Like many health systems, UChicago medicine is experimenting with artificial intelligence. It has external partners and internal teams examining use cases to address critical problems and evaluate return on investment.
The opportunities are endless; but so are the risks. Even after evaluating the AI solution and finding it solves a true pain point for the system, Cheng-Kai Kao, MD, chief medical information officer at UChicago Medicine, said his team still has a very thoughtful process for validating AI.
“We look for accuracy, completeness and all types of evaluations around effectiveness of AI,” he said during an interview with the “Becker’s Healthcare Podcast.” “There’s also a lot of risk assessments, especially nowadays with a lot of cybersecurity risk. We make sure we are reviewing this AI in a fair way to adjust to their risk level based on whether the use of AI is susceptible or not susceptible and a high or low risk.”
The health system also has a process to continuously monitor the AI for continued accurate performance.
“The health system is constantly changing. All these changes, regardless of spending on new units, building a new hospital, adding additional outpatient sites, all potentially change how the AI model works,” said Dr. Kao. “We want to have some line of sight into that too.”
The AI governance process includes leaders in data science, analytics, informatics, security, privacy and other stakeholders to ensure new technology roll-out and monitoring is equitable. UChicago Medicine formed the Center for Digital Transformation to lead digital health and AI employment in clinical care for improved outcomes and patient experience.
“We are taking a very thoughtful approach when we develop this technology,” said Dr. Kao. “We talk about the patient community we serve. It’s not exactly the most digitally savvy population. So we make sure every time we roll something out, we’ll always have digital equity in mind. We want to make sure that the things we deploy are benefiting all the patient population, maybe not all through the same venue, but always making sure there’s a way to engage and help them navigate and benefit from all the new solutions we deploy throughout the systems.”
Not all patients have digital devices they need for care at home and communication with providers. UChicago Medicine has made progress in helping patients secure those devices since the pandemic, but they’re still working on process improvement.
“Now with AI, there’s even more to think about around digital equity, like biases based on gender, race, populations, etcetera,” said Dr. Kao. “How does AI use being properly governed to make sure that we are free of those biases or ethical concerns? It will be the other challenge we need to constantly test out and be mindful of in our processes.”
UChicago Medicine has been trying to solve equity challenges for years to close gaps in access to care. New technology can make a difference.
“How can we use these tools in the right way to actually mitigate some of these gaps and make things better, not the other way around?” said Dr. Kao. “It’s just something we need to constantly be careful about when we march forward.”
There is a continuous stream of new companies and use cases for AI in healthcare, and many warn that physicians who use AI will replace physicians who don’t use it. The future will include more AI integration and automation. How can physicians prepare?
“What we want to do is make sure we elevate the AI literacy in our institution. We want to train people to know how to make the best use of AI,” said Dr. Kao. “AI is not perfect. We are training them to understand what posting counts are, what’s the limitation, what do we mean by AI hallucinations. In the meantime, with all of that, we’re also being able to leverage AI as our most inexperienced intern. It’s someone on our team, almost like your digital worker who is inexperienced and new to the team. How do you train the AI to actually work with you to get things done more efficiently? That will be our goal.”
Dr. Kao is building a HIPAA-compliant generative AI platform that is secure for internal use and will allow UChicago staff and clinicians to upload information about patients and execute certain tasks, including drafting messages to send in different formats or languages at different literacy levels, which can improve communication.
His team is also developing modules to draft denial letters of appeals and answer questions about policies and procedures. He expects to have the platform ready in the next few months and then pair it with staff communication and education about AI, alongside prompting instructions and AI governance.
“I’m going to spend quite some time to help stand it up and think about creating different use cases, and training people to make use of it,” said Dr. Kao. “We’re going to be really improving a lot of efficiency issues we have and really allow people to do the things that are at the top of their license. Let AI do repetitive tasks or something that’s less challenging per se in terms of complexity or originality. Reserve all your energy for something that requires it and the AI can help you save time.”
