In a recent decision with important implications for artificial intelligence (AI) driven innovation, the Patent Trial and Appeal Board (PTAB) denied a patent for an AI-based medical tool.[1] The refusal was not because the invention was not new or inventive. Rather, the refusal was because the invention did not meet a fundamental rule of U.S. patent law. In Ex parte Michalek, the PTAB specifically acknowledged that the patent claims at issue recited new information about the nexus between certain biomarkers and the development of lung cancer as facilitated by machine learning. In fact, prior to appeal, the applicant had successfully refuted all arguments raised by the patent examiner that the invention was not new or nonobvious. That said, based on U.S. Patent Office guidance and a related example from that guidance, the PTAB still determined the claims were flawed based on the legal principle of subject matter eligibility. Although the facts in this decision concern medical health innovation, the decision is helpful to inform patent strategy for AI-enabled inventions across various disciplines and industries.
In its patent application, the applicant submitted patent claims covering a machine learning enabled capability to predict a disease state of a human based on certain biomarkers. During prosecution, the applicant had overcome all prior art rejections. Thus, the issue of novelty and nonobviousness of the claims had been specifically raised, considered, and resolved in the applicant’s favor. Patentability rested on the only remaining issue of subject matter eligibility.
Subject matter eligibility refers to whether an invention is of the required type to qualify for patent protection under U.S. patent law. Processes, machines, manufactures, and compositions of matter are patentable but natural laws, mathematical concepts, and abstract ideas, for example, are not. In practice, distinguishing between the two categories has proven difficult. Because of this difficulty and the unique complexities posed by AI driven innovation, the U.S. Patent Office has issued specific guidance on subject matter eligibility of AI-related inventions. The guidance sets forth principles that inform how patent examiners should assess whether AI-driven innovations meet subject matter eligibility requirements. To illustrate these principles, the USPTO has provided various specific “examples” demonstrating when AI-related inventions are patent-eligible and when they are not.
Although it acknowledged that the invention involved a new idea, the PTAB in Michalek found that the invention was a natural law and a mathematical concept. The PTAB relied on an example from Patent Office guidance that characterizes an invention relating to determination of patient risk for a medical condition as ineligible because it involved an improvement to an abstract idea, not to the functioning of a computer or other technology. According to the cited example, the recital of a treatment for the medical condition in theory could have helped the applicant to demonstrate the subject matter eligibility of the invention. However, the PTAB did not discuss this option and, in any event, no evidence indicated that a treatment had been described in the patent application. As perhaps more relevant, the PTAB did not discuss other examples from Patent Office guidance that might have better applied to save the invention.
While this case involved medical health technology, the implicated issues inform patent strategies for AI-enabled inventions across all industries. Patent applicants working with AI should be prepared for the Patent Office to apply similar reasoning to their applications. Patent applicants should be prepared to address strained reliance on certain examples from Patent Office guidance or, better yet, highlight more analogous examples. It is critical for patent applicants at the preparation stage to proactively devise an application drafting strategy supported by the guidance and examples to invite smoother prosecution.
FOOTNOTES
[1] Ex parte Michalek, Appeal No. 2023-004204 (PTAB Dec. 27, 2024).