AI in Healthcare: The Growing Promise - and Potential Risks

Dr. Eric Liederman of Kaiser Permanente on Balanced Approaches to AI
AI in Healthcare: The Growing Promise - and Potential Risks
Eric Liederman, medical informatics and national privacy and security leader, Kaiser Permanente

Exciting advancements in medicine through the use of AI are already happening, and many more are in the pipeline. But they need to be approached carefully and vetted properly for risk, said Dr. Eric Liederman, medical informatics and national privacy and security leader at Kaiser Permanente.

Some of the greatest AI developments in healthcare so far are in medical imaging, he said.

"With mammography, there have been recent studies - and we've been doing a lot of work in our organization, as well - where we're reaching a point where the artificial intelligence machine learning tools may be as good as replacing one out of a pair of radiologists," he said.

"So, AI - plus a radiologist - may be as good as two radiologists reading images together," he said. "That's amazing especially because we have a workforce shortage," he said in an interview with Information Security Media Group.

Other encouraging use cases for AI in healthcare include tools that can alert clinicians when patients are about to take a turn for the worse.

"For some of the sickest patients - those with sepsis, an infection of the bloodstream infecting the whole body - we want to identify those people as soon as possible because they tend to get really sick, really fast, and oftentimes die," he said.

Also, "in general, being able to identify patients in the hospital that are going to have a code - such as a cardiac arrest or other life-ending event, in the near future. We want to identify those as early as possible, even hours before it happens. We have increasingly powerful tools that allow us to do all those things," Liederman said.

With large language models, organizations also can tackle challenging, time-consuming but necessary work in healthcare, such as generating notes after patient encounters and visits and creating messages to patients. "These are really important, and we have to get these right," he said.

"Large language models hold promise. But like anything else, all this artificial intelligence also holds the possibility of peril," he said.

"What if the AI misses cancer, or flags it where it isn't cancer? Similarly, with septic patients or patients that are going to code," he said. "Therefore, it is incumbent on us not only to focus on the promise but also the risks and to mitigate them."

In this interview with Information Security Media Group (see audio link below photo), Liederman also discussed:

  • Ways to avoid having AI drive shadow IT within large organizations;
  • Risks involving AI used against healthcare entities for cyberattacks, spear-phishing campaigns, deepfakes and other security compromises;
  • Deterring insider privacy breaches involving patient records - and the dramatic effect health data breaches have had on patients' trust in and transparency with their healthcare providers.

Liederman, an internal medicine physician, serves as director of medical informatics for The Permanente Medical Group and national leader of privacy, security and IT Infrastructure for The Permanente Federation. In these roles at Kaiser Permanente, which serves over 12 million members across the U.S., Liederman is accountable for privacy and security, IT investment, large program governance and IT infrastructure delivery and resilience.

Around the Network

Our website uses cookies. Cookies enable us to provide the best experience possible and help us understand how visitors use our website. By browsing, you agree to our use of cookies.