their respective AI agents.
Let’s just hope the patients benefit from the outcome.
This article was originally published on the Dr. Nick – The Incrementalist blog and is republished here with permission.