The healthcare industry is primed for artificial intelligence to reshape the delivery of medical care, according to a group of independent scientists. But the technology can only live up to its hype if AI algorithms have access to high-quality data sets.
In a report (PDF) commissioned by the Office of the National Coordinator for Health IT (ONC) and the Agency for Healthcare Research and Quality (AHRQ), an advisory group of scientists and academics known as JASON acknowledged the significant hype surrounding AI. But they also pointed to a “confluence” of forces that are likely to drive AI adoption, including frustration with legacy systems, widespread use of networked devices and the public’s broader acclimation to services like Amazon that emphasize convenience.
RELATED: Stanford researchers: Bring expectations for artificial intelligence back down to earth
“Most importantly, the report indicates that the use of artificial intelligence in health and healthcare is promising—and doable,” officials with ONC, AHRQ and the Robert Wood Johnson Foundation wrote in a blog post.
Wrangling the massive amounts of data generated by health IT systems, and integrating new data streams, will be critical to AI’s success in healthcare. JASON raised specific concerns about the quality and availability of relevant EHR data, along with the lack of interoperability across the industry.
Industry experts have previously highlighted the need for greater access to easily digestible health data for AI to make a bigger impact on clinical care.
“If EHR data are to be used to support AI applications, understanding this quality, and how AI algorithms react given the quality issues will be important,” the report stated. “To date, very little research has looked at this issue.”
RELATED: In a new pharmacy initiative, Sutter Health reimagines its approach to real-time analytics
Officials with ONC and AHRQ said the agencies plan to work with the National Institutes of Health and the Food and Drug Administration to identify areas where AI can improve research and medical care.
Although healthcare AI solutions are still in their infancy, there have been pockets of success, particularly involving imaging specialties. In the last year, Stanford researchers have built algorithms that can identify pneumonia in x-rays or instances of skin cancer better than trained medical professionals. Meanwhile, Harvard researchers have called for updates to the medical education system to incorporate analytics and digital tools.