Artificial intelligence use in healthcare is only as safe and accurate as the governance and trust frameworks surrounding it, said Dave Bailey, vice president at healthcare privacy and security consultancy Clearwater. That’s particularly the case in clinical environments where errors or hallucinations can directly impact patient care.
“Any autonomous use of AI in clinical decision-making can be very dangerous, and organizations that are implementing AI need to have the processes in place, have the understanding that they’ve tested the data, they trust it,” he said.
Should a nurse find an error in an AI-generated clinical summary within an electronic health record, there needs to be a process and plan in place to address the issue before a potentially dangerous medical mishap occurs.
“How do you respond to those things? It’s extremely important for organizations to understand how they’re going to use [AI] and certainly plan around those adverse events when ultimately they don’t trust the data.”
Also, it’s one thing to put governance around getting an AI system implemented, but managing governance and risks around the life cycle of the technology and its use is equally as critical, he said.
“What do you do after it’s in?” he asked. “It’s going to take the use of new emerging technologies in order to monitor the models, the outcomes and help organizations provide indicators when the model drifts, or it hallucinates.”
In the interview (see audio link below photo), Bailey also discussed:
- Evolving threats and risks involving the use of AI in healthcare;
- The use of AI in healthcare for improving cybersecurity, and the associated risks;
- Other important AI trust and governance issues for healthcare sector entities to consider.
Bailey is vice president of consulting services and strategy at Clearwater, where he leads enterprise-level cybersecurity and risk management services for healthcare organizations nationwide. He has more than 24 years of cybersecurity experience, including 14 years focused on healthcare. Bailey previously served 13 years as a communications and information officer in the United States Air Force, with leadership assignments spanning the Pentagon, domestic bases and overseas operations.
