Califf Sees Adaptive Algorithms, Capable of Change, in Medical Devices

FDA Commissioner Robert Califf wants to push the agency toward approving AI assisted medical devices that contain algorithms capable of learning, adapting and changing as they gather and analyze more data — which is not the technology in the AI-assisted devices FDA has approved thus far.

“This idea of fixed algorithms that don’t change is probably not a good idea for most of the exciting areas that are coming up,” said Califf in an interview at the Consumer Electronics Show in Las Vegas on Wednesday. “If you have a chance for algorithms to learn and it’s supervised correctly, it will get better and better over time.”

Algorithms that use fixed or locked algorithms that can’t pivot in response to new data but can only use the data they were trained on. The advantage that adaptive algorithms offer lies within their ability to learn continuously and self-improve from real-world experience without input from their developers.

To read the whole story, click here to subscribe.

Related Topics