A split panel of the Patent Trial and Appeal Board recently upheld an AI (artificial intelligence) patent after it was previously ruled invalid nine months earlier by U.S. District Judge Alan Albright of the Western District of Texas. The patent, U.S. 7,542,959, is directed towards using Support Vector Machines (SVM) and Recursive Feature Elimination (RFE) for identifying genes underlying certain medical conditions. In the ruling, the PTAB determined that Intel Corp. failed to show that the six claims it challenged in the patent owned by Health Discovery Corp. were obvious.
The panel’s 2-1 split decision was handed down on September 12, 2022 and comes nine months after Judge Albright held that the patent, along with three others, were invalid under the U.S. Supreme Court’s decision in Alice v. CLS Bank, which held that abstract ideas are not patentable subject matter under Section 101.
The central issue before the Board was whether the proposed prior art combination disclosed a key element of Health Discovery’s patent, ranking features according to corresponding values.

Fig. 36 of U.S. Patent 7,542,959, Feature Selection Method Using Support Vector Machine Classifier
Intel had argued that skilled artisans would have been motivated to combine elements of an earlier patent and two technical publications to arrive at the claimed invention. While the majority found that the prior art references could be combined, it said that Intel’s evidence showing that a skilled artisan would have been motivated to do so was lacking.
“We find petitioner’s evidence and reasoning demonstrates nothing more than a skilled artisan, once presented with the separate pieces of highlighted information … may have understood that they could be combined in the manner claimed,” the majority held (emphasis supplied).
Administrative Patent Judge Garth D. Baer wrote a three-paragraph dissent. Judge Baer disagreed with the majority’s analysis and conclusion that Intel failed to show a lack of motivation to combine the prior art. Judge Baer further stated that he believed that Intel met its burden under the 2007 decision by the U.S. Supreme Court in KSR International v. Teleflex, which rejected a rigid test for obviousness.
KSR held that “when a patent ‘simply arranges old elements with each performing the same function it had been known to perform’ and yields no more than one would expect from such an arrangement, the combination is obvious.”
Judge Baer said that was the case here, agreeing with Intel “that the claimed invention is an obvious combination of known techniques applied to a known device, yielding only predictable results and thus obvious under KSR’s framework.”
In the parallel suit in the Western District of Texas, Judge Albright found that the claims in the patent are directed to an abstract mathematical concept and lacked an inventive concept that would transform them into something patent eligible under Alice. Claim 1 of the asserted patent is exemplary:

Claim 1 of U.S. Patent 7,542,959, Feature Selection Method Using Support Vector Machine Classifier
Health Discovery appealed Judge Albright’s decision in January, but moved to dismiss the appeal in April, telling the court it had filed a new complaint against Intel in Judge Albright’s court with new allegations it believes will overcome the rationale under which the previous suit was dismissed.
FINAL WRITTEN DECISION: Intel Corp. v. Health Discovery Corporation, PTAB, IPR2021-00552, Patent 7,542,959 B2, Final Decision, September 12, 2022
+++++
Richard A. Catalina, Jr. is a senior partner and Chair of the Intellectual Property and Complex Litigation Department of Jardim, Meisner and Susser, P.C. Mr. Catalina specializes in litigating patent, trademark, trade secret and related Lanham Act disputes, and inter partes proceedings before the Patent Trial and Appeal Board. Mr. Catalina has been practicing law for more than 33 years and has litigated more than one hundred matters in federal and state courts across the U.S, as well as inter and ex partes matters before the Trademark Trial and Appeal Board and the Patent Trial and Appeal Board.