As health systems aspire to extend care beyond the visit, AI tools are becoming essential to orchestrating pre and post-visit outreach without adding to the burden on the clinical workforce. This was a main theme at a panel discussion sponsored by Nabla, an ambient AI assistant provider, at Becker’s 16th Annual Meeting.
Panelists were:
- Tomi Kolade, assistant chief medical information officer, UTHealth, The University of Texas Health Science Center at Houston
- Ryan Sadeghian, MD, chief medical information officer and medical director of AI, University of Toledo (Ohio)
- Matt Sakumoto, MD, chief clinical product officer, Nabla (moderator)
These were three key insights from the session:
1. Health systems need AI tools that understand and facilitate physicians’ work
Such tools must integrate into existing clinical workflows, have high clinical accuracy through prior validation and “understand” clinical context — not just execute automation for the sake of automation. “As AI becomes more embedded into clinical workflows, there’s a risk that complex clinical reasoning gets reduced to if/then automated things,” Dr. Sakumoto cautioned.
Effective AI tools must also encompass multiple workflows, allowing clinicians to do everything from a unified interface without switching between screens, which can add cognitive burden.
2. Care coordination remains a major challenge; AI tools that help orchestrate care will be valued
Tasks most in need of orchestration include improving patient intake, avoiding delays in care follow-up, ensuring that patients do not access lab results before their clinician and enabling better communication within and between care teams.
“These are opportunities for AI,” Dr. Kolade said. He noted that addressing these gaps goes beyond simple automation and requires identifying who should work next on a particular task — which is the essence of longitudinal care orchestration. “The difference is in ownership and timing,” he said.
Orchestration is also about considering how automating certain tasks affects other processes downstream. For example, the University of Toledo has a process that evaluates the impact of automation in the pre-registration and prior-authorization phases of the revenue cycle and financials, Dr. Sadeghian pointed out.
The best AI tools and implementations have a positive effect on both the patient and provider experience by making it possible that “while the total number of minutes [in a patient-provider encounter] isn’t changed, the value of those minutes is improving,” Dr. Sakumoto said.
3. AI governance and strong vendor partnerships are crucial to success
This involves oversight of AI tool implementation, input from multiple departments, prioritization of key initiatives, and in some cases, designation of departmental leaders.
For example, UTHealth is developing a governance model that enlists departmental technical leaders — beyond the IT department — to manage AI implementations. At the University of Toledo, a similar system appoints “trusted advisors” from each department to formally propose and oversee AI projects.
Beyond internal governance, organizations should carefully evaluate potential vendors for each vendor’s capacity to partner over the long term, not just sell a product. “This is a conversation we normally have in the beginning [of vendor selection], especially when testing out new products,” Dr. Sadeghian said. “As the providers bring feedback, we compile it, communicate it to the vendor and wait to see if the change is made and when.”
At the Becker's 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health, taking place September 14–17 in Chicago, healthcare executives and digital leaders from across the country will come together to explore how AI, interoperability, cybersecurity, and revenue cycle innovation are transforming care delivery, strengthening financial performance, and driving the next era of digital health. Apply for complimentary registration now.
