FDA Needs Help in Regulating Algorithms: Califf
Speaking at the Consumer Technology Association’s “Ask A Commissioner” program at last week’s Consumer Electronics Show in Las Vegas, FDA commissioner Robert Califf said the agency can’t be solely responsible for effectively regulating products with adaptive algorithms that can learn and change. “If you have an algorithm that can learn and is supervised correctly, it will get better over time,” he said. “It will be better than a fixed algorithm. We have to be able to regulate them effectively.”
He said there is a need to be able to follow people over time to see if the algorithm gets better or worse. “FDA can’t do that alone,” he said. “We need a community of entities that do assessments and certify whether an algorithm is doing good or harm.”
Asked about changes in technology between his time as commissioner in the Obama administration and now, Califf said today is “a different world. The changes are so dramatic, it’s hard to characterize them.” He said it’s now possible to see a time when healthcare will be guided much more by artificial intelligence (AI). “We have a great chance for good,” he said, “but also a real chance to harm people.”
Reflecting on his visit to the show’s exhibit area, Califf said he was very excited about platforms being developed for use in homes, operating rooms, and clinics.
Asked how FDA will try to keep biases from creeping into algorithms, Califf said that should always be part of the evaluation of any AI tool.
He then took the opportunity to take a much broader view of the problems and opportunities. “The U.S. health system is structurally designed to advantage people with money and power,” he declared. “AI is only as good as the system in which it is deployed.”
Califf said one of the greatest opportunities in AI is to “get the gatekeepers out of the way who right now are suppressing a lot of information. There’s a reason why the U.S. is at the bottom of the list of high-income countries in terms of life expectancy. We’re inventing all the technology that’s fueling the world, and other people are using it better. A better analysis of data would show why, but we’re not seeing that. It’s not getting to the primary needs.”
He continued that the country is not putting health clinics where there is the greatest need, but rather where the profit is the highest. “We have a lot of work to do if we want to get bias out of the way,” he concluded.