DCT

3:26-cv-05050

Evidium Inc v. BMJ Publishing Group Ltd

Log In
Key Events
Amended Complaint
complaint Intelligence

I. Executive Summary and Procedural Information

  • Parties & Counsel:
  • Case Identification: 4:26-cv-05050, N.D. Cal., 09/03/2026
  • Venue Allegations: Venue is alleged to be proper as Defendant is an alien defendant over whom the court has personal jurisdiction.
  • Core Dispute: Plaintiff alleges that after a failed business collaboration, Defendant misappropriated Plaintiff's trade secrets and infringed five of its patents to develop a competing artificial intelligence platform that makes medical evidence computational.
  • Technical Context: The technology involves using neuro-symbolic AI and graph databases to convert unstructured medical literature and clinical data into a structured, computable, and reusable format for applications in clinical decision support and diagnostics.
  • Key Procedural History: The complaint alleges that the dispute arises from a business relationship that began in 2020. Plaintiff alleges it sent multiple pre-suit letters notifying Defendant of infringement of U.S. Patents 11,869,674, 12,518,882, and 12,518,883. The complaint also notes that claims for several of the asserted patents were allowed during prosecution only after amendments were made to overcome rejections under 35 U.S.C. § 101 (patent eligibility).

Case Timeline

Date Event
2020-04-01 Evidium and BMJ discussions regarding a potential business relationship begin
2020-09-01 Evidium allegedly shares initial core design frameworks with BMJ
2021-06-01 "Best Practice Agreement" executed between parties
2021-09-11 Evidium provides proposal to BMJ to make its "Action Sets" computable
2021-10-01 "Technology Licence Agreement" executed between parties
2021-12-09 "Tech and Product Deep Dive" meeting held between parties
2022-01-19 Additional deep-dive meetings held between parties
2022-04-15 Priority Date for '674, '882, '883, '274, and '716 Patents
2024-01-09 U.S. Patent 11,869,674 ('674 Patent) Issues
2024-03-01 Alleged public launch of BMJ's competing "Clinical Intelligence" product
2024-11-01 Evidium alleges it discovered BMJ's "Clinical Intelligence" product
2024-12-12 Evidium counsel sends first letter to BMJ alleging infringement of '674 Patent
2025-03-01 BMJ demonstrates "Clinical Intelligence" at HIMSS 2025 Global Conference
2025-04-08 Evidium counsel sends letter with further evidence of '674 Patent infringement
2026-01-06 U.S. Patent 12,518,882 ('882 Patent) Issues
2026-01-06 U.S. Patent 12,518,883 ('883 Patent) Issues
2026-04-08 Evidium counsel sends letter notifying BMJ of infringement of '882 and '883 Patents
2026-05-26 U.S. Patent 12,640,274 ('274 Patent) Issues
2026-08-21 Evidium provides notice to BMJ that the '716 patent will issue and be asserted
2026-08-24 BMJ allegedly receives actual notice of '274 Patent infringement
2026-09-01 U.S. Patent 12,725,716 ('716 Patent) Issues
2026-09-03 Complaint Filed

II. Technology and Patent(s)-in-Suit Analysis

U.S. Patent No. 11,869,674 - "Translation of Medical Evidence Into Computational Evidence and Applications Thereof"

  • Patent Identification: U.S. Patent No. 11,869,674, "Translation of Medical Evidence Into Computational Evidence and Applications Thereof," issued January 9, 2024.

The Invention Explained

  • Problem Addressed: The patent describes the problem that traditional medical evidence exists primarily in narrative, document-based forms designed for human interpretation (('674 Patent, col. 1:59-63)). This format is slow, inefficient, and not scalable for use in modern computational systems like AI, leading to delays in care, clinician burnout, and misdiagnosis Compl. ¶¶45-47
  • The Patented Solution: The invention is a method for transforming this narrative medical evidence into structured, reusable "computational elements" (('674 Patent, col. 2:31-34)). This process involves using Natural Language Processing (NLP) to extract clinical concepts, which are then reviewed, grouped, and stored as distinct nodes in a graph database (('674 Patent, Fig. 3)). This creates a computational knowledge base where the evidence itself is machine-readable, allowing software applications to query the data to determine a patient's condition or recommend next-best actions (('674 Patent, abstract)).
  • Technical Importance: This approach creates a computable and traceable foundation for healthcare AI, an improvement over text-only AI architectures that may lack explainability Compl. ¶56

Key Claims at a Glance

  • The complaint asserts independent claim 1 and dependent claims 2-9 and 11-30 Compl. ¶89
  • Independent Claim 1 of the '674 patent recites a method with the essential elements of:
    • Inputting electronic text from a medical evidence source pertaining to a medical condition.
    • Processing the text with an NLP engine to produce a plurality of diagnostic factors.
    • Displaying the diagnostic factors to a person and receiving input that indicates a "factor group" subsuming the factors and a rank for that group.
    • Storing the medical condition, factor group, and diagnostic factors as separate entries in a database, with specific linkages between them and a requirement that the stored factor entries do not include information specific to the medical condition.

U.S. Patent No. 12,518,882 - "Translation of Medical Evidence Into Computational Evidence and Applications Thereof"

  • Patent Identification: U.S. Patent No. 12,518,882, "Translation of Medical Evidence Into Computational Evidence and Applications Thereof," issued January 6, 2026.

The Invention Explained

  • Problem Addressed: The '882 patent shares a common specification with the '674 patent and addresses the same problem: medical evidence is typically represented in narrative forms unsuitable for direct machine computation Compl. ¶44 Compl. ¶¶46-47
  • The Patented Solution: The '882 patent claims a system for implementing the computational evidence platform. It recites a computer system with processors and memory configured to receive medical text, retrieve diagnostic factors, and create a cross-linked graph database representing medical conditions and factors (('882 Patent, claim 1)). A key feature of the claimed system is the establishment of one or more Application Programming Interfaces (APIs) that allow an external software application to integrate with and access the graph database to "generate a healthcare determination" (('882 Patent, claim 1); Compl. ¶58).
  • Technical Importance: This system architecture provides a persistent, machine-operable representation of medical knowledge that can be accessed by other software, enabling functionality that prior document-oriented approaches could not provide Compl. ¶55

Key Claims at a Glance

  • The complaint asserts independent claim 1 and dependent claims 2-9 and 11-41 Compl. ¶99
  • Independent Claim 1 of the '882 patent recites a computer system with the essential elements of:
    • Processors and memory storing instructions.
    • Receiving electronic text from medical evidence sources.
    • Retrieving a plurality of diagnostic factors associated with a medical condition, where the condition and a symptom share the same name in the source text but are arranged in different hierarchies.
    • Creating a graph database that cross-links the condition and factors as separate entries to disambiguate the concepts.
    • Establishing one or more APIs for the graph database, allowing an application to integrate with it to "generate a healthcare determination."

Multi-Patent Capsule: U.S. Patent No. 12,518,883 ('883 Patent)

  • Technology Synopsis: The '883 patent, sharing a common specification with the patents above, covers methods and systems for translating medical evidence into computational elements Compl. ¶35 The complaint notes its claims were allowed after amendments reciting the outputting of data to a software application via an API "to generate a medical determination," which was argued to be a software construct a human cannot practically perform Compl. ¶59
  • Asserted Claims: Claims 1-34 Compl. ¶110
  • Accused Features: The BMJ Clinical Intelligence platform and related products using the BMJ knowledge graph are accused of infringement Compl. ¶111

Multi-Patent Capsule: U.S. Patent No. 12,640,274 ('274 Patent)

  • Technology Synopsis: The '274 patent covers systems for translating medical evidence into computational elements Compl. ¶38 The complaint highlights that its claims were allowed after amendments reciting, among other things, the use of an NLP engine and a structure to "disambiguate" clinical concepts by representing them in different nodes in a graph database Compl. ¶49 Compl. ¶60
  • Asserted Claims: Claims 1-25 Compl. ¶121
  • Accused Features: The BMJ Clinical Intelligence platform and related products using the BMJ knowledge graph are accused of infringement Compl. ¶122

Multi-Patent Capsule: U.S. Patent No. 12,725,716 ('716 Patent)

  • Technology Synopsis: The '716 patent covers methods and systems for translating medical evidence into computational elements Compl. ¶41 The complaint states its claims were allowed after amendments reciting a knowledge graph that includes investigation nodes linked to condition nodes and ontology links to different electronic medical ontologies that are traversable from the knowledge graph Compl. ¶61
  • Asserted Claims: Claims 1-25 Compl. ¶132
  • Accused Features: The BMJ Clinical Intelligence platform and related products using the BMJ knowledge graph are accused of infringement Compl. ¶133

III. The Accused Instrumentality

Product Identification

  • The primary accused instrumentality is the "BMJ Clinical Intelligence" platform, which relies on an underlying "BMJ knowledge graph" Compl. ¶¶77-78

Functionality and Market Context

  • The complaint alleges that BMJ Clinical Intelligence transforms clinical content into "computable evidence" delivered through a "dynamic knowledge graph" Compl. ¶80 This platform is described as supporting APIs, workflow integrations, and large language model systems Compl. ¶80 BMJ allegedly markets the platform as providing "computable evidence at scale" and "trusted recommendations" that are "coded and converted into computable form in a knowledge graph" Compl. ¶78 Compl. ¶80
  • The platform is positioned as foundational infrastructure for AI-enabled clinical reasoning systems and is allegedly used for BMJ's own products and by its customers Compl. ¶78 Compl. ¶80 The complaint asserts that it competes directly with Evidium's own platform and that BMJ mimicked Evidium's marketing language Compl. ¶77 Compl. ¶79
  • No probative visual evidence provided in complaint.

IV. Analysis of Infringement Allegations

The complaint references claim charts attached as Exhibits O, P, Q, R, and S, but these exhibits were not provided for this analysis Compl. ¶90 Compl. ¶100 Compl. ¶111 Compl. ¶122 Compl. ¶133 In lieu of a table, the narrative infringement theory is summarized below.

  • '674 Patent Infringement Allegations
    The complaint's narrative theory alleges that BMJ's Clinical Intelligence platform practices the patented method. It alleges BMJ took its own medical content (the "medical evidence source"), processed it using technologies like NLP to extract clinical concepts ("diagnostic factors"), and stored the results in its "BMJ knowledge graph" (the "database") Compl. ¶78 Compl. ¶80 This process, Evidium contends, is the same as its patented method for transforming narrative evidence into structured, computational elements. The infringement allegation rests on the premise that BMJ, after learning of Evidium's proprietary methods during their collaboration, created a system that performs the same steps to achieve the same result of making medical evidence computable Compl. ¶¶79-80

  • '882 Patent Infringement Allegations
    The infringement theory for the '882 patent is that the BMJ Clinical Intelligence platform is a system embodying the patented invention. The complaint alleges BMJ's platform constitutes a computer system that creates and uses a "dynamic knowledge graph" (the "graph database") Compl. ¶80 It further alleges that this system supports APIs, which corresponds to the '882 patent's requirement for APIs that allow an application to integrate with the database to "generate a healthcare determination" Compl. ¶58 Compl. ¶80 The core of the allegation is that the entire BMJ Clinical Intelligence architecture-from the knowledge graph to its external interfaces-is a system built according to Evidium's patented design Compl. ¶79

  • Identified Points of Contention:

    • Scope Questions: A central issue may be whether BMJ's "knowledge graph" embodies the specific structure claimed in the patents. For instance, the patents claim a structure where elemental factors are stored independently of their condition-specific context to enable reuse and disambiguation (('674 Patent, col. 10:49-51); Compl. ¶49). A point of contention may be whether BMJ's system actually creates such "elemental" and "reusable" concepts, or if it uses a different data structure that does not meet this limitation.
    • Technical Questions: The case may turn on evidence of how the BMJ Clinical Intelligence platform actually functions. For the '674 method patent, what evidence shows that BMJ's process involves the claimed human-in-the-loop step of a "person" reviewing and grouping NLP-generated factors (('674 Patent, claim 1))? For the '882 system patent, what is the specific function of the accused system's API, and does it "generate a healthcare determination" as the claim requires, or does it perform a more basic data retrieval function?

V. Key Claim Terms for Construction

  • The Term: "diagnostic factor... wherein an entry of each of said diagnostic factors in said database does not include information specific to said medical condition" '674 Patent, claim 1

  • Context and Importance: This limitation is the technical core of the claimed invention, defining the "elemental" and "reusable" nature of the computational evidence. The outcome of the case may depend on whether the nodes in BMJ's knowledge graph are structured this way. Practitioners may focus on this term because it distinguishes a specific, inventive data architecture from a generic knowledge graph where concepts might be stored with context.

  • Intrinsic Evidence for Interpretation:

    • Evidence for a Broader Interpretation: The specification describes "diagnostic factors" broadly to include signs, symptoms, and vital signs, and notes that they may be "caused by a condition" (('674 Patent, col. 9:10-24); ('674 Patent, col. 9:59-62)). This could support an argument that any clinical concept node in a graph is a "diagnostic factor."
    • Evidence for a Narrower Interpretation: The specification repeatedly emphasizes that these factors are made "elemental and independent" by separating out condition-specific attributes into linked "demographics" and "factor group" nodes (('674 Patent, col. 11:4-11); ('674 Patent, Fig. 4B)). This supports a narrower construction requiring a specific, multi-node data structure where the factor node itself is context-free.

  • The Term: "generate a healthcare determination" '882 Patent, claim 1

  • Context and Importance: This term defines the function of the claimed system's API, which was a key addition to overcome a §101 (patent eligibility) rejection during prosecution for related patents Compl. ¶¶58-59 Its construction is critical to determining if the API for BMJ's Clinical Intelligence platform performs the claimed function.

  • Intrinsic Evidence for Interpretation:

    • Evidence for a Broader Interpretation: The term itself is not explicitly defined, which could leave room for it to encompass any health-related output, such as retrieving a relevant document.
    • Evidence for a Narrower Interpretation: The complaint highlights prosecution history for the related '883 patent, where the applicant and examiner agreed that using an API to "generate a medical determination" was a software construct a human cannot practically perform in their mind Compl. ¶59 This suggests the term requires a computational output beyond simple data retrieval, such as a diagnosis, a ranked list of conditions, or a next-best action recommendation as described in the specification (('882 Patent, col. 2:56-63)).

VI. Other Allegations

  • Indirect Infringement: The complaint alleges inducement of infringement for all five patents. The basis for this allegation is that BMJ, with knowledge of the patents from pre-suit notice letters, "encourages or enables its customers to use" the accused Clinical Intelligence platform in the United States, thereby acting with specific intent to cause infringement by others Compl. ¶91 Compl. ¶101 Compl. ¶112 Compl. ¶123 Compl. ¶134
  • Willful Infringement: Willfulness is alleged for all five patents. The claims are based on alleged pre-suit knowledge. The complaint specifies the dates Evidium provided actual notice of infringement for each patent: December 22, 2024 for the '674 patent; April 8, 2026 for the '882 and '883 patents; August 24, 2026 for the '274 patent; and knowledge as of the issue date of September 1, 2026 for the '716 patent, following prior notice Compl. ¶94 Compl. ¶105 Compl. ¶116 Compl. ¶127 Compl. ¶138 The complaint alleges BMJ continued its infringing activities despite this knowledge.

VII. Analyst's Conclusion: Key Questions for the Case

  • A core issue will be one of architectural equivalence: does BMJ's "dynamic knowledge graph" implement the specific data structure required by the claims, where clinical concepts are abstracted into "elemental" factors stored independently of their condition-specific context? The case may turn on whether BMJ's system is merely a graph of linked medical terms or if it embodies the patents' specific, multi-node architecture for enabling computational reuse and disambiguation.
  • A second central question will be one of patent eligibility: given the complaint's extensive pre-emptive arguments regarding 35 U.S.C. § 101, the court will likely need to decide if the claims are directed to the abstract idea of organizing medical information, or to a concrete, technological improvement in computer functionality. The analysis will likely focus on whether the claimed combination of NLP, a specific graph structure, and APIs provides a specific, non-abstract solution to a technical problem inherent in prior art computer systems.
  • A key evidentiary question will be one of proving infringement beyond the narrative of copying: while the complaint presents a compelling story of misappropriation, the patent infringement claims will require technical proof. A critical question is what evidence Evidium will be able to discover and present to show that the accused BMJ Clinical Intelligence platform, as sold and used, practices each limitation of the asserted claims, particularly the specific API functions and internal data structures.