How to Take a Comprehensive Approach to AI Governance
By Christopher Cheney
Providence is taking a comprehensive approach to AI governance, the health system’s chief clinical officer says.
“Our approach to the use of AI tools is methodical and anchored in our mission, values, and organizational vision and priorities,” says Hoda Asmar, MD, MBA, executive vice president and system chief clinical officer for Providence. “While we believe AI advancements have the potential to elevate quality of care and allow our caregivers to perform at the top of their license, the safety and security of our patients and their data will always be our top priority.”
HealthLeaders is conducting its AI in Clinical Care Mastermind program through December. The program brings together nearly a dozen healthcare executives to discuss their AI strategies and offerings. As part of the program, each of the panelists are talking with HealthLeaders about the use of AI in clinical care.
According to Asmar, Providence has established AI governance to get ahead of several issues raised by AI models,
“Providence proactively assembled an AI governance structure to ensure alignment around priorities and strategy, and ensure safety, privacy, security, equity, and the ethical use of AI,” Asmar says. “This governance structure will evolve as our experience and knowledge around AI deepens.”
“Providence has put together an AI guardrails workgroup led by our system’s chief data officer; an Information Protection committee led by our chief information security officer; and a Data Ethics Council, led by our chief ethicist,” Asmar says. “The work of these three teams feed into the Generative AI Leadership Council that oversees our responsible use of AI and advances our AI strategy.”
The health system has also convened expert groups to manage AI governance, Asmar explains.
“We have stood up four subject matter expert groups: clinical, patient and consumer, workforce and administration, and back office,” Asmar says. “These groups identify and prioritize key use cases for their areas and leverage the guardrails, data protection, and other governance structures to guide work to develop and implement AI solutions.”
AI models at Providence
The health system is integrating AI tools into daily work to accelerate decision-making, simplify workflows, and reduce non-clinical task burdens.
This includes internally generated innovations that leverage AI to enhance patient experience and reduce clinician burnout, according to Asmar.
“Some examples include automating non-direct patient care tasks in the form of ambient documentation; clinical decision support by bringing together multiple information sources in a way that allows the clinician to easily analyze and make decisions; and employing predictive analytics that speed up access to just-in-time data and information,” Asmar says.
ProvARIA (automated realtime in-basket assistant) organizes inbox messages to physicians and caregivers in the ambulatory setting based on acuity, urgency, and content. The AI model uses a natural language processing engine to organize the messages, and a tailored user interface integrated into the electronic health record.
“ProvARIA supports clinicians in responding to their in-basket messages by augmenting them and supporting them in understanding a patient in-basket message, triaging, and responding to the message more readily,” Asmar says.” ProvARIA has not only eased the way for our care teams but also improved the response time to patients by 50%.”
AI’s impact on Providence care teams
Some of the clinical AI tools being used at Providence are still new, and the health system is assessing how they are impacting care teams.
“Early indicators suggest these tools have a high level of engagement and satisfaction with the care teams by allowing clinicians to spend more time with their patients, reducing stress and administrative task burdens, and allowing clinicians to focus on what matters most to them and their patients,” Asmar says.
Computer-assisted physician documentation and ambient technology, in which physician-patient conversations are automatically transcribed and uploaded directly to EPIC and Providence’s EHR through tools such as Nuance’s DAX, remove the technological barrier that typically inhibits personal connection in exam rooms and allows caregivers to deliver even better care, according to Asmar.
Another example of an AI tool that is having a positive impact on care teams is an AI-powered surgical scheduling tool, according to Asmar.
“Surgical teams have indicated improved role satisfaction as surgery scheduling is easier and more efficient, and this satisfaction is also shared by the surgeons and their office staff,” Asmar says.
Christopher Cheney is the CMO editor at HealthLeaders.