Webinar: How do you patent healthcare AI?
Judging by today’s news cycle, you might think that the world of IP is quite advanced when it comes to artificial intelligence (AI).
Some people, like Ryan Abbott’s team at the University of Surrey, are trying to move the debate on to whether AI itself can be listed as an inventor on patent applications.
So far, the US Patent and Trademark Office, European Patent Office, and UK Intellectual Property Office have said no, it can’t be.
But before we worry about whether AI can invent patented technologies, let’s figure out what kinds of AI can be patented technologies.
WIPR discussed some of the main issues in AI patentability, specifically in healthcare inventions, with Andrew McGettrick, patent director at HGF, in a webinar last Thursday, April 23.
As McGettrick noted, AI has been mostly grabbing the headlines in the fields of transportation for its use in self-driving cars.
That’s reflected in the World Intellectual Property Organization’s 2019 Technology Trends, which show that patent filings for transport-related AI applications rose by 143% between 2013 and 2016.
Telecommunications is in second place with an increase of 84%. But quietly taking third spot is AI in healthcare inventions, having grown by a considerable 43%.
AI has been put to use in healthcare for things like breast cancer screening, stroke detection, patient monitoring, and disease prediction.
The US has become notorious for being a bit of a minefield when it comes to patent eligibility and computer-related inventions. To what extent inventions reliant on mathematical methods and concepts can be patented is the subject of much debate, and it seems likely Congress will have to step in at some point in order to clear it up.
But what about in Europe? McGettrick explained that, under the European Patent Convention, mathematical methods and computer programs are excluded from patentability.
But these exclusions are “quite narrow and it can often be possible to work around them,” McGettrick said. The key is to place mathematical methods and computer programs in a technical context, in a way that’s “useful in the real world”.
“Very often a claim that’s couched in those terms will not face an objection that it relates to excluded subject matter,” he said.
In healthcare, the main exclusions to patentability to bear in mind are methods of treatment and diagnosis. These, McGettrick said, “can be more difficult to overcome”.
Surgical and therapeutic steps can be considered to “pollute” a patent claim, he advised.
This is primarily attributable to the policy reasons that the exclusions exist in the first place—to allow healthcare professionals to carry out their work without fear of infringing patents.
Ultimately, McGettrick said, the best way to understand what AI applications can and can’t be patented is by looking at examples and ran through several in his presentation.
Listen to the full webinar, "Patenting computer-implemented healthcare inventions in AI and machine learning", here.
LSIPR has launched a spring series of webinars in partnership with law firms on a wide range of topics including invention harvesting, second medical use, and patenting antibodies.
For more information on opportunities to participate in a webinar, contact Sarah Gooding on sgooding@newtonmedia.co.uk.
To listen to LSIPR's and WIPR's back catalogue, visit our BrightTALK channel.
Did you enjoy reading this story? Sign up to our free daily newsletters and get stories sent like this straight to your inbox.
