Will AI transform health care administration, specialty care?

Administrative costs account for 15-25% of all health care spending in the United States, totaling potentially more than $1 trillion per year.

A new review in Clinical Gastroenterology and Hepatology investigates whether the current landscape of artificial intelligence administration products may help reduce these costs, specifically in the context of gastroenterology practices.

In this Q&A, senior author Eric D. Shah, M.D., M.B.A., Associate Professor of Internal Medicine at the University of Michigan Medical School and Chair Elect, AGA Committee for GI Innovation and Technology, discusses the present and future of specialty-specific AI tools. 

What are the main findings of this latest study in Clinical Gastroenterology and Hepatology?

Shah: There are a lot of new platforms and technologies that are coming out focused on the administrative side of health care.

That focus is not surprising, as administration accounts for a quarter of all health care expenditure now. There is a clear need to reduce administrative bloat.

Some of these products exist purely to assist with administration as we know it. 

There’s potentially a larger opportunity, however, to use AI to look at whether administrative costs are adding value and how we could improve those processes.

Our study was focused on the AI tools now available in one clinical area: gastroenterology. For now — and this is common when things are new — there's not much there yet. 

We're seeing a lot of promises as to what can be delivered. But when you lift the hood, the level of evidence to support claims is not quite there yet.

Do the products that do exist — GI-specific or not — seem to be addressing administrative costs?

Shah: Current AI products focused on health care administration often have a specific conception of efficiency.

Those tools are purely focused on getting more patients in, which can add to health care worker burnout, and does not necessarily improve patient outcomes or health.  

We all must ask: Is more better? Or is more just more?

As the number of appointments increases, there’s also the possibility these tools lead to even more complexity, more administrative spending.

From a physician side, we already worry that the physician-patient relationship is turning too much into a physician-patient-administrative relationship.

The most promising products are the ones that are taking that on directly. But those tend to be the more early-stage products at this point.

How would AI administration tools ideally improve the status quo?

Shah: Research shows that during a routine clinic visit, patients and physicians have about eight minutes to talk about everything, including the physical exam.

How are you supposed to get everything done in eight minutes, even with an artificial intelligence tool?

What AI seems to be hoping to do within a decade is remove all the administrative concerns from that time, including billing issues. That time could be completely replaced with the patient talking to a doctor.

That’s a noble goal: Time with patients builds trust and trust creates a situation where you can uncover even more and get more done.

My concern is that may not happen. We may just end up having more eight-minute visits.

What are some GI-specific opportunities for AI tool development?

Shah: There’s an opportunity for a liver transplant app. It would be great to see products that are able to help patients get through that process.

Another possibility might be a tool that can identify appropriate clinical trial candidates based on what the physician sees during a routine endoscopy procedure.

We did identify some tools that are in development specifically for colonoscopy preparation. 

That's important because there are a lot of patients that are undergoing non-invasive tests for colon cancer screening who don’t necessarily follow up with a colonoscopy.

How can AI improve specialty care?

Shah: One reason that the specialty-specific AI solutions might be more effective is they can capture the nuances of what patients with certain diseases and certain symptoms are going through.  

This possibility is especially important in gastroenterology, in which the ability to describe symptoms can be difficult but essential. Putting the way your gut feels into words is hard.

Some of these GI-specific AI solutions are trying to parse a patient’s words to help both the patient and physician to ask the right questions of each other.

That also underscores the importance of how we use the time we might get back from administrative AI tools. 

Gastroenterology appointments, especially, benefit from time spent talking with the physician. 

If artificial intelligence can save time in health care, we want to give that time back to our patients, so we can better understand how we can help. 

Paper cited: “Artificial Intelligence-Enabled Opportunities to Reduce Administrative Bloat in Gastroenterology Practices: Scoping Review,” Clinical Gastroenterology and Hepatology. 10.1016/j.cgh.2026.06.036

Sign up for Health Lab newsletters today. Get medical tips from top experts and learn about new scientific discoveries every week.

Sign up for the Health Lab Podcast. Add us wherever you listen to your favorite shows.