The U.S. FDA’s approach to bias covers a large swath of territory, including the potential for bias to creep invisibly into artificial intelligence (AI) products. Yarmela Pavlovic, vice president for global regulatory affairs at Medtronic plc, said at this year’s Artificial Intelligence Summit that regulators may be more wary of the potential hazards of bias in AI compared to non-AI software simply because of the difficulty in anticipating how bias might affect the function of these advanced algorithms.

The America Invents Act of 2011 was designed to provide a durable overhaul of the U.S. patent system, but the inter partes review (IPR) process has drawn fire from inventors as a patent-killing machine. The Senate is considering a new bill to address some of these concerns, but witnesses at a hearing this week were anything but united in their assessment of the status quo, making it difficult to forecast the fate of this latest effort at patent reform.

A number of recent developments in artificial intelligence (AI) have sent some reassurance that these algorithms will not hit the market completely devoid of regulation, but a Nov. 8 hearing in the U.S. Senate makes clear that Capitol Hill is intent on legislating on AI, even if only belatedly.

As reported ipreviously in coverage by BioWorld, the U.S. FDA’s latest guidance on cybersecurity elevates the agency’s demands for medical device cybersecurity, but the agency advised industry in a recent webinar that hospital IT systems are fraught with cybersecurity hazards of their own, and thus device makers should view these IT systems as potentially hostile environments where cybersecurity is concerned.