Cybersecurity is a key risk factor for emerging applications of artificial intelligence in medical devices. Products that fail to address security risks may remain vulnerable to cyberattacks. Malicious attackers are targeting hospitals and healthcare systems, especially through ransomware, credential harvesting, or device theft. According to IBM’s report, the average cost of a medical data breach is now $10.1 million per incident, with the healthcare industry having the highest data breach cost of any industry surveyed in the report. AI brings new potential responsibilities to the field as researchers show that AI models can be fooled using falsified medical images. Industry players should be aware that while the FDA has generally treated the healthcare sector as a victim when it comes to cyberattacks, regulators may be adjusting their views. In 2022, the FDA released draft cybersecurity guidance stating that it will withhold premarket authorization or approval of medical devices if cybersecurity is not properly addressed. To mitigate these risks, companies may consider adversarial training of his AI models. Pre-generating adversarial data and teaching the model that this data is being manipulated can make the device’s diagnostic tools more robust.
This post is part of a series on trends in artificial intelligence for 2023 written by MoFo attorneys.
[View source.]
