On Monday, April 27, 2020, the USPTO published a decision that claims that a machine cannot be an inventor for the purposes of obtaining patent protection. Rather, an inventor must be human or “natural person.”
This decision reminded me of a discussion I had earlier this year when I participated on a conference panel in Dubai to explore patent issues in the new and exciting field of AI. My focus specifically was AI as it applies to biotechnologies, pharmaceutical technologies, and medical devices. While the prospect for developing AI-based technologies seems promising, obtaining valuable patent protection on such technologies can be challenging.
Here I will examine some of the challenges presented in patenting AI in the life sciences field as well as discuss five strategies that may improve the chances for patenting these new technologies.
AI can have many applications ranging from voice-powered personal assistants like Siri and Alexa to trying to identify ones of the most infamous serial kills of all time, the Zodiac Killer. Within the life sciences field, AI has so far been used to detect cancer in radiology, improve accuracy of CRISPR technology by predicting and preventing CRISPR from accidentally editing genomic regions similar to target region, generate and validate new drug compounds, and even to calculate the impact of various treatments on brain tumors by analyzing MRIs.
Because of its unique nature, AI-based technologies in life sciences face problems from both the AI side as well as the life sciences side. From the AI side, Section 101 patent eligibility issues, namely Alice, and Section 112 written description present major hurdles. In addition, there are questions stemming from DDR and Enfish about patenting software. From the life sciences side, Section 101 patent eligibility, namely from Myriad, Section 112 written description, and issues related to inherently, obviousness, and double patenting all present hurdles. Navigating these hurdles individually can be challenging by itself but navigating multiple at the same time can be even more so.
Moreover, the past several years have see uncertainties in court decisions and USPTO guidances and thus the eligibility criteria for AI-based technologies is unclear. This uncertainty stems from the Mayo and Alice cases.
So what do you do when you try to patent a technology with so many uncertainties? Below are five strategies that can help
First, narrow the claims to a specific product. This is not the time to try to carve as much white space for yourself as possible. Focus on the product and make sure the claims cover that product. As the law develops around AI, broadening the scope of such claims will likely be possible but until then, patent eligibility and written description are hurdles that, more often than not, will demand a more tailored approach to claim drafting.
Second, while you want to narrow your claim to cover your particular product, do not just rely on one claim to do so. Instead, draft multiple claims with different claim scopes. This will give you the best chance of not only getting a claim allowed, but also potentially allowing some claims to withstand a post-grant challenge should one arise. In my book, Billion Dollar Patents, I discuss the importance of having varying claim scopes as a way to increase the likelihood that some variations could withstand challenges even when others may not.
Third, draft the specification to include as much detail as possible about the invention, specifically about its novel and nonobvious features. During prosecution of the application, you may need to rely on these features to amend the claims and include limitations. Since these types of inventions, by their nature, present written description challenges, the more information you provide about its features, the better the chances to overcome Section 112 hurdles.
Fourth, draft the specification to show how the invention provides a solution to a problem. This is a follow up to the previous point about providing as much detail as possible about the invention. If you can show that the invention solves a problem, the invention becomes more concrete because it has a purpose. In addition, the 2018 European patent eligibility guidelines require that the applicant “establish a causal link to the technical purpose,” so following the guidelines from a country that is more evolved in its eligibility guidelines can be helpful.
Finally, choose the jurisdictions in which you want to patent in carefully. You may want to first protect in countries where the laws are more clear, such as in Europe, and then enter other countries, including the U.S., when the rules are still being developed. Moreover, you can avoid countries where it does not seem that AI will be allowed at all and focus time and effort on those which could allow it.
Overall, AI presents a lot of exciting opportunities in the field of life sciences and patenting technologies in this space presents some unique challenges. Until the rules surrounding patent eligibility become more clear, drafting AI-based applications with a focus on providing as much detail about the invention as possible, narrowing the claims to the invention, and looking for guidance from more established AI jurisdictions can improve the likelihood of of obtaining patent protection at home.
BioPharma Law Blog posts updates and analyses on IP topics, FDA regulatory issues, emerging legal developments, and other news in the constantly evolving world of biotech, pharma, and medical devices.