Judge Albright Grants Motion to Dismiss AI Patent Claims on § 101 Grounds

In Health Discovery Corp. v. Intel Corp., Judge Alan Albright granted, in part, Intel’s motion to dismiss infringement claims of AI machine-learning patents on § 101 grounds. The court did so without prejudice, however, noting the difference between “failing to plead adequate facts addressing the analytical steps called for in Alice” and proving ineligibility by clear and convincing evidence.

Health Discovery Corp. (Health Discovery) accused Intel of infringing its patented machine-learning method of generating datasets using support vector machines (SVM) and recursive feature elimination (RFE). Intel filed a 12(b)(6) motion to dismiss, arguing patent ineligibility under § 101. In an opinion lamenting the difficulty of applying § 101 jurisprudence consistently and coherently, the court applied the two-step framework for determining patent eligibility as established by the Supreme Court in Alice Corp. v. CLS Bank International, 573 U.S. 208 (2014).

Under step one of Alice, the court assessed whether the claims are directed to an abstract idea. At the outset, the court discussed the difficulty of reconciling the Federal Circuit’s jurisprudence on Alice step one and extensively reviewed recent relevant Federal Circuit cases—In re Board of Trustees of Leland Stanford Junior University, 991 F.3d 1245 (Fed. Cir. 2021) (Stanford II); SAP America, Inc. v. InvestPic, LLC, 898 F.3d 1161 (Fed. Cir. 2018); CardioNet, LLC v. InfoBionic, Inc., 955 F.3d 1358 (Fed. Cir. 2020); and Koninklijke KPN N.V. v. Gemalto M2M GmbH, 942 F.3d 1143 (Fed. Cir. 2019). If only Koninklijke and CardioNet applied, the court observed that the claims would pass Alice step one. However, the inconsistency in § 101 jurisprudence required the court to look to cases on patents most similar to the patents at issue. The court, therefore, relied on SAP and Stanford II in its analysis because they analyzed patents most similar to those at issue here.

The Health Discovery patents’ written descriptions describe a way of improving a mathematical analysis, which the court found analogous to the claims in Stanford II and SAP. In those cases, the generated patents improved data when compared to previous methods. But the Federal Circuit found these patents to be a mere improvement to the abstract idea of data analysis and calculation. In this case, the court observed that the patent uses SVM‑RFE to rank and eliminate features within datasets, which is not the conventional industry method. While this different method may be novel, the court ruled that it is still an abstract mathematical technique that improves data quality, as in Stanford II and SAP. Additionally, Health Discovery conceded that SVM is an abstract, mathematical idea. Health Discovery added RFE, which is also an abstract idea, to SVM. The court found that the resulting SVM‑RFE combination of two abstract ideas is still an abstract idea.

Health Discovery argued that its claims are patent eligible because the claims increase computational speed. The court was not persuaded by this argument. Citing Intellectual Ventures I LLC v. Capital One Bank (USA), the court applied the established precedent that an increase in speed through added computer functionality does not render an abstract idea patent eligible.

At Alice step two, the court analyzed the claims for an inventive concept that could transform the abstract idea into patent eligible claims. Health Discovery argued that the patents have an inventive concept because they improve data quality when compared to the conventional methods. But the court referred to the Federal Circuit’s holding in Stanford II that an abstract idea does not become patent eligible merely because a “different combination of mathematical steps” produces better data when compared to the prior art. Health Discovery also argued that SVM‑RFE has a variety of applications, as demonstrated by the fact that the initial paper describing SVM‑RFE has been cited thousands of times. But, relying on SAP, the court dismissed this argument, noting that even a groundbreaking mathematical idea can be patent ineligible. The court acknowledged that the claims could be patent eligible at Alice step two if Health Discovery could identify a nonabstract application of the innovation, but Health Discovery failed to do so.

As a result of the Alice analysis, the court dismissed Health Discovery’s claims under § 101, but did so without prejudice. The court ruled that dismissal with prejudice was not justified in this case because Health Discovery simply failed to plead adequate facts for the Alice analysis. The court explained that dismissal with prejudice is justified only where the defendant proves that the claims are patent ineligible by clear and convincing evidence. The court also denied as moot Intel’s motion to dismiss for failure to sufficiently plead direct and indirect infringement under 12(b)(6).