Examining Art Units To Avoid Subject Matter Eligibility Challenges For Bioinformatics And AI-related Patents – Intellectual Property

Overview of Subject Matter Eligibility Challenges

Computer-based inventions – especially in the machine learning
(ML), bioinformatics, and artificial intelligence (AI) fields – are
susceptible to subject matter eligibility challenges. Subject
matter eligibility challenges may prevent a patent application from
being granted by the United States Patent and Trademark Office
(USPTO) and may even be asserted to invalidate a patent post-grant.
Subject matter eligibility challenges include categorizing the
computer-related invention as an abstract idea, which includes
mental processes (concepts capable of being performed on pen and
paper), methods of organizing human activity (such as managing
interactions between persons), and mathematical concepts.

In recent years, the Federal Circuit has implemented a
multi-step test to determine whether patent claims would survive a
subject matter eligibility challenge. In some cases, the Federal
Circuit has upheld the validity of claims reciting an abstract idea
when the claims integrate the abstract idea into a practical
application or recite additional elements that amount to
significantly more than the abstract idea. Similarly, the USPTO
implements a similar multi-step test with various guidelines to
determine whether patent claims in an application will survive a
subject matter eligibility rejection. Nevertheless, 101
jurisprudence remains unsettled in regards to the key determining
criteria and is continually evolving with new guidelines and
emerging case law. As a result, the fate of bioinformatics and
machine learning-based patents and patent applications remains
uncertain.

As we previously discussed in Patenting Considerations for Artificial
Intelligence in Biotech and Synthetic Biology Part 1,
computer-based applications for inventions covering a gamut of life
science disciplines – from sequencing and functional genomics to
drug design, discovery, and testing – have realized tremendous
benefits due to the use of machine learning and AI. But the patent
rights protecting these advanced inventions are susceptible to the
same subject matter eligibility vulnerabilities mentioned
above.

Subject Matter Eligibility Challenges in AI and
Bioinformatics

Many ML, bioinformatics, and AI patents face an uphill battle
for patentability due to the use of computer systems and
algorithms, and the rapidly evolving law surrounding subject matter
eligibility. These computation-heavy areas face subject matter
eligibility challenges, especially for their ML features or
proximity to mathematical calculations.

For example, the Federal Circuit issued a decision that
manifests the troubling eligibility landscape in the fields of
machine learning and bioinformatics. In In re Board of Trustees
of the Leland Stanford Junior Univ., No. 2020-1288 (Fed. Cir.
Mar. 25, 2021), the Federal Circuit affirmed the rejection of the
patentability of claims directed to computerized methods to
generate genetic data. Here, Stanford sought to patent claims
directed to a “computerized method for inferring haplotype
phase in a collection of unrelated individuals” that created
“new data” with the use of specific rules and machine
learning techniques. These machine learning techniques included
steps such as building a data structure describing a Hidden Markov
Model, repeatedly randomly modifying at least one of the imputed
initial haplotype phases, and automatically replacing an imputed
haplotype phase. Id. at 4-5.

The Federal Circuit held that the claims did not “involve
practical, technological improvements extending beyond improving
the accuracy of a mathematically calculated statistical
prediction.” Id. at 10. Additionally, the Federal
Circuit found that “the recited steps of receiving,
extracting, and storing data amount to well-known, routine, and
conventional steps taken when executing a mathematical algorithm on
a regular computer,” and the claims recite generic computer
components that were in no way “specialized.”
Id. at 12. As such, Stanford’s patent application for
its computerized method for inferring haplotype phase did not get
granted as a patent. At the time of publication, the court
proceedings have terminated for this patent application.

As previously discussed here and here, there are a number of ways to mitigate
the likelihood of a subject matter eligibility rejection. For
example, an effective drafting technique includes adding explicit
descriptions in the specification that identifies specific
industries or applications where the AI may be particularly useful
and explaining AI’s advantages over existing systems and
processes.

In this article, we discuss another technique that includes
drafting a patent application for placement into a specific art
unit, which is possibly one of the most important drafting
considerations that few patent practitioners think about. This
article provides insight to USPTO art units to reduce the
likelihood of abstract idea attacks in the first place for your ML,
bioinformatics, AI, and computational patent.

Art Units Covering Technologies Related to AI and
Bioinformatics

The USPTO assigns each U.S. patent application to one of many
art units, which are organizational units of technology subclasses.
Some art units at the USPTO may behave more aggressively than
others in asserting subject matter eligibility challenges.
Let’s look at art units 1631 and 2129 to compare the
aggressiveness of different art units and their assertiveness
relating to subject matter eligibility challenges.