AI garners focus at annual health care event

Katie Towns: Springfield-Greene County Health Department is yet to utilize AI.

Harnessing the power of artificial intelligence in health care while taking caution of its potential risks took center stage at a May 7 event organized by the Springfield Area Chamber of Commerce.

A 2022 National Academy of Medicine report notes potential benefits of AI in health care include improving outcomes for patients and clinical teams, lowering health care costs and benefiting population health. Panelists at the annual Health Care Outlook event agreed and noted technology use is widespread in the industry. Dr. Sadaf Sohrab, chief medical officer with Mercy Springfield Communities, and Dr. Shawn Usery, senior vice president and chief medical officer for CoxHealth, joined Jarad Johnson, founder and CEO of digital marketing agency Mostly Serious LLC, on the panel moderated by Jonathan Groves, professor and chair of communication at Drury University.

When it comes to health care usage, Usery said patient safety is the most important consideration. As AI needs large amounts of diverse data to be most effective, he said it’s imperative not to use all the internet content to train it.

“You guys have looked at the internet, and there’s a lot of junk out there, and more every day,” he said to the audience of 320.

Additionally, he said using a tool like ChatGPT to seek medical information about a patient is unsafe.

“Because all that’s doing is taking that patient information and exposing it to the world,” he said.

Johnson said his firm trains clients, including those in the health care sector, about AI tools, such as ChatGPT and Microsoft Copilot.

“I don’t think that what we’re facing right now with generative AI is overhyped,” he said of the type of AI that uses algorithms to create new content and ideas, such as text, images and video. “It’s easy to believe that this is something fundamentally different, but it’s also something that isn’t brand new.”

Still, Johnson said the technology comes with different considerations.

“One, we all have access to this insanely powerful tool that often makes mistakes, and that is dangerous. Two, we don’t always know exactly why it works,” he said.

Setting guardrails
Sohrab said it was only a matter of time before AI was used in health care, adding that Mercy recognizes the technology is the future.

“This is something that we as humans have to equip ourselves with, learn about it and also make sure that we have the right guardrails,” she said, noting that includes learning how to use deidentified data to better serve patients. “What are the ethical issues that may arise?”

AI has so many potential applications in health care, Sohrab said, adding that includes diagnoses, operational purposes and scheduling patients.

Roughly 65% of U.S. hospitals reported using AI-assisted predictive models, according to a 2025 study from the University of Minnesota School of Public Health. These models were most employed to predict inpatient health trajectories at 92%. That was followed by 79% using it to identify high-risk outpatients and 51% to facilitate scheduling.

However, while 61% of hospitals evaluated their predictive models for accuracy, the study said only 44% conducted similar evaluations for bias.

Sohrab said making sure unconscious bias isn’t fed into AI models is extremely important.

“If there is bias in health care and in our data, then that’s what the system will learn, and that’s what the algorithm will then produce,” she said. “It’s a question of trust. So, when a person trusts you with their information, it is our responsibility to make sure that that information does not get into the wrong hands.”

Usery said rather than using a “whack-a-mole” approach to address issues as they arise, he said organizations need to consider how AI can best make sense for their needs.

“I encourage organizations to take a step back and say, ‘What are our gaps? What are our opportunities? What do we need to do? How do we get more efficient at fulfilling our mission?’” he said.

Johnson said a mistake he commonly sees organizations make – both in and out of health care – is trying to find a solution to a problem that doesn’t exist and using generative AI software to do so. He recommends companies introduce more basic AI tools for a small cost to their employees, such as ChatGPT, rather than investing in high-dollar custom solutions.

“Let’s get everybody in the organization onboarded to some of these tools that are a little lower effort that we can teach them what works and what doesn’t work,” he said, adding that includes more safeguards for his firm and clients in the health care space, such as nondisclosure agreements. “We have a few more gates to pass through to make sure that it’s compliant.”

Springfield-Greene County Health Department Director Katie Towns, who was among those who attended the May 7 event, said more businesses need to learn how to use AI. That includes hers, as she said the health care organization is yet to utilize the technology. As the department also is a government agency, she said it isn’t a simple decision to just start using AI, but added it could help with capacity for fielding phone calls.

“We have to clear a lot of hurdles,” she said.

In action
While Usery said health care is much more reserved than some organizations may be to AI usage, both he and Sohrab said their health care systems are utilizing the technology.

Sohrab said Mercy refers to AI as augmented intelligence. The technology can help free up time for physicians and improve patient care, she said. For example, Mercy is using DAX Copilot, an AI-powered tool that automatically documents patient-physician encounters. She said those who have used it reported saving at least 60-90 minutes a day of documentation.

“That means you can see five extra patients per doctor per day,” she said, adding that alternately could mean more time for those employees seeking a better work-life balance. “So, it saves you a lot of time and it’s accurate.”

Mercy also is using AI technology to assist radiologists in prioritizing and analyzing imaging studies through Aidoc, a platform which enables a comprehensive additional review of most diagnostic imaging to identify suspected conditions, including brain hemorrhage, cervical spine fracture, lung nodules and pulmonary embolism.

“That way, the radiologist who has to go through 20 or 30 images for the day can prioritize and start with those that are flagged by AI so that those patients can get help sooner rather than later, because time is of the essence,” Sohrab said.

Usery said the technology also can help sustain health care careers, which is amid a nationwide shortage. According to the Association of American Medical Colleges, an 86,000 physician shortfall is projected in the U.S. by 2036. 

Johnson said there’s still a fear factor of technology that some need to overcome to more fully embrace AI.

“There are going be reservations about these things,” he said. “The more we can educate people on how they work, give them the tools in people’s hands so they can experiment with it, the more we lower that fear and people understand what they are.”