The Food and Drug Administration wants on how to overhaul approval of medical device software to better incorporate artificial intelligence tools.
Outgoing FDA Commissioner Scott Gottlieb Tuesday that currently any changes made to software in a medical device requires FDA approval, which doesn't work for AI.
"With artificial intelligence, because the device evolves based on what it learns while it's in real-world use, we're working to develop an appropriate framework that allows the software to evolve in ways to improve its performance while ensuring that changes meet our gold standard for safety and effectiveness," Gottlieb said in a statement.
The agency has approved an AI-based device for detecting an eye disease called retinopathy. Another device uses AI technology to alert providers of a possible stroke in patients.
However, these devices use "locked" algorithms that are updated by the manufacturers to incorporate new data. The FDA doesn't want future AI-incorporating devices to require those kinds of updates.
"For example, an algorithm that detects breast cancer lesions on mammograms could learn to improve the confidence with which it identifies lesions as cancerous or may learn to identify specific subtypes of breast cancer by continually learning from real-world use and feedback," Gottlieb said.
This is the latest approach by the agency to improve its regulations on digital health. Last year, the agency put out regulations for digital products that encourage prescription medication adherence.
Gottlieb's announcement comes about a week after the CMS announced a $1.65 million challenge to develop understandable AI products for healthcare.
A key requirement of the challenge is that the product has to be understandable by doctors to ensure it is actually used.
However, developing and approving new products aren't the only barriers to adoption of AI in healthcare. A major hurdle is gathering enough real-world, trustworthy data for an AI device to learn and help influence clinical decisions.
The CMS and Office of the National Coordinator for Health IT put out proposed rules in February that seek to implement open data-sharing technology and prevent hospitals from blocking information.