Nebraska Medicine, based in Omaha, Neb., is using generative AI to develop its own tools internally — a move that is reducing reliance on outside technology vendors and accelerating innovation timelines.
“What’s super exciting now, when you think about the introduction of generative AI, is all of that time to train a model and label the data — that’s already been done for us,” Michael Hasselberg, PhD, chief transformation and digital officer at Nebraska Medicine, told Becker’s. “Big companies like OpenAI, Amazon, Google and Anthropic have already spent billions of dollars to build these models that come to us pre-trained.”
Before the advent of generative AI, developing an internal natural language processing model could take six months to a year, Dr. Hasselberg said — and often with disappointing results due to the complexity and variability of healthcare data. Now, by using pre-trained foundation models, his team can prototype tools “in days to weeks.”
“You no longer need to be a traditional programmer to build something,” he said. “You can just use natural chat and prompt engineering to build tools. The playing field has been leveled for health systems — and if anything, skewed toward health systems — because we have the same technology now to build off of, and we have the data, which vendors don’t.”
Nebraska Medicine has already rolled out 22 AI tools developed entirely in-house, focusing on the administrative side of healthcare, such as capacity management, staffing and scheduling.
One example is an AI-driven system that creates “baseball cards” for staff members — digital profiles that include each person’s strengths and specialties.
“We can take those baseball cards to move our staff around to meet the needs of our patients,” Dr. Hasselberg said.
Another area of focus is optimizing surgical throughput and scheduling to ensure the system’s surgical department, which serves patients across the state and beyond, operates efficiently from preoperative scheduling through recovery.
On the revenue side, Dr. Hasselberg said the health system has targeted “low-hanging fruit” by using AI to streamline prior authorizations, denials and appeals, and clinical coding — areas where vendors have traditionally offered standalone solutions.
Despite Nebraska Medicine’s push toward internal development, Dr. Hasselberg said AI is not creating distance between health systems and vendors — rather, it’s changing the nature of the partnership.
“I think it’s making deeper partnerships with vendors,” he said. “Gone are the days where healthcare providers are at the mercy of the vendors — of them driving where to go, how fast we can move. Now we’re directing the ship.”
He said vendors that want to remain competitive should broaden their offerings and move beyond single-use tools, since health systems cannot sustain “a thousand different bolt-ons” to their tech stacks.
“It’s financially not sustainable, and from a cybersecurity standpoint, it’s not viable,” he said.
Dr. Hasselberg believes the next phase of AI in healthcare will require both local customization and strategic collaboration.
“There’s still a very local part to AI,” he said. “Even with large language models, you really need to hone those models to the patient population you serve. Who we serve in Omaha, Nebraska, is very different from health systems in Southern California or New York City.”
