Skip to main content
AI, machine learning in health care need diverse datasets

Artificial intelligence and machine learning platforms that do not include demographically representative data could produce biased results and reinforce existing clinical biases, biomedical ethics and medicine professor David Magnus said in the keynote address at a radiology oncology association meeting. In a moderated discussion after the address, machine learning and health care professor Suchi Saria said algorithms must be designed more holistically to account for biases.

Full Story: