Local health care takes selective approach to Ai

But they are taking baby steps toward utilizing the new technology.

AI is cutting diagnostic turnaround times in radiology — particularly for treatment of strokes — reducing wait times at clinical reception desks and could augment pharmacists’ and physicians’ oversight of prescriptions.

All this is being done to increase interpersonal face time with patients and make it more efficient for everyone, they say. It will support, not replace, personal medical personnel with years of training and expertise.

Brian Haase of Denver, a Waterloo native who heads up imaging for UnityPoint Health, is seeing some of the greatest use of AI in his field. He worked at UnityPoint Health-Allen Hospital in Waterloo 25 years, and now oversees imaging across the entire organization.

It’s happened largely without a massive recapitalization of equipment.

“We’ve been able to work with the equipment we have for most of the AI we’re talking about,” Haase said. “A lot of the AI runs inside the hospital information system. So the computers that we use, the patient’s chart, the AI works right in there.

“What AI does inside of the chart is it helps guide ordering practitioners, doctors, to the right exams, the most appropriate (imaging) exams,” he said. “They can put in the patient’s information. The AI pulls in the patient’s demographics, the patient’s age, the basics about the patient, just to guide the physician to the most appropriate exam, the exam that has the least radiation exposure, or even the exam that’s most affordable.”

AI is also helping radiologists “re-prioritize” exams, Haase said. “It looks at the exam before the radiologist does and says, ‘Hey there may be an incidental or a critical finding in this exam.’ So it re-prioritizes it up to the top of the radiologist’s work list, highlights it, says, ‘Hey, you need to look at this one next.’

“It’s definitely a timesaver for the radiologist, but even more about prioritization, just making sure the right stuff gets looked at first,” he said.

“It’s measured progress in radiology,” Haase said. “How it helps the staff is we don’t get the phone calls we used to get. Doctors used to call X-ray departments” for test consultation. “A lot of that guidance is in the medical record now in the form of AI that helps them find the exams.”

With companies promoting a plethora of AI platforms and products, “the hospital systems are being real careful to make sure we’re choosing the right one that still allows the radiologist to be the one in control.”

Detection, treatment and prevention of strokes through AI “is really the very first application that caught fire in health care,” Haase said. “The rest of these things, the time factor isn’t so pressing. Stroke, they need decisions right now. These AI tools, we send them the CT data before the radiologist looks at it and they send the result back in less than five minutes. The AI takes that patient’s stroke information, looking for a brain bleed or looking for an area that’s not getting the blood it needs, and it compares it to normal populations. It looks for outliers.

“You need to take so many things into account,” in diagnosing and treating a stroke, Haase said. “AI can take all those things into account, based on the patient’s history, based on prior imaging. It takes everything into account. Then it can classify the stroke” and tell the radiologist how to treat the patient — surgery, medication, catheterization, transfer to University Hospitals and Clinics in Iowa City, whatever is most appropriate. Most hospitals use the same software UIHC does in that regard.

“For stroke the important part is that we get the ordering doctor information within 45 minutes of the patient hitting the door,” Haase said. “You can imagine with a five-minute turnaround on the imaging result, that makes all the difference in the world.

Other AI platforms are slowly working their way into health care.

“There’s so many applications of these broader-based AI platforms, looking for things like aneurysms, gastro-intestinal bleeds. I’m not kidding. There’s an AI platform for everything,” Haase said.

The technology is also useful in ensuring health equity.

“It helps us make sure we’re giving every patient the same level of care” no matter their socio-economic or educational status. “So it elevates the whole field,” Haase said.

There also are AI tools to manage follow-up exams and treatment, making sure patients don’t fall through the cracks.

The AI tools available to radiologists will expand.

“There’s work being done across every modality for ultrasound and MRI and CT and mammography. It’s really those critical exams like stroke where the work started,” Haase said.

Brent Bovy of Parkersburg Pharmacy and Reinbeck Pharmacy said what’s known as AI now is not really new. It’s sometimes hard to distinguish between what’s called AI and what isn’t.

“We’ve always relied on technology for a lot of different stuff as far as drug interactions, just the processing of claims, things like that,” he said. But AI is only as good as the data inputted and has to be balanced with human oversight and direct interaction with patients.

“One place we’ve used AI already is inventory control,” Bovy said.

Programs can predict what drugs and what quantities pharmacies are going to use over the next 30 days. “So that’s a big part of our costs here. Inventory control is pretty important,” Bovy said.

“Typically, we don’t have much control over our price once it goes though insurance. But cash price, which is a small part of our business, those we can control. There are different companies you can contract with that use AI, they can see your market area and see competitors and things like that,” Bovy said.

Bovy can see AI helping monitor drug interactions with patients, but there always has to be a human element. Electronic pill counters, for example, provide “a nice double check” on filling prescriptions.

Mary Greenley is chief revenue officer at Cedar Valley Medical Specialists. Based in Waterloo, it serves patients from 16 counties in 27 clinics.

“Where we’re at right now, from a clinical aspect, we’re not utilizing a lot,” Greenley said.” I think radiology is one of your first areas that you heard about AI. We’re looking at it more from the administrative side of things. And not to reduce staff, but to help supplement some of the other duties they have, just because it’s a growing business.

At one of the large clinics, AI answers phone calls and handles simple questions. “And that has really reduced the flow to reception area — they’re your front-facing-patient people.”

For a patient, “It’s hard to come up to a counter and have someone sitting on a phone when you want to check in,” Greenley said. “So that (AI use) has been really successful, and that just rolled out in the past two or three months. It reduced their calls by 40% to 50% just within the first week.

“It was actually at our digestive health center,” she noted. “So a lot of patients are there for colonoscopies. They’ve gone through their prep the day before, and they just want to get to the back of the building,” she said.

She is an advocate of “the human element” and was concerned about patients calling in, especially the elderly, getting caught in a phone tree. “It was actually implemented very well and it’s very user friendly. So my skepticism was put to rest a bit,” she said. “You can hit a live body if you need to, and the chatbot that we’ve worked with is very realistic.

AI is also helping with patient literature, especially with elderly patients. The clinician obtains information through AI and edits depending on patient’s needs. The literature reinforces one-on-one conversation with a patient and family members.

She said AI may be used in the future for clinical financial reports, numbers of patients seen and collections tied to patient visits. “And with that, some intuitive appointment analysis, looking at no shows and how to get those slots filled, rather than a provider having a two-hour open spot on their appointment schedule because of no shows.”

As far as AI in general goes, “I think we’ll see a lot more traction this year,” she said.

“I think that’s where you read about AI, when it comes to your imaging, when it comes to a ‘bot helping with diagnosis, that type of stuff,” Greenley said. “I don’t know if we’ll get the buy-off on that from the physicians. It’s the clinical expertise. They went to school for those things, and they are the experts in those arenas.”

“Everybody’s being real careful to say AI is not going to replace radiologists or doctors,” Haase of UnityPoint said. “There have been real innovative ways technology is helping us. We can do more with less than we ever have before” and it helps radiologists and potentially others spend more time with patients.