Reed Smith LLP

International Comparative Legal Guides’ fifth edition of its Digital Health guide

Assessing Product Liability and Related Considerations for Artificial Intelligence/Machine Learning-Enabled Medical Devices

Subscribe
Home Perspectives Assessing Product Liability and Related Considerations for Artificial Intelligence/Machine Learning-Enabled Medical Devices
As technology continues to advance the provision of healthcare to patients, more and more medical devices are incorporating Artificial Intelligence (“AI”), including a subset of AI known as Machine Learning (“ML”). AI has been defined as the science and engineering of computer systems capable of performing tasks that historically required human intelligence. Generally speaking, AI works by ingesting large amounts of data, analysing that data for patterns, and using those patterns to make predictions about future states. AI relies on various techniques, including models based on statistical analysis of data, expert systems and ML. ML is a branch of AI focused on building software algorithms that learn from and act on data. Generally speaking, an algorithm is a process that takes given inputs, and following defined rules, produces an output. Software developers use ML to create an algorithm that is “locked”, so that it provides the same result each time the same input is entered, or “adaptive”, so its behaviour has the ability to change over time using a defined learning process. For example, when Netflix recommends programmes to a user, it does so based on ML algorithms that analyse various factors, such as the user’s viewing history, preferences and behaviour.

With respect to medical devices, AI/ML can be used to glean insights from the extensive amount of data brought about during the daily delivery of healthcare. To date, the United States Food & Drug Administration (“FDA”) has cleared nearly 700 algorithms employing AI/ML. Examples include cardiac ultrasound software that uses AI to guide users, wearable technology for remote patient monitoring, radiology software that helps interpret CT scan images, software that generates 3D-printed models to better plan surgery, and doctor-prescribed video game treatment for children with ADHD. These and similar devices offer potential large-scale benefits to the provision of healthcare, including greater efficiency, improved patient outcomes, improved collection of meaningful physiological data, and an increased ability for healthcare providers and patients to consistently monitor for and detect health issues.

Regulatory bodies across the world recognise that the traditional regulatory framework for medical devices is inadequately equipped to effectively regulate AI/ML-enabled devices. As a result, FDA and other regulatory bodies have been developing new and/or additional frameworks for such devices. This chapter will first discuss FDA’s traditional regulatory framework for medical devices. It will then provide an overview of some of the recent developments in the AI/ML-enabled medical device regulatory space in the United States and elsewhere. Lastly, the chapter will identify potential product liability implications for manufacturers of AI/ML-enabled devices.

To continue reading this chapter, please click to view the PDF linked in the "Attachments" section below. Alternatively, you can visit the International Comparative Legal Guides (ICLG) website, where this chapter was first published in ICLG – Digital Health.