Patent-Eligibility Considerations & Prosecution Strategies for AI-Based Life Sciences Patent Applications

This post is part of MoFo's 2025 Intersection of AI and Life Sciences blog series. In this blog series, we explore how artificial intelligence is revolutionizing research, innovation, and patient care in the life sciences. Stay tuned for expert insights regarding the impact of AI on intellectual property, licensing, contracts, regulatory policy, enforcement, privacy, and venture markets in life sciences.

I. Inventions at the intersection of AI and Life Sciences

Based on recent, dramatic advances in its predictive and generative capabilities, artificial intelligence (AI) is poised to revolutionize a variety of fields, including business, finance, manufacturing, information technology, and healthcare, with the potential to transform the global economy.1,2 AI is also increasingly being utilized to tackle challenging problems in the life sciences.3 Examples include the use of machine-learning (ML) models for prediction of 3D protein structures and protein-protein interactions based on input amino acid sequences, optimization of genetically engineered protein sequences for improved expression in host cells, in silico drug design, design of metabolic pathways to synthesize a drug based on a specified precursor, detection of subtle changes in cellular and/or tissue morphology in digital pathology images, and analysis of medical record data.

As a consequence, the number of AI-related patent applications granted globally (including those that straddle AI and life sciences) had skyrocketed to over 62,000, as of 2022.4 However, along with patent applications directed to software in general, patent applications directed to AI-based inventions can sometimes run into subject-matter eligibility rejections under 35 U.S.C. § 101 at the USPTO, with recent court decisions holding that claims directed to "generic machine learning" technologies are directed to abstract ideas (see, e.g., Recentive Analytics, Inc. v. Fox Corp., No. 2023-2437 (Fed. Cir. Apr. 18, 2025), and a related blog). Thus, it is imperative for patent practitioners, in-house IP counsel, and inventors to stay up to date on relevant USPTO guidelines and case law relating to subject-matter eligibility.

II. Recent USPTO guidance on subject-matter eligibility:

In an effort to improve and standardize the examination of patent applications directed to AI-based inventions, the USPTO issued an updated guidance document relating to subject-matter eligibility on July 17, 2024.5 The recent guidance is intended to supplement the USPTO's previous guidance on subject-matter eligibility under 35 U.S.C. § 101, which is summarized in Sections 2103 through 2106.07 of the Manual of Patent Examining Procedure (MPEP).

A. Overview of the USPTO's patent subject-matter eligibility guidance:

A modified flowchart that combines the two flowcharts provided in MPEP 2106, Subsection III, and MPEP 2106.4, Subsection II is presented below. The flowchart summarizes the USPTO's guidance on subject-matter eligibility under 35 U.S.C. § 101. The guidance is applicable to all statutory categories of invention (i.e., processes, machines, manufactures, or compositions of matter).

The analysis comprises four steps:

Step 1 determines whether the claimed invention falls into at least one of the four statutory categories of invention recited in 35 U.S.C. 101.

Step 2A, Prong One determines whether the claim recites a judicial exception (i.e., an abstract idea, a law of nature, or a natural phenomenon).

Step 2A, Prong Two determines whether the claim integrates the recited judicial exception into a practical application of the judicial exception (in which case the claim is eligible) or whether the claim is "directed to" the judicial exception (in which case the claim requires further analysis at Step 2B). The determination of whether the claim integrates the recited judicial exception into a practical application requires an evaluation of additional considerations, such as whether the additional elements of the claim are directed to insignificant extra-solution activity or are mere instructions to apply the judicial exception to the specified problem, or whether the claim is directed to an improvement in the functioning of a computer or an improvement to another technology.

Step 2B of the analysis evaluates whether the claim as a whole (including the additional elements) amounts to significantly more than the recited judicial exception. Although Step 2A, Prong Two and Step 2B may overlap, Step 2B includes a consideration of whether the additional elements of the claim amount to a well-understood, routine, or conventional activity. If the claim as a whole is determined to amount to significantly more than the judicial exception in Step 2B, the claim is patent-eligible. If the additional elements do not result in a claim that amounts to significantly more than the judicial exception, the claim will be rejected under 35 U.S.C. 101 as lacking patent eligibility.

B. Applicability of USPTO subject-matter eligibility guidance to AI-assisted inventions:

When considering the subject-matter eligibility of AI-based inventions, there are two particular areas for focus: (i) the evaluation of whether a claim recites an abstract idea in Step 2A, Prong One; and (ii) the evaluation of whether the claim as a whole integrates the judicial exception into a practical application of the exception in Step 2A, Prong Two.

B1. Evaluation of whether a claim recites an abstract idea (Step 2A, Prong One):

At Step 2A, Prong One, examiners must determine whether a claim recites an abstract idea (and thus requires further eligibility analysis) or merely involves-or is based on-an abstract idea. MPEP 2106.04(a)(1) provides instructions and hypothetical examples of claims that do and do not recite an abstract idea to assist with this determination. The determination of whether a claim recites an abstract idea is based on: (i) identifying specific claim limitations that the examiner believes recite an abstract idea, and (ii) determining whether the identified limitations fall within at least one of the abstract idea groups defined in MPEP 2106.04(a)(2), i.e., mathematical concepts, certain methods of organizing human activity, or mental processes. Of these abstract idea groups, the two that are most often cited in subject-matter eligibility rejections for AI-based life sciences applications are the "mathematical concepts" and "mental processes" groups.

The "mathematical concepts" group is defined as including mathematical relationships, mathematical formulas or equations, and mathematical calculations. Importantly, the guidance indicates that a claim does not recite a mathematical concept if it is only based on, or involves, a mathematical concept (see, e.g., MPEP 2106.04(a)(2), Subsection I). This suggests that when drafting applications directed to AI-based life sciences inventions, the patent practitioner should emphasize the practical application that the claimed invention addresses and, to the extent possible, include narrative that describes the concepts, models, and/or data structures used to implement the invention in words as well as in terms of mathematical equations. Similarly, the independent claims should recite the concepts, models, and/or data structures used to implement the invention in words, with specific equations relegated to the dependent claims. For example, if the training of a machine-learning model involves the use of a novel loss function, one should recite in words the relationships and parameters that are captured by the loss function rather than include just the mathematical equation itself.

The "mental processes" group includes concepts that can be performed in the human mind, e.g., observations, evaluations, judgments, and opinions. The USPTO guidance explains that "claims do not recite a mental process when they contain limitations that cannot practically be performed in the human mind, for instance when the human mind is not equipped to perform the claim limitations" (see, e.g., MPEP 2106.04(a)(2), subsection III). During drafting, the patent practitioner should include a description of why the problem being addressed is exceptionally complex on a technical level, requires evaluation of an exceptionally large number of parameters, has historically been intractable, etc. For example, in the context of an application directed to a model trained to predict three-dimensional protein structure based on an input amino acid sequence, one might include a description of how prediction of tertiary protein structure has historically been difficult due to the large number of structural permutations that can be generated based on parameters such as bond angles, bond lengths, intramolecular and intermolecular electrostatic, hydrogen bonding, and Van der Waals interactions, environmental factors, etc., and how the problem is thus too complex to solve as a mental exercise or by using pencil and paper.

B2. Evaluation of whether the claim as a whole integrates the judicial exception into a practical application of that exception (Step 2A, Prong Two):

If the claim is determined to recite a judicial exception in Step 2A, Prong One, the analysis proceeds in Step 2A, Prong Two to an evaluation of whether the claim as a whole integrates the recited judicial exception into a practical application of the exception. If so, the claim is not "directed to" the judicial exception. Integration into a practical application is evaluated by: (i) identifying whether there are any additional elements recited in the claim beyond the judicial exception(s), and (ii) evaluating those additional elements individually and in combination to determine whether they integrate the exception into a practical application of that exception. The analysis is performed based on considerations such as whether the additional elements improve the functioning of a computer or another technology, whether the claim links the judicial exception to a particular technological environment or field of use, or whether the claim includes a step that uses the judicial exception to effect a particular treatment or prophylaxis for a disease or medical condition. The analysis in Step 2A, Prong Two specifically excludes consideration of whether the additional claim elements represent well-understood, routine, or conventional activity, which is performed in Step 2B.

As noted, integration of a judicial exception into a practical application can be demonstrated by showing that the claimed invention improves the functioning of a computer or improves another technology, i.e., that the claimed invention "provides a technological solution to a [specific] technological problem." Thus, it is important to indicate that the problem to be solved by the invention is technical in nature, as well as how the claimed invention addresses the technical problem in a way that confers advantage(s) over prior art solutions. Again, in the context of an application directed to a model trained to predict three-dimensional protein structure based on an input amino acid sequence, for example, one might include a description of how the model output (i.e., the predicted three-dimensional structure of the protein) is used in one or more downstream applications (e.g., to identify receptor-ligand binding sites, to evaluate protein-protein binding interaction probabilities, etc., to facilitate drug design and development). One might also include a description and examples of data that illustrate the performance of the trained model in terms of one or more performance metrics (e.g., comparisons of predicted three-dimensional protein structures to known crystal structures for selected test cases, comparison of the predicted solvent accessibility of amino acid residues to experimentally determined solvent accessibility data, comparison of predicted ligand binding affinities or protein-protein binding interaction probabilities with experimentally determined binding affinity data, etc.), and how the performance of the claimed model compares to that of prior art prediction models.

The recent USPTO guidance emphasizes that "A key point of distinction to be made for AI inventions is between a claim that reflects an improvement to a computer or other technology described in the specification (which is eligible) and a claim in which the additional elements amount to no more than (1) a recitation of the words "apply it" (or an equivalent) or are no more than instructions to implement a judicial exception on a computer, or (2) a general linking of the use of a judicial exception to a particular technological environment or field of use (which is ineligible)." That is, vaguely worded claims directed to, e.g., applying a specified machine-learning model (e.g., an artificial neural network) to a specified problem or field of use (e.g., protein structure or function prediction) may be held ineligible under 35 U.S.C. § 101. The patent practitioner should ensure that the claim language indicates what specific improvement has been achieved, and how technical elements of the claimed invention achieve that improvement.

III. Patent-eligibility considerations & prosecution strategies for AI-based life sciences inventions:

The recent USPTO guidance on subject-matter eligibility under 35 U.S.C. § 101 provides several considerations when drafting and prosecuting U.S. patent applications directed to AI-based life sciences inventions, including:

• When drafting the specification and claims:
• • Include a description of the technical problem to be solved and how the claimed invention (i.e., the AI-based method or system) provides a technical solution to that problem (to address potential Step 2A, Prong Two eligibility issues).
  • Include description that illustrates why the claimed invention is not directed to an abstract idea (e.g., a mental process or mathematical concept) to address potential Step 2A, Prong One eligibility issues. For example, emphasize the complexity of the problem to be solved. Explain why the invention involves performing steps that cannot be practically performed in the human mind or using pencil and paper, or that involve specific data preprocessing steps, data structures, or model training techniques that are not simply mathematical operations. Describe underlying mathematical concepts in words rather than simply using equations.
  • Include description that illustrates how the claimed invention is directed to a practical application to address potential Step 2A, Prong Two eligibility issues. Include descriptions of upstream (e.g., sample collection, nucleic acid sequencing, image acquisition, etc.) and downstream process steps (e.g., detecting sequence variants or changes in tissue morphology that serve as biomarkers, designing or manufacturing a drug, identifying or administering a treatment, operating a medical device, etc.) to help tie the claimed invention to the practical application.
  • Include description of the advantages conferred by the claimed invention compared to prior art solutions to help address potential Step 2A, Prong Two eligibility issues. Whenever possible, also include examples of data that illustrate the performance of the trained model. Include support for values of-or ranges for-model performance metric(s) (e.g., accuracy, precision, recall (sensitivity), specificity, etc.) to illustrate the advantages conferred by the claimed invention relative to prior art solutions.
  • Include independent claim(s) directed to the use of the model for inference rather than directed simply to the training of the model to avoid potential Step 2A, Prong Two eligibility issues. Describe the input data structures, relationships, models, output data structures, etc., used to implement the invention in words in the independent claim, and include dependent claims directed to specific mathematical equations where necessary.
  • Include dependent claims directed to specific model architectures (e.g., neural networks, regression models, clustering models, decision tree models, etc.), training modes (e.g., supervised, unsupervised, reinforcement learning, etc.), training techniques (e.g., gradient descent, backward propagation, the use of specific loss functions, etc.), training data components (e.g., specific combinations of input data), etc., if not already incorporated into the independent claim to avoid potential Step 2A, Prong One eligibility issues. Focus on the novel aspects of the abovementioned features, rather than reciting generic characteristics that are broadly applicable to all machine-learning models.
  • Include dependent claims directed to upstream and downstream process steps to help tie the claimed invention to the practical application to avoid potential Step 2A, Prong Two eligibility issues.
• During prosecution:
• • If a claim is rejected under 35 U.S.C. 101, schedule an examiner interview. Use the interview to provide an overview of the claimed invention to the examiner and discuss the alleged subject-matter eligibility. This is particularly important, given that the likelihood of receiving a rejection under 35 U.S.C. 101 is so examiner-dependent. Point the examiner to descriptions in the specification of the technical advantages provided by the claimed technical solution and how the outputs of the AI techniques are tied to a practical application. Ask the examiner for feedback on the claim language.
  • When drafting a response, be sure to focus arguments on the specific step(s) in the analysis of patent eligibility under 35 U.S.C. 101 that the examiner has cited as the basis for the rejection. Include separate arguments for why the claims are not directed to an abstract idea under Step 2A, Prong One, and why the claim as a whole is directed to a practical application of the alleged abstract idea under Step 2A, Prong Two.

Finally, keep abreast of changes at the USPTO and potential congressional action, including updates to guidance documents and examples. Recently, a new director for the USPTO has been nominated, which may lead to the issuance of a new guidance documents or examples by the USPTO and changes in how patent examiners vet AI inventions. Recently, in Congress, bills have been introduced that seek to change the statutory language of Section 101.

IV. References:

1. K. Georgieva (2024), "AI Will Transform the Global Economy. Let's Make Sure It Benefits Humanity."

2. C. Dilmegani (2025), "Top 120 Generative AI Applications with Real-Life Examples."

3. Colombe et al. (2023), "Opportunities for Generative AI in Biotechnology."

4. N. Maslej et al., "The AI Index 2024 Annual Report," AI Index Steering Committee, Institute for Human-Centered AI, Stanford University, Stanford, CA, April 2024.

5. 2024 Guidance Update on Patent Subject Matter Eligibility, Including on Artificial Intelligence.