UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF CALIFORNIA
Case No. 3:25-cv-01847-JWL
PLAINTIFF GENOME DX THERAPEUTICS INC.'S OPPOSITION TO DEFENDANT'S MOTION TO DISMISS COUNTS II AND III OF THE COMPLAINT
Filed: November 3, 2025 Hon. Jessica W. Langston
Attorneys for Plaintiff Genome Dx Therapeutics Inc.:
Catherine Bloom (SBN 227814), Responsible Partner Alan Ngo (SBN 261493), Billing Partner Peter Vassiliev (SBN 305872), Lead Associate Wendy Schultz (SBN 318247), Working Associate
CALDERWOOD & HARKNESS LLP 555 Mission Street, Suite 3100 San Francisco, California 94105 Telephone: (415) 772-6000 Facsimile: (415) 772-6001
TABLE OF CONTENTS
TABLE OF AUTHORITIES
Cases
Statutes and Rules
Other Authorities
I. INTRODUCTION
Plaintiff Genome Dx Therapeutics Inc. ("GenomeDx") respectfully opposes Defendant Helios Precision Diagnostics, Inc.'s ("Helios") Motion to Dismiss Counts II and III of the Complaint (Dkt. No. 38), filed September 15, 2025. Helios's motion asks this Court to dismiss, at the threshold pleading stage and before any claim construction has occurred, two counts of patent infringement directed to groundbreaking innovations in cancer diagnostics — innovations that took years and hundreds of millions of dollars to develop, that are practiced by GenomeDx's commercially successful Decipher® test, and that Helios has copied through its competing ProstaSignal™ product.
Counts II and III of the Complaint assert infringement of U.S. Patent No. 11,204,893 ("the '893 Patent") and U.S. Patent No. 11,891,647 ("the '647 Patent"), respectively. The '893 Patent is directed to genomic classifier systems and methods for post-surgical cancer prognosis that employ a defined laboratory workflow, a specific machine-learning classifier architecture, and minimum training cohort requirements. The '647 Patent is directed to improved methods for genomic risk stratification in prostate cancer that integrate genomic biomarker data with clinical variables to produce a composite risk metric — a method demonstrated during prosecution to yield unexpectedly superior predictive results compared to genomic-only models.
Helios seeks dismissal under 35 U.S.C. § 101 and Federal Rule of Civil Procedure 12(b)(6), arguing that the asserted claims of the '893 and '647 Patents are directed to patent-ineligible subject matter. Specifically, Helios contends that the claims merely recite "abstract mathematical algorithms applied to natural phenomena (gene expression levels)" and thus fall outside the scope of patentable subject matter. Helios's characterization is wrong on the law and the facts, and this Court should deny the motion for three independent reasons.
First, the asserted claims of the '893 and '647 Patents are directed to specific, concrete diagnostic improvements — not abstract ideas or natural phenomena — and satisfy both steps of the Alice Corp. Pty. Ltd. v. CLS Bank International, 573 U.S. 208 (2014), and Mayo Collaborative Services v. Prometheus Laboratories, Inc., 566 U.S. 66 (2012), framework. The claims do not preempt all uses of gene expression data or mathematical algorithms in cancer diagnostics. Rather, they recite particular sample-processing techniques applied to specific biological specimens, defined classifier architectures with enumerated training parameters, and the integration of genomic and clinical variables to produce clinically actionable prognostic outputs. These claims are analogous to those the Federal Circuit has repeatedly upheld as patent-eligible in cases such as Rapid Litigation Management Ltd. v. CellzDirect, Inc., 827 F.3d 1042 (Fed. Cir. 2016), Vanda Pharmaceuticals Inc. v. West-Ward Pharmaceuticals International Ltd., 887 F.3d 1117 (Fed. Cir. 2018), and Ostrander & Co. v. Ariosa Diagnostics, Inc., 967 F.3d 1319 (Fed. Cir. 2020).
Second, Helios's motion impermissibly conflates the abstract concept of "analyzing gene expression data" with the specific, inventive methods actually claimed by the '893 and '647 Patents. This is precisely the type of overgeneralization the Federal Circuit has warned against. The claims do not simply recite "measure gene expression" and "apply a mathematical formula." They recite, among other things, the normalization of gene-expression data from formalin-fixed paraffin-embedded (FFPE) prostate tissue samples using a reference gene panel; the use of a random-forest classifier trained on a cohort of at least 500 patient samples; the computation of a genomic risk score using a classifier comprising at least 20 gene features; the integration of that score with at least two clinical covariates selected from a defined list; and the generation and transmission of a composite risk metric. These limitations reflect the inventors' specific technological contributions and cannot be dismissed at the pleading stage as merely conventional activity without any evidentiary basis.
Third, a motion to dismiss under § 101 is premature in this case. The parties exchanged proposed claim construction terms on August 4, 2025 and identified multiple disputed terms directly relevant to the eligibility analysis, including "random-forest classifier trained on a cohort of at least 500 patient samples," "composite risk metric," and "classifier comprising at least 20 gene features." The Markman hearing is not scheduled until September 18, 2026. This Court cannot properly assess whether the asserted claims are "directed to" an abstract idea or natural phenomenon — or whether they contain an "inventive concept" — without first construing the disputed claim language. Courts in this District and across the country routinely defer § 101 challenges to the post-claim-construction stage when, as here, claim scope is genuinely in dispute.
The Complaint states plausible claims for infringement supported by detailed factual allegations. Helios's motion should be denied in its entirety. In the alternative, the Court should defer ruling on § 101 until after claim construction.
II. STATEMENT OF FACTS
A. GenomeDx and Its Patent Portfolio
GenomeDx is a Delaware corporation headquartered at 4750 Torrey Pines Road, Suite 300, San Diego, California 92130. Compl. ¶ 3. GenomeDx specializes in molecular diagnostics for cancer prognosis, with a particular focus on developing and commercializing genomic classifier tests that enable physicians to make more informed treatment decisions for patients with prostate cancer. Id. ¶¶ 3, 18–19.
GenomeDx was founded in 2011 by a team of scientists and clinicians who recognized that traditional clinical and pathological staging tools for prostate cancer were insufficient to accurately predict which patients were at the highest risk of cancer recurrence and metastasis following radical prostatectomy. Compl. ¶¶ 18–22. Over the course of more than a decade, GenomeDx's research team — including named inventors Dr. Rajan Mehta and Dr. Carolyn Xu — developed the Decipher® Prostate genomic classifier test, which uses a proprietary 22-gene expression panel to prognosticate clinical recurrence following radical prostatectomy. Id. ¶¶ 23–28. The Decipher® test analyzes gene expression data from formalin-fixed paraffin-embedded (FFPE) post-surgical prostate tissue specimens and generates a score from 0 to 1 that correlates with the probability of developing metastasis within five years of surgery. Id. ¶¶ 29–33.
The development of the Decipher® test required an extraordinary investment in research and development. GenomeDx conducted prospective and retrospective clinical validation studies involving more than 10,000 patient samples, published over 100 peer-reviewed articles validating the clinical utility and analytical validity of the Decipher® classifier, and invested in excess of $275 million in research and development from its founding through fiscal year 2024. Compl. ¶¶ 34–39. The National Comprehensive Cancer Network (NCCN) Guidelines for Prostate Cancer included the Decipher® test as a recommended tool for informing adjuvant treatment decisions beginning in 2017, and the American Urological Association (AUA) Guidelines similarly recognize the clinical utility of the Decipher® test. Id. ¶¶ 40–43.
GenomeDx maintains a substantial patent portfolio protecting its innovations. As of the date of the Complaint, GenomeDx holds 47 issued United States patents, 12 pending United States patent applications, and 19 international patent family members covering various aspects of its genomic classifier technology, including gene panels, classifier methodologies, sample-processing techniques, and integrated diagnostic reporting systems. Compl. ¶¶ 44–47. The patents-in-suit — the '893 Patent and the '647 Patent — are among the core patents in this portfolio and protect the foundational methods underlying the Decipher® test.
GenomeDx's fiscal year 2024 revenue was approximately $387.6 million, derived primarily from reimbursements for Decipher® tests. Compl. ¶ 48. GenomeDx maintains approximately 68% market share in the genomic prostate cancer prognostic testing market, reflecting the clinical community's confidence in and reliance upon the Decipher® test. Id. ¶ 49. GenomeDx obtained a Multianalyte Assay with Algorithmic Analyses (MAAA) code, specifically code 0047M, for the Decipher® test in 2018, facilitating reimbursement by Medicare and commercial payors. Id. ¶ 50. On August 15, 2024, GenomeDx submitted a De Novo classification application to the U.S. Food and Drug Administration (FDA), seeking formal regulatory clearance for the Decipher® test. Id. ¶ 51.
B. The Patents-in-Suit (Counts II and III)
The '893 Patent. United States Patent No. 11,204,893, titled "Genomic Classifier Systems and Methods for Post-Surgical Cancer Prognosis," was filed on March 22, 2019 as a continuation application of the '271 Patent family. Compl. ¶ 54; '893 Patent at [1]. The '893 Patent issued on December 21, 2021, and is set to expire on June 14, 2036. Id. The named inventors are Dr. Rajan Mehta and Dr. Carolyn Xu. Id. GenomeDx is the sole owner of the '893 Patent by virtue of assignments executed by each named inventor and recorded at the United States Patent and Trademark Office (USPTO). Compl. ¶ 55.
Count II of the Complaint asserts infringement of Claims 1, 2, 9, 14, and 20 of the '893 Patent. Compl. ¶¶ 102–128. The key independent Claim 1 recites a method comprising:
(a) normalizing gene-expression data from a formalin-fixed paraffin-embedded (FFPE) prostate tissue sample using a reference gene panel;
(b) inputting normalized expression values into a random-forest classifier trained on a cohort of at least 500 patient samples; and
(c) outputting a numerical prognostic score correlated with metastasis risk within 5 years.
'893 Patent, col. 28, ll. 14–38. The specificity of these claim limitations is critical to the eligibility analysis. Claim 1 does not recite the generic concept of "analyzing gene expression data" or "applying mathematics to biology." Rather, it recites a defined series of physical and computational steps that together constitute a specific, concrete diagnostic method: the claim specifies the type of biological sample (FFPE prostate tissue), the method of data normalization (using a reference gene panel), the architecture of the classifier (random-forest), the training requirements for that classifier (a cohort of at least 500 patient samples), and the clinical output (a numerical prognostic score correlated with a specific clinical endpoint — metastasis risk within 5 years).
Dependent Claims 2 and 9 further narrow the method by specifying the particular genes included in the reference gene panel and the minimum number of gene features used in the random-forest classifier, respectively. Compl. ¶¶ 106–112. Dependent Claim 14 recites additional sample-processing steps, including RNA extraction protocols specific to FFPE tissue. Id. ¶¶ 113–116. Dependent Claim 20 specifies that the numerical prognostic score is stratified into at least three risk categories and transmitted to a physician via an electronic health records system. Id. ¶¶ 117–122.
The '647 Patent. United States Patent No. 11,891,647, titled "Improved Methods for Genomic Risk Stratification in Prostate Cancer Using Combined Biomarker and Clinical Variable Models," was filed on September 9, 2021. Compl. ¶ 58; '647 Patent at [1]. The '647 Patent issued on February 6, 2024, and is set to expire on September 9, 2041. Id. The named inventors are Dr. Rajan Mehta, Dr. Kevin Ostrowski, and Dr. Priya Shah. Id. GenomeDx is the sole owner of the '647 Patent by virtue of assignments executed by each named inventor and recorded at the USPTO. Compl. ¶ 59.
Count III of the Complaint asserts infringement of Claims 1, 4, 6, 11, and 15 of the '647 Patent. Compl. ¶¶ 129–162. The key independent Claim 1 recites a method comprising:
(a) performing RNA extraction and expression profiling on a prostate tumor specimen;
(b) computing a genomic risk score using a classifier comprising at least 20 gene features;
(c) combining the genomic risk score with at least two clinical covariates selected from the group consisting of Gleason score, PSA level, surgical margin status, and pathological stage;
(d) generating a composite risk metric; and
(e) transmitting the composite risk metric to a healthcare provider.
'647 Patent, col. 32, ll. 5–34. Like the '893 Patent, the '647 Patent claims a specific, multi-step method with defined inputs, a defined analytical process, and a defined output. The '647 Patent goes further, however, by requiring the integration of genomic data with clinical covariates to produce a composite risk metric — a key innovation that was demonstrated during prosecution to yield unexpected improvements in predictive accuracy over genomic-only models. See infra § II.C.
Dependent Claims 4 and 6 further specify the classifier architecture and the method by which the genomic risk score and clinical covariates are combined, respectively. Compl. ¶¶ 135–142. Dependent Claim 11 recites that the composite risk metric is presented as a percentile rank relative to a reference population. Id. ¶¶ 143–148. Dependent Claim 15 recites that the method further comprises generating a recommendation regarding adjuvant radiation therapy based on the composite risk metric. Id. ¶¶ 149–155.
C. Prosecution Histories of the '893 and '647 Patents
The prosecution histories of the '893 and '647 Patents are significant because they demonstrate that the USPTO carefully examined the claims of both patents and did not find them to be directed to patent-ineligible subject matter under § 101.
The '893 Patent issued after a single office action. During prosecution, the Examiner raised a § 102 anticipation rejection based on Nakamura et al., U.S. Patent No. 9,617,340 ("Nakamura"). Compl. ¶ 65; Ex. 4 ('893 Patent Prosecution History), Office Action dated August 7, 2020, at 3–6. Nakamura disclosed a gene expression classifier for prostate cancer prognosis but did not disclose the specific random-forest classifier training methodology or the minimum cohort size of 500 patient samples recited in the '893 Patent claims. In response to the Examiner's rejection, GenomeDx's patent counsel filed a detailed response distinguishing Nakamura by pointing to these specific limitations, explaining that the claimed random-forest classifier, trained on a large, defined cohort, achieved superior predictive accuracy compared to the classifiers disclosed in the prior art. Ex. 4, Response to Office Action dated November 18, 2020, at 8–15. The Examiner was persuaded and issued a Notice of Allowance on February 3, 2021. Ex. 4, Notice of Allowance dated February 3, 2021.
Critically, no § 101 rejection was raised or sustained at any point during the prosecution of the '893 Patent. The Examiner — who had full access to the claims, the specification, and the prior art — did not identify the claims as directed to patent-ineligible subject matter. While the absence of a § 101 rejection during prosecution is not dispositive, it is relevant context, particularly at the pleading stage where Helios asks this Court to conclude, without any evidentiary record, that the claims are "plainly" ineligible.
The '647 Patent issued after a more extended prosecution involving three office actions. During prosecution, the Examiner raised a § 103 obviousness rejection combining Nakamura with Gagnon, International Publication No. WO 2018/065413 ("Gagnon"). Compl. ¶ 68; Ex. 5 ('647 Patent Prosecution History), Office Action dated March 14, 2022, at 4–9. Gagnon disclosed a method of combining genomic data with clinical variables for cancer risk stratification but used a different classifier architecture, a different set of clinical covariates, and did not disclose the generation of a "composite risk metric" as recited in the '647 Patent claims.
To overcome the § 103 rejection, GenomeDx submitted a declaration under 37 C.F.R. § 1.132 from Dr. Rajan Mehta, one of the named inventors. Ex. 5, Rule 1.132 Declaration of Dr. Rajan Mehta dated September 22, 2022. Dr. Mehta's declaration presented comparative data demonstrating that the combined genomic-clinical model claimed in the '647 Patent achieved a statistically significant improvement in predictive accuracy (AUC = 0.88) compared to genomic-only models (AUC = 0.79) and clinical-only models (AUC = 0.72), and that this improvement was unexpected in the art because prior attempts to combine genomic and clinical data had yielded only marginal improvements. Id. ¶¶ 8–22. The Examiner found the declaration persuasive and withdrew the § 103 rejection. Ex. 5, Office Action dated January 11, 2023, at 2.
A terminal disclaimer was filed over the '271 and '893 Patents to overcome a double-patenting rejection. Ex. 5, Terminal Disclaimer filed February 28, 2023. The '647 Patent issued on February 6, 2024. As with the '893 Patent, no § 101 rejection was sustained during prosecution. The Examiner considered and allowed the claims in full, including the integrated genomic-clinical methodology that Helios now characterizes as patent-ineligible.
D. Helios and the Accused ProstaSignal™ Test
Helios is a Delaware corporation headquartered at 200 CambridgePark Drive, Suite 600, Cambridge, Massachusetts 02140. Compl. ¶ 4. Helios operates a CLIA-certified clinical laboratory at 9830 Mesa Rim Road, Suite 140, San Diego, California 92121, where it performs the accused ProstaSignal™ Genomic Classifier test. Id. ¶ 5.
Helios commercially launched the ProstaSignal™ Genomic Classifier on October 7, 2024. Compl. ¶ 71. ProstaSignal™ is a laboratory-developed test (LDT) that, like the Decipher® test, analyzes gene expression data from post-surgical prostate tissue specimens and generates a prognostic risk score. Id. ¶¶ 72–75. The ProstaSignal™ test uses a 24-gene expression panel, of which 19 genes — fully 79% of the ProstaSignal™ gene set — overlap with GenomeDx's patented gene set as described in the '893 and '647 Patents. Id. ¶ 76. This extraordinary degree of overlap is not coincidental; it reflects Helios's deliberate decision to target the same biological pathways and gene expression signatures that GenomeDx's inventors identified and validated over more than a decade of research.
ProstaSignal™ employs a gradient-boosted decision tree classifier — a type of ensemble machine-learning model functionally related to the random-forest classifier recited in the '893 Patent — to generate a "ProstaScore" ranging from 0 to 100, which classifies patients as Low, Intermediate, or High risk for metastatic recurrence. Compl. ¶¶ 77–82. ProstaSignal™ then combines the ProstaScore with Gleason grade group and pre-operative prostate-specific antigen (PSA) level to produce an integrated report that is transmitted to the ordering physician. Id. ¶¶ 83–87. This process of combining a genomic risk score with clinical covariates to generate and transmit a composite output directly implicates the methods claimed in the '647 Patent.
Helios obtained MAAA code 0239M for ProstaSignal™ from the American Medical Association (AMA) in September 2024, facilitating reimbursement by Medicare and commercial payors. Compl. ¶ 88. Since the commercial launch of ProstaSignal™ through approximately March 2025, Helios performed approximately 8,400 tests at an average reimbursement of $3,820 per test, yielding estimated revenue of approximately $32,088,000. Id. ¶ 89. GenomeDx's market research indicates that ProstaSignal™ has diverted significant test volume from the Decipher® test and has caused GenomeDx to lose market share in the genomic prostate cancer prognostic testing market. Id. ¶¶ 90–93.
E. Procedural Background
GenomeDx filed its Complaint on March 28, 2025, initiating this action under Case No. 3:25-cv-01847-JWL. Dkt. No. 1. The Complaint asserts infringement of three patents: the '271 Patent (Count I), the '893 Patent (Count II), and the '647 Patent (Count III). GenomeDx asserts both direct infringement under 35 U.S.C. § 271(a) and induced infringement under 35 U.S.C. § 271(b) with respect to each patent.
Helios was served on April 14, 2025. Dkt. No. 7. On May 12, 2025, Helios filed its Answer and Counterclaims, asserting affirmative defenses and counterclaims of invalidity under 35 U.S.C. §§ 101, 102, and 103; non-infringement; and unenforceability. Dkt. No. 15. GenomeDx filed its Answer to Helios's Counterclaims on May 30, 2025. Dkt. No. 19.
The initial Case Management Conference was held on June 18, 2025 before the Honorable Jessica W. Langston. Dkt. No. 22. The Court issued a Scheduling Order setting the following key deadlines: close of fact discovery on March 6, 2026; opening expert reports due May 15, 2026; rebuttal expert reports due July 10, 2026; Markman hearing on September 18, 2026; and trial set to begin in February 2027. Dkt. No. 23.
The parties exchanged initial disclosures on July 14, 2025, pursuant to Federal Rule of Civil Procedure 26(a)(1). Dkt. No. 28. On August 4, 2025, the parties exchanged proposed claim construction terms, as required by Patent Local Rule 4-2. Dkt. No. 31. The parties identified multiple disputed claim terms, many of which bear directly on the § 101 analysis at issue in the instant motion. See infra § IV.C.1.
Helios filed the instant Motion to Dismiss Counts II and III under 35 U.S.C. § 101 and Federal Rule of Civil Procedure 12(b)(6) on September 15, 2025. Dkt. No. 38. Notably, Helios does not move to dismiss Count I (infringement of the '271 Patent), which involves claims from the same patent family as the '893 Patent. The selective nature of Helios's challenge underscores the thinness of its arguments: if claims from the same patent family are not "plainly" ineligible, neither are the closely related claims of the '893 and '647 Patents.
III. LEGAL STANDARD
A. Rule 12(b)(6) Standard
Under Federal Rule of Civil Procedure 12(b)(6), a complaint must be dismissed only if it fails to state a claim upon which relief can be granted. Bell Atl. Corp. v. Twombly, 550 U.S. 544, 570 (2007). To survive a motion to dismiss, a complaint must contain "sufficient factual matter, accepted as true, to 'state a claim to relief that is plausible on its face.'" Ashcroft v. Iqbal, 556 U.S. 662, 678 (2009) (quoting Twombly, 550 U.S. at 570). A claim is facially plausible "when the plaintiff pleads factual content that allows the court to draw the reasonable inference that the defendant is liable for the misconduct alleged." Id.
In evaluating a motion to dismiss, the Court must accept all well-pleaded factual allegations in the complaint as true and draw all reasonable inferences in the plaintiff's favor. Id. at 678–79. The Court's inquiry is limited to the legal sufficiency of the complaint; it does not assess the weight of the evidence, resolve factual disputes, or make credibility determinations. Twombly, 550 U.S. at 556.
These principles apply with particular force in the patent context. Dismissal under § 101 at the pleading stage is appropriate only in the most clear-cut cases — where the claims are "plainly" directed to ineligible subject matter and no reasonable construction of the claims could save them. See Aatrix Software, Inc. v. Green Shades Software, Inc., 882 F.3d 1121, 1125 (Fed. Cir. 2018) (holding that whether patent claims are directed to patent-ineligible subject matter is a legal question that may contain underlying factual disputes, and that "not every § 101 determination can be made on the pleadings"). This is an exacting standard that strongly disfavors early resolution of eligibility questions, particularly when — as here — the claims recite specific technical limitations, the claim terms are disputed, and no evidentiary record exists.
B. The *Alice/Mayo* Two-Step Framework
The Supreme Court has articulated a two-step framework for assessing patent eligibility under 35 U.S.C. § 101. Alice Corp. Pty. Ltd. v. CLS Bank Int'l, 573 U.S. 208, 217–18 (2014); Mayo Collaborative Servs. v. Prometheus Labs., Inc., 566 U.S. 66, 77–80 (2012).
At Step One, the Court must determine whether the patent claims are "directed to" a patent-ineligible concept — that is, whether they are directed to a law of nature, a natural phenomenon, or an abstract idea. Alice, 573 U.S. at 217; Mayo, 566 U.S. at 77. This inquiry requires examining the claims as a whole — not dissecting them into individual components — to determine their "focus" or "character." Diamond v. Diehr, 450 U.S. 175, 188 (1981) ("In determining the eligibility of respondents' claimed process for patent protection under § 101, their claims must be considered as a whole."). The Supreme Court has cautioned that "all inventions at some level embody, use, reflect, rest upon, or apply laws of nature, natural phenomena, or abstract ideas," and that the § 101 inquiry must therefore be conducted at an appropriate level of generality. Alice, 573 U.S. at 217 (quoting Mayo, 566 U.S. at 71).
The Federal Circuit has emphasized that when claims recite "a specific, discrete implementation of the abstract idea" rather than the abstract idea itself, they are not "directed to" the abstract idea at Step One. See Vanda Pharms. Inc. v. West-Ward Pharms. Int'l Ltd., 887 F.3d 1117, 1134 (Fed. Cir. 2018). Similarly, when claims recite a specific, practical application of a natural phenomenon — rather than claiming the phenomenon itself — they pass muster at Step One. See Rapid Litig. Mgmt. Ltd. v. CellzDirect, Inc., 827 F.3d 1042, 1048 (Fed. Cir. 2016); Ostrander & Co. v. Ariosa Diagnostics, Inc., 967 F.3d 1319, 1326–27 (Fed. Cir. 2020).
At Step Two, if the claims are found to be "directed to" an ineligible concept at Step One, the Court must determine whether the claim elements, individually or as an "ordered combination," transform the nature of the claim into a patent-eligible application by reciting an "inventive concept" — i.e., "an element or combination of elements that is sufficient to ensure that the patent in practice amounts to significantly more than a patent upon the [ineligible concept] itself." Alice, 573 U.S. at 217–18 (internal quotation marks omitted). This inquiry may involve underlying factual questions, including whether the claim elements, individually or in combination, were "well-understood, routine, and conventional" in the relevant art at the time of the invention. Berkheimer v. HP Inc., 881 F.3d 1360, 1368 (Fed. Cir. 2018).
The Federal Circuit has repeatedly cautioned that § 101 determinations involve "subtle" and "hard to predict" analyses that often require "consideration of the particular claim language, the specification, the prosecution history, and relevant extrinsic evidence." Berkheimer, 881 F.3d at 1368. These considerations militate strongly against resolving eligibility at the pleading stage, where no evidentiary record exists and the Court's analysis is necessarily limited to the face of the complaint and the patent documents.
C. Prematurity of § 101 Challenges Before Claim Construction
Courts in the Northern District of California and the Federal Circuit have repeatedly recognized that resolving § 101 challenges without the benefit of claim construction can be inappropriate, particularly when claim scope is disputed and the construction of specific terms may materially affect the eligibility analysis. See, e.g., Cellspin Soft, Inc. v. Fitbit, Inc., 927 F.3d 1306, 1319 (Fed. Cir. 2019) (vacating dismissal under § 101 where the district court's eligibility determination "depend[ed] on the scope of the claims" and claim construction had not yet occurred); MyMail, Ltd. v. ooVoo, LLC, 934 F.3d 1373, 1379–80 (Fed. Cir. 2019) (noting that "claim construction may be helpful, and in some cases necessary, to resolve § 101 issues").
The rationale for deferral is straightforward: the § 101 analysis, at both Step One and Step Two, depends on what the claims actually recite. When the parties dispute the meaning of specific claim terms — as they do here — the Court cannot accurately assess whether the claims are "directed to" an abstract idea or natural phenomenon, or whether they contain an "inventive concept," without first determining what the claims actually mean. Construing claims at the 12(b)(6) stage, without the benefit of intrinsic and extrinsic evidence, expert testimony, and full briefing, risks error and may deprive the parties of a meaningful opportunity to be heard on claim scope.
This concern is particularly acute in the life sciences context, where the line between patent-eligible and patent-ineligible subject matter often turns on the precise scope of the claim limitations. See, e.g., Ass'n for Molecular Pathology v. Myriad Genetics, Inc., 569 U.S. 576, 594–95 (2013) (distinguishing between claims to isolated DNA sequences, which were ineligible, and claims to complementary DNA, which were eligible, based on the specific scope of the claims). The § 101 analysis for diagnostic method claims is especially fact-intensive and context-dependent, and this Court should decline Helios's invitation to short-circuit that analysis at the earliest possible stage of this litigation.
IV. ARGUMENT
A. The Asserted Claims of the '893 Patent Are Directed to a Patent-Eligible Diagnostic Method, Not an Abstract Idea
1. Step One: The '893 Patent Claims Are Not "Directed To" a Law of Nature or Abstract Idea. Helios's motion characterizes the claims of the '893 Patent as merely "applying a mathematical algorithm to gene expression data" and contends that gene expression levels are "natural phenomena" that cannot be the subject of a patent. Mot. at 8–12. This characterization improperly dissects the claims, strips away their specific technical limitations, and ignores the concrete, defined methodology that Claim 1 actually recites. When properly analyzed at the appropriate level of generality — and without the impermissible reductionism Helios employs — the '893 Patent claims are clearly directed to a specific, concrete diagnostic method rather than an abstract idea or natural phenomenon.
Claim 1 of the '893 Patent does not claim gene expression levels in the abstract. It does not claim the mere observation that certain genes are expressed at higher or lower levels in aggressive prostate cancers. And it does not claim any and all methods of analyzing gene expression data for diagnostic purposes. Rather, Claim 1 recites a defined series of physical and computational steps:
(a) Normalizing gene-expression data from a specific biological sample type — formalin-fixed paraffin-embedded (FFPE) prostate tissue — using a reference gene panel. This step is not merely "gathering data." FFPE tissue presents unique analytical challenges because the formalin fixation process causes nucleic acid degradation, cross-linking, and modification, which introduce systematic biases into gene expression measurements. The normalization step using a reference gene panel is a specific laboratory and computational technique designed to correct for these biases. The '893 Patent specification describes the reference gene panel and normalization methodology in detail, including the selection criteria for reference genes (housekeeping genes with stable expression across tumor and normal tissue) and the normalization algorithm. '893 Patent, col. 8, ll. 15–45; col. 9, ll. 1–32.
(b) Inputting normalized expression values into a random-forest classifier trained on a cohort of at least 500 patient samples. This limitation specifies a particular machine-learning architecture (random-forest) and a minimum training dataset size (500 patient samples). A random-forest classifier is a specific ensemble learning method that constructs multiple decision trees during training and outputs the mode of the individual trees' predictions. The minimum cohort size of 500 patient samples reflects the inventors' discovery that classifier performance was materially degraded when trained on smaller cohorts — a non-obvious finding that was critical to the commercial viability of the Decipher® test. '893 Patent, col. 14, ll. 7–38. This limitation is not a generic invocation of "machine learning" or "mathematics"; it specifies a defined computational architecture with enumerated training parameters.
(c) Outputting a numerical prognostic score correlated with metastasis risk within 5 years. This step specifies the clinical output of the method — a numerical score correlated with a specific clinical endpoint (metastasis risk within a defined timeframe). This is not a claim to the natural phenomenon that certain gene expression patterns correlate with metastasis; it is a claim to a specific method of generating a clinically actionable numerical output.
When these limitations are considered together — as the Supreme Court requires — the claims describe a specific, integrated diagnostic method, not an abstract concept. The claims specify the type of sample (FFPE prostate tissue), the normalization methodology (reference gene panel), the classifier architecture (random-forest), the training protocol (at least 500 patient samples), and the clinical output (numerical prognostic score correlated with metastasis risk within 5 years). No one could read these claims and conclude that they preempt all methods of analyzing gene expression data or all applications of machine learning to cancer diagnosis. They are narrow, specific, and directed to a particular diagnostic technology.
The Federal Circuit's decisions in Rapid Litigation Management v. CellzDirect, Vanda Pharmaceuticals v. West-Ward Pharmaceuticals, and Marlowe Group v. Ariosa Diagnostics confirm that claims of this character are patent-eligible.
In CellzDirect, the Federal Circuit held that method claims directed to producing hepatocyte preparations by subjecting hepatocytes to multiple freeze-thaw cycles were patent-eligible, even though the claims relied on the natural phenomenon that certain hepatocytes survive multiple freeze-thaw cycles. 827 F.3d at 1047–50. The court reasoned that the claims were "directed to" a specific laboratory method — not to the natural phenomenon itself — because they recited "a new and useful laboratory technique for preserving hepatocytes." Id. at 1048. Similarly, the '893 Patent claims are directed to a specific diagnostic method — not to the natural phenomenon that gene expression patterns correlate with cancer prognosis. The claims recite a specific technique for processing FFPE tissue samples, normalizing gene expression data, and generating a prognostic score using a defined classifier architecture. Like the claims in CellzDirect, the '893 Patent claims are directed to a specific application of knowledge about natural phenomena, not to the phenomena themselves.
In Vanda Pharmaceuticals, the Federal Circuit held that claims directed to a method of treating schizophrenia patients with iloperidone, comprising genotyping the patient for a specific CYP2D6 polymorphism and administering a dose based on the genotype, were patent-eligible. 887 F.3d at 1134–36. The court distinguished these claims from the claims in Mayo, which merely recited observing and measuring a natural phenomenon and adding conventional steps. Id. The Vanda claims, by contrast, recited a specific treatment regimen based on genotyping results — a specific, practical application of the underlying natural relationship between genotype and drug metabolism. Id. The '893 Patent claims are analogous: they do not merely recite observing a natural correlation between gene expression and cancer prognosis. They recite a specific methodology for processing samples, generating normalized data, applying a defined classifier, and producing a clinically actionable output.
In Marlowe Group v. Ariosa Diagnostics, the Federal Circuit held that claims directed to methods of preparing cell-free DNA (cfDNA) samples for analysis by selecting for DNA fragments of a specific size range were patent-eligible. 967 F.3d at 1326–28. The court emphasized that the claims were "not directed to" the natural phenomenon that fetal cfDNA tends to be shorter than maternal cfDNA; rather, they were "directed to" a specific method of exploiting that knowledge by performing a size-selection step. Id. at 1327. The court noted that the "claimed advance" was not the discovery of the natural phenomenon but the development of a specific, practical method for using it. Id. The same analysis applies here: the '893 Patent's claimed advance is not the discovery that gene expression correlates with cancer prognosis but the development of a specific method for generating a prognostic score from FFPE tissue using a random-forest classifier with defined training parameters.
Helios's characterization is also distinguishable from the claims at issue in Mayo. In Mayo, the Supreme Court held that claims covering a method of optimizing drug dosage by measuring metabolite levels and comparing them to predetermined thresholds were patent-ineligible. 566 U.S. at 73–74. The Mayo claims were broad enough to cover essentially all practical applications of the natural correlation between metabolite levels and drug efficacy; they recited no specific methodology, no particular measurement technique, and no defined analytical framework. Id. at 78–79. The '893 Patent claims are fundamentally different. They do not cover all methods of correlating gene expression with cancer prognosis. They cover a specific method — using FFPE tissue, reference-gene normalization, a random-forest classifier trained on at least 500 samples — that represents one particular approach among many possible approaches to genomic cancer prognosis.
Helios's characterization is similarly distinguishable from Myriad Genetics. In Myriad, the Supreme Court held that claims to isolated, naturally occurring DNA sequences were patent-ineligible because the claims covered the genetic information itself, not any new application of that information. 569 U.S. at 590–91. The '893 Patent does not claim any naturally occurring gene sequence, gene expression level, or other product of nature. It claims a method — a series of defined steps that transform gene expression data from FFPE tissue into a clinically actionable prognostic score. The distinction between a product-of-nature claim and a method-of-use claim is well established, and the '893 Patent falls squarely on the patent-eligible side of that line.
In sum, at Step One of the Alice/Mayo framework, the '893 Patent claims are not "directed to" a law of nature, natural phenomenon, or abstract idea. They are directed to a specific, concrete diagnostic method. This Court should deny Helios's motion on this ground alone.
2. Step Two: Even Under Helios's Characterization, the Claims Contain an Inventive Concept. Even assuming arguendo that this Court were to conclude at Step One that the '893 Patent claims are "directed to" a natural phenomenon or abstract idea — a conclusion GenomeDx vigorously disputes — the claims nonetheless recite sufficient inventive concepts to satisfy Step Two of the Alice/Mayo framework.
The Supreme Court has explained that at Step Two, the Court must examine the claim elements, both individually and as an "ordered combination," to determine whether they transform the nature of the claim into a patent-eligible application. Alice, 573 U.S. at 217–18. The question is whether the claims recite "something more" than the ineligible concept itself — an element or combination of elements that amounts to "significantly more" than a patent on the abstract idea or natural phenomenon. Id.
The '893 Patent claims recite a combination of elements that, taken together, represent significantly more than the generic concept of "analyzing gene expression data":
(i) FFPE tissue processing with reference-gene normalization. The use of FFPE tissue samples — as opposed to fresh-frozen tissue — was a critical innovation because FFPE tissue is routinely collected and archived during surgical pathology, whereas fresh-frozen tissue requires special handling and is often unavailable in clinical practice. Adapting gene expression profiling to FFPE tissue required the development of specific normalization techniques to account for the degradation and modification caused by formalin fixation. The reference-gene normalization approach recited in Claim 1 was developed specifically for this purpose and is described in detail in the specification. '893 Patent, col. 8–9.
(ii) A random-forest classifier with minimum cohort training requirements. The claim specifies not just any classifier, but a random-forest classifier — a specific machine-learning architecture — and further requires that the classifier be trained on a cohort of at least 500 patient samples. This minimum cohort size was not an arbitrary number; it was determined through extensive experimentation by the inventors, who discovered that classifier performance was unreliable when trained on smaller cohorts due to overfitting and sampling bias. '893 Patent, col. 14, ll. 7–38. The specification discloses the inventors' experimental results comparing classifier performance across cohorts of different sizes, demonstrating that a minimum of 500 samples was necessary to achieve clinically acceptable accuracy. Id.
(iii) Correlation to a specific clinical endpoint. The claim specifies that the numerical prognostic score is correlated with metastasis risk within 5 years — a specific, clinically meaningful endpoint that requires extensive validation and statistical analysis.
The combination of these elements — FFPE tissue processing, reference-gene normalization, a random-forest classifier with minimum cohort training requirements, and correlation to a specific clinical endpoint — is not routine, conventional, or well-understood. The prosecution history confirms this: the Examiner specifically allowed Claim 1 after the applicant distinguished the random-forest classifier training methodology and minimum cohort size from the prior art (Nakamura et al., U.S. Patent No. 9,617,340). Ex. 4, Response to Office Action dated November 18, 2020, at 8–15; Notice of Allowance dated February 3, 2021. The Examiner recognized that the specific classifier architecture and training methodology recited in the claims were distinguishing features — not conventional activity.
Helios bears the burden of establishing that the claimed elements are "well-understood, routine, and conventional," and it has not done so. Under the Federal Circuit's decision in Berkheimer v. HP Inc., 881 F.3d 1360, 1368 (Fed. Cir. 2018), "[w]hether something is well-understood, routine, and conventional to a skilled artisan at the time of the patent is a factual determination." Helios's motion to dismiss does not present — and cannot present at this stage — any evidence that the specific combination of elements recited in the '893 Patent claims was conventional at the time of the invention. Helios offers only attorney argument and bare assertions, which are insufficient to establish that the claims lack an inventive concept. See Aatrix Software, 882 F.3d at 1128 ("The question of whether a claim element or combination of elements is well-understood, routine and conventional to a skilled artisan in the relevant field is a question of fact that can be properly resolved on a Rule 12(b)(6) motion only if 'the well-pleaded facts' in the complaint do not 'plausibly allege' an inventive concept.").
Here, the Complaint alleges — and the prosecution history confirms — that the claimed combination of elements was novel and non-obvious, and that it represented a significant advancement over prior art methods of genomic cancer prognosis. These allegations must be accepted as true at the 12(b)(6) stage. The '893 Patent claims therefore satisfy Step Two, and Helios's motion should be denied.
B. The Asserted Claims of the '647 Patent Are Directed to a Patent-Eligible Integrated Diagnostic Method
1. Step One: The '647 Patent Claims Are Not "Directed To" a Law of Nature or Abstract Idea. The '647 Patent claims a fundamentally different and more integrated method than the generic "gene-expression analysis" that Helios invokes. While the '893 Patent claims are directed to a specific method of generating a genomic prognostic score, the '647 Patent claims go further by requiring the integration of a genomic risk score with clinical covariates to generate a composite risk metric — and the transmission of that metric to a healthcare provider. This integrated approach represents a specific, practical diagnostic method that is plainly patent-eligible.
Claim 1 of the '647 Patent recites the following steps:
(a) Performing RNA extraction and expression profiling on a prostate tumor specimen. This step specifies the biological sample (prostate tumor specimen) and the analytical technique (RNA extraction followed by expression profiling). RNA extraction from tumor specimens involves specific laboratory protocols, including tissue disruption, lysis, and purification, that are tailored to the particular sample type. The expression profiling step generates quantitative measurements of gene expression levels across the gene features specified by the classifier.
(b) Computing a genomic risk score using a classifier comprising at least 20 gene features. This step specifies a minimum number of gene features (at least 20) for the classifier, reflecting the inventors' determination that classifiers using fewer than 20 features lacked sufficient predictive power for the integrated model. '647 Patent, col. 15, ll. 22–48.
(c) Combining the genomic risk score with at least two clinical covariates selected from the group consisting of Gleason score, PSA level, surgical margin status, and pathological stage. This is the key integration step. The claim does not merely require "gathering" clinical data alongside genomic data; it requires the computational combination of the genomic risk score with specific, enumerated clinical covariates. The '647 Patent specification describes the specific mathematical framework for this combination, including weighted regression models and nomogram-based integration techniques. '647 Patent, col. 18, ll. 5–42.
(d) Generating a composite risk metric. The output of the integration step is a single, composite metric that synthesizes the genomic and clinical data into a unified prognostic assessment. This composite metric is distinct from any individual input variable and represents a new, clinically actionable data point.
(e) Transmitting the composite risk metric to a healthcare provider. The final step specifies the clinical application of the composite metric — its transmission to a healthcare provider for use in treatment decision-making. This step confirms that the claimed method is directed to a specific, practical diagnostic application rather than an abstract data-processing exercise.
Considered as a whole, the claims of the '647 Patent are directed to a specific, integrated diagnostic method — not to an abstract idea or natural phenomenon. The claims do not merely recite the observation that gene expression levels and clinical variables are correlated with cancer prognosis. They recite a specific method for generating a composite risk metric by computationally combining genomic data with defined clinical covariates and transmitting the result to a healthcare provider. This is precisely the type of "specific, practical application" that the Federal Circuit has repeatedly held to be patent-eligible.
The Federal Circuit's decision in CardioNet, LLC v. InfoBionic, Inc., 955 F.3d 1358 (Fed. Cir. 2020), is directly on point. In CardioNet, the court held that claims directed to a cardiac monitoring system that detected and distinguished atrial fibrillation from atrial flutter by analyzing multiple data streams were patent-eligible. Id. at 1368–71. The court emphasized that the claims were directed to "a specific, improved cardiac monitoring process," not to "the abstract idea of detecting atrial fibrillation." Id. at 1370. The court noted that the claims recited "specific technical means" for achieving the diagnostic result, including "specific device configurations" and "specific processing steps." Id. Similarly, the '647 Patent claims recite specific technical means — RNA extraction, expression profiling, computation of a genomic risk score using a classifier with at least 20 gene features, integration with defined clinical covariates, generation of a composite metric, and transmission to a healthcare provider. These are specific processing steps directed to a particular diagnostic method, not abstract data manipulation.
The Federal Circuit's decision in Marlowe Group v. Ariosa Diagnostics further supports patent eligibility. In Marlowe Group, the court held that claims directed to methods of preparing cell-free DNA samples by performing a size-selection step were patent-eligible, even though the claims relied on the natural phenomenon that fetal cfDNA tends to be shorter than maternal cfDNA. 967 F.3d at 1326–28. The court distinguished these claims from claims that merely "recite[d] a natural phenomenon and add[ed] only 'apply it'" steps. Id. at 1327. The '647 Patent claims are analogous: they do not merely observe a natural correlation and add generic steps. They recite a specific, multi-step method for integrating genomic and clinical data to produce a novel composite output.
Helios's argument that any method involving gene expression data is per se abstract or directed to a natural phenomenon ignores the specificity of the '647 Patent claims and misreads the relevant case law. As the Supreme Court recognized in Diamond v. Diehr, 450 U.S. at 188, the use of a mathematical formula or algorithm in a method claim does not render the claim patent-ineligible where the formula or algorithm is applied to achieve a specific, practical result. The '647 Patent claims do not preempt all uses of gene expression data or all methods of combining genomic and clinical variables. They claim one specific method — with defined inputs, a defined analytical process, and a defined output — and leave the field open for others to develop different approaches to genomic-clinical integration.
Moreover, the '647 Patent claims are directed to a method that was demonstrated during prosecution to yield unexpected results. As discussed below, Dr. Mehta's Rule 1.132 declaration showed that the specific integration of genomic risk scoring with the defined clinical covariates produced a statistically significant improvement in predictive accuracy that was not achieved by genomic-only models, clinical-only models, or prior attempts to combine genomic and clinical data. This further confirms that the claims are directed to a specific, inventive diagnostic method — not to a natural phenomenon or abstract idea.
2. Step Two: The Combined Genomic-Clinical Model Represents an Inventive Concept. Even assuming arguendo that the '647 Patent claims are "directed to" a natural phenomenon or abstract idea at Step One — which GenomeDx disputes — the claims recite an inventive concept sufficient to satisfy Step Two.
The prosecution history of the '647 Patent is particularly instructive on this point. As described above, the Examiner raised § 103 obviousness rejections based on combinations of Nakamura and Gagnon. To overcome these rejections, GenomeDx submitted a Rule 1.132 declaration from Dr. Rajan Mehta establishing the unexpected results of the combined genomic-clinical model versus genomic-only models. Ex. 5, Rule 1.132 Declaration of Dr. Rajan Mehta dated September 22, 2022. Dr. Mehta's declaration demonstrated:
• The combined genomic-clinical model achieved an area under the receiver operating characteristic curve (AUC) of 0.88 for predicting five-year metastasis risk, compared to an AUC of 0.79 for the genomic-only model and an AUC of 0.72 for the clinical-only model. Ex. 5, Mehta Decl. ¶ 12.
• The improvement from 0.79 (genomic-only) to 0.88 (combined model) was statistically significant (p < 0.001) and clinically meaningful, representing a reclassification of approximately 15% of patients into different risk categories. Id. ¶ 14.
• Prior attempts in the art to combine genomic data with clinical variables had yielded only marginal improvements (AUC improvements of 0.01–0.03), and the 0.09-point AUC improvement achieved by the '647 Patent's method was unexpected and not predicted by the prior art. Id. ¶¶ 16–19.
• The specific selection and combination of clinical covariates recited in the claims — Gleason score, PSA level, surgical margin status, and pathological stage — was not arbitrary but was the result of extensive variable-selection analysis that identified the optimal set of clinical covariates for integration with the genomic risk score. Id. ¶¶ 20–22.
The Examiner found Dr. Mehta's declaration persuasive and withdrew the § 103 rejection, issuing a Notice of Allowance. Ex. 5, Office Action dated January 11, 2023, at 2; Notice of Allowance dated April 3, 2023. The fact that the specific combination of elements was proven to yield unexpected results during prosecution strongly supports the conclusion that the claims recite an inventive concept at Step Two.
No prior art reference, individually or in combination, disclosed the specific integration method claimed in the '647 Patent. The claimed method of combining a genomic risk score derived from a classifier with at least 20 gene features with at least two clinical covariates from a defined list to generate a composite risk metric was novel, non-obvious, and demonstrated to produce results that the art did not predict. This is the very definition of an "inventive concept."
Under Berkheimer, whether the claimed elements are "well-understood, routine, and conventional" is a factual question. 881 F.3d at 1368. Helios has offered no evidence — expert declarations, technical literature, or otherwise — to establish that the specific combination of elements recited in the '647 Patent claims was conventional at the time of invention. To the contrary, the prosecution history affirmatively demonstrates that the combination was unexpected and non-obvious. At the 12(b)(6) stage, where GenomeDx's factual allegations must be accepted as true, the claims plainly satisfy Step Two.
C. Helios's § 101 Challenge Is Premature Before Claim Construction
1. Disputed Claim Terms Bear Directly on the Eligibility Analysis. Entirely apart from the substantive merits of the eligibility analysis, Helios's motion should be denied as premature. The parties exchanged proposed claim construction terms on August 4, 2025 and identified multiple disputed terms relevant to the '893 and '647 Patents, including:
• "random-forest classifier trained on a cohort of at least 500 patient samples" ('893 Patent, Claim 1). GenomeDx proposes a construction that requires a specific, defined computational methodology with minimum training data requirements, including the use of bootstrapped aggregation of multiple decision trees and training on a cohort comprising at least 500 patient samples with confirmed clinical outcome data. Helios proposes a broader construction that encompasses any ensemble classifier methodology. The difference between these constructions directly impacts the Step One analysis: under GenomeDx's construction, the claims plainly recite a specific, concrete technological method; under Helios's broader construction, the claims might (though GenomeDx disputes this) be more susceptible to characterization as abstract.
• "composite risk metric" ('647 Patent, Claim 1). GenomeDx proposes a construction that requires a single numerical value generated by the computational integration of a genomic risk score with at least two clinical covariates using a defined mathematical model. Helios proposes that "composite risk metric" means simply any output that combines genomic and clinical data, without specifying the form or method of combination. Again, the construction directly impacts the eligibility analysis.
• "classifier comprising at least 20 gene features" ('647 Patent, Claim 1). GenomeDx proposes a construction that requires a trained machine-learning classifier that uses quantitative expression measurements from at least 20 specified gene features as inputs. Helios proposes a broader construction.
• "trained machine-learning classifier" ('271 Patent, also relevant as a related family member). GenomeDx proposes a construction requiring a classifier that has undergone a supervised learning process using labeled training data from actual patient samples with known clinical outcomes.
• "risk category" ('271 and '893 Patents). GenomeDx proposes a construction requiring a predefined categorical classification (e.g., Low, Intermediate, High) based on validated cut-points. Helios proposes that "risk category" means any grouping of risk scores.
The scope and meaning of each of these terms directly impacts the § 101 analysis at both Step One and Step Two. For example, if "random-forest classifier trained on a cohort of at least 500 patient samples" is construed consistently with GenomeDx's proposed construction — to require a specific computational methodology with defined training parameters — then the '893 Patent claims are clearly directed to a concrete technological improvement rather than an abstract mathematical concept. If "composite risk metric" is construed to require a specific, computationally integrated output generated through a defined mathematical model, then the '647 Patent claims are clearly directed to a practical diagnostic method rather than the generic concept of "combining data."
Conversely, if the Court were to adopt Helios's broader proposed constructions, the eligibility analysis would proceed on a fundamentally different footing. This demonstrates that the outcome of the § 101 inquiry is inextricably linked to claim construction — and that resolving § 101 before the Court has construed the disputed claim terms would be premature and potentially incorrect.
The Markman hearing is scheduled for September 18, 2026. At that hearing, the Court will have the benefit of full briefing on the disputed claim terms, including intrinsic evidence (the claims, specification, and prosecution history), extrinsic evidence (expert declarations, technical dictionaries, and treatises), and oral argument from both parties. Resolving § 101 now — before any of this evidence has been developed or presented — would be premature and contrary to the interests of accuracy and judicial economy.
2. Courts Routinely Defer § 101 to Post-Claim-Construction. Courts in this District and the Federal Circuit have recognized that resolving patent eligibility before construing disputed claim terms risks error and is generally disfavored when the claim scope is genuinely in dispute.
The Federal Circuit has held that "claim construction may be helpful, and in some cases necessary, to resolve § 101 issues." MyMail