17:03
News Story
Biotech panel: Patient needs must come first in adoption of artificial intelligence, advanced tech
SIOUX FALLS – Virtual visits have saved Sanford Health patients more than 25 million miles of travel by allowing them to skip the trip to a clinic, the company’s chief digital officer said on Thursday.
But electronic medical records have made for less-efficient health care operations – at Sanford and across the health care industry, said Sanford’s Jared Antczak.
There’s often far more data collected than necessary, and its collection can gum up the gears of patient care.
Change is necessary, Antczak said, “so providers can spend less time looking at their screens, and more time interacting with patients.”
“That’s what they went to medical school for, that’s what many of them were motivated by, that’s what they aspired to when they became physicians,” he said.
The gulf between the possible and the practical in health technology was a recurring theme in a panel discussion at the South Dakota Biotech summit and annual meeting at the Minnehaha Country Club in Sioux Falls.
Antczak was joined by Jon Oh, the vice president of data and analytics at Avera, in a panel moderated by Avera’s vice president of hospital pharmacy and lab services, Tom Johnson.
Data analysis, AI
The mountains of electronic health care data now available and the emerging tools to analyze them are potentially transformative, the panelists said, not actually transformative.
Not yet, anyway.
“We are on the cusp of being able to use data at a large scale as technology advances and data science advances continue to move forward,” Antckzak said. “But it has to be done in an ethical way. It has to be done in a way that our patients’ interests are first and foremost.”

Sanford has more than a decade of information on 1.5 million patients across four states, which Antczak called “a treasure trove for data scientists.” One avenue being explored, he said, is “risk stratification,” through which the data can be analyzed to be proactive and address likely health issues.
Social factors account for half of any patient’s health risks, he said, so connecting patterns of geography and demography to health indicators could help providers connect people with preventative care before problems emerge.
“We’re just scratching the surface of that,” he said.
Sanford is also piloting the use of some existing artificial intelligence technologies. Artisight is an operating system that uses multiple cameras to collect data and analyze factors like a patient’s gait to predict falls, and Sanford has begun testing it in some nursing units.
Camera monitoring, or “telesitting,” has been around for years, Antczak said.
“This is taking that to another level, where I’m proactively nudging the nurse to say, ‘Hey, go pay attention to room 334. There’s a potential issue here, make sure that patient is assisted,’” Antczak said.
Potential vs. implementation
But Antczak and Oh, the Avera data lead, said several times that a notable share of existing and emerging health care technologies seem more helpful than they are.
Oh cited a radiology result alert he received from his patient portal app a few years ago as an example. The test results included a finding tagged with an unfamiliar medical term. Concerned, he asked a radiologist friend, who told him it was probably nothing to worry about.
Ask your doctor to be sure, his friend said.
“Well, the follow-up appointment wasn’t for two weeks,” Oh said. “So you know what I did for those two weeks? I searched Google. Word of caution, folks: Do not do that.”
The notification, ostensibly a means by which to keep Oh informed, played out as an anxiety-inducing portal to misinformation. The anecdote shows why connecting the dots between tech and patient needs ought to be at the heart of any decisions to adopt or innovate, Oh said.
GET THE MORNING HEADLINES DELIVERED TO YOUR INBOX
“Our only goal is to provide the best care for patients, and that’s it,” Oh said. “Nothing else should be considered in my opinion.
AI has emerged as a go-to suggestion for health care issues, Oh said, but it’s not always necessary. Sometimes, simpler is better.
The inefficiencies borne of electronic record keeping, he said, can be cataloged without complex algorithms. Avera recently ran a study on the number of clicks its providers made in the course of their day. A third of those clicks had nothing to do with patient care, and more to do with answering questions or ticking boxes wholly unrelated to a patient’s issue.
“Technology in itself, I think, has a lot of value – if it is understood and if it is implemented correctly,” Oh said. “So far, there’s not a whole lot of that.”
Obstacles to adaptation
Regulatory issues and the high price of approved technology tools stand as other hurdles to effective tech use in hospitals.
Avera trained an AI model on 10,000 radiology scans to identify pneumonia, Oh said. But the health system can’t use that information without FDA approval. That would require a legal team and a team of academic researchers to hit all the expected marks to earn that approval.
That’s expensive and time-consuming enough to nearly match the unaffordable price point for the existing technologies that can do the same work, Oh said.
“We can do it, but we can’t really get it to work for us,” he said.
Oh also said it’s important to remember that not all patients have the same comfort level with or access to the kinds of tools now available. Care at a clinic isn’t helpful if the patient doesn’t have transportation. Virtual care only works for people with internet access, smartphones or computers.
Solutions that leave out groups of patients or push them beyond their comfort level aren’t ideal.
“We need to make sure we’re not making assumptions about our patients,” Oh said.
Virtual care and online patient portals are now common and expected, said Johnson, the panel’s moderator.
That doesn’t mean everyone wants to use them.
“My parents, who are in their mid-80s, are not interested in virtual,” Johnson said. “Visits to the physician has become sort of one of their hobbies. I guess it works for them.”
Our stories may be republished online or in print under Creative Commons license CC BY-NC-ND 4.0. We ask that you edit only for style or to shorten, provide proper attribution and link to our web site. Please see our republishing guidelines for use of photos and graphics.