DCT
1:18-cv-00901
MModal Services Ltd v. Nuance Communications Inc
I. Executive Summary and Procedural Information
- Parties & Counsel:
- Plaintiff: MModal Services Ltd. (New Jersey)
- Defendant: Nuance Communications, Inc. (Delaware)
- Plaintiff’s Counsel: Duane Morris LLP; Latham & Watkins LLP
- Case Identification: 1:18-cv-00901, N.D. Ga., 02/28/2018
- Venue Allegations: Plaintiff alleges venue is proper in the Northern District of Georgia because Defendant has a "regular and established place of business" in the district and has committed alleged acts of infringement there.
- Core Dispute: Plaintiff alleges that Defendant’s speech recognition and clinical documentation products infringe four patents related to remote speech recognition architecture, automated document compliance checking, text replacement, and interactive data verification.
- Technical Context: The technologies at issue operate in the domain of cloud-based speech recognition and computer-assisted physician documentation (CAPD), a significant market focused on improving the efficiency and accuracy of clinical record-keeping in the healthcare industry.
- Key Procedural History: The complaint alleges a long history between the parties, noting that Defendant has made three proposals to acquire Plaintiff or its predecessor companies since 1998. Subsequent to the filing of this complaint, two of the asserted patents underwent inter partes review (IPR) at the U.S. Patent and Trademark Office. In IPR2019-00499, all asserted claims of the ’829 Patent, including independent claim 1, were cancelled. Similarly, in IPR2019-00496, all asserted claims of the ’524 Patent, including independent claim 1, were cancelled. These cancellations may render the infringement counts for the ’829 and ’524 patents moot.
Case Timeline
| Date | Event |
|---|---|
| 2006-05-25 | ’524 Patent Priority Date |
| 2007-06-21 | ’040 Patent Priority Date |
| 2010-05-11 | ’040 Patent Issue Date |
| 2011-06-13 | ’786 Patent Priority Date |
| 2011-06-19 | ’829 Patent Priority Date |
| 2013-04-02 | ’524 Patent Issue Date |
| 2014-07-15 | ’829 Patent Issue Date |
| 2015-12-08 | ’786 Patent Issue Date |
| 2018-02-28 | Complaint Filing Date |
II. Technology and Patent(s)-in-Suit Analysis
U.S. Patent No. 9,208,786 - “Speech Recognition Using Loosely Coupled Components,” issued December 8, 2015
The Invention Explained
- Problem Addressed: The patent addresses the challenge of providing high-quality, efficient automatic speech recognition (ASR) across a growing variety of computing devices like smartphones, tablets, and desktops (Compl. ¶31). Prior systems often required a microphone to be physically connected to the computer performing the ASR, or involved transmitting raw audio over a network to a remote server, which could degrade quality and introduce latency (’786 Patent, col. 1:11-2:40).
- The Patented Solution: The invention describes a system architecture with "loosely coupled" components that can be distributed across different devices. An "audio capture component" on a user's local device (e.g., a smartphone) captures speech and can optimize the audio signal specifically for recognition before it is sent to a remote "speech recognition processing component." A "context sharing component" then identifies the correct "result processing component" (e.g., an application running on a desktop) to receive the transcribed text, based on the user's current context (Compl. ¶¶27, 32-33; ’786 Patent, Abstract).
- Technical Importance: This distributed architecture allows for more flexible and efficient use of computing resources, enabling scenarios like using a smartphone as a high-quality wireless microphone for dictation into an application on a separate computer without a direct physical connection (Compl. ¶33).
Key Claims at a Glance
- The complaint asserts at least independent claim 1 (Compl. ¶34).
- Essential elements of Claim 1 include:
- An audio capture component for capturing a user's speech.
- A speech recognition processing component for performing ASR.
- First and second result processing components to receive the recognition results.
- A "context sharing component" that identifies which result processing component is associated with the user's current context.
- The context sharing component further comprises means for identifying a list of authorized components and means for determining that a component in the list is associated with the user's current context.
- A "speech recognition result provision means" for sending the results to the identified component.
U.S. Patent No. 8,781,829 - “Document Extension in Dictation-Based Document Generation Workflow,” issued July 15, 2014
The Invention Explained
- Problem Addressed: In medical dictation, prior art systems were often limited to producing either unstructured free-form text or requiring users to tediously provide input for structured documents field-by-field. These systems lacked a way to automatically check documents generated from free-form speech for compliance with clinical "best practices," such as including specific details required for billing or quality of care metrics (Compl. ¶¶67-68; ’829 Patent, col. 2:14-25).
- The Patented Solution: The patent describes a method where a structured document is generated from a user's dictated speech. The system then analyzes this document to determine if it complies with a "plurality of best practices." Based on this analysis, the system can automatically insert required content or generate prompts to the user, guiding them to provide additional information needed to make the document compliant (’829 Patent, Abstract; Compl. ¶66).
- Technical Importance: This invention aimed to increase the accuracy and completeness of clinical reports, which is critical for patient safety, regulatory compliance, and obtaining proper reimbursement, while also educating clinicians on documentation requirements (Compl. ¶¶67-68).
Key Claims at a Glance
- The complaint asserts at least independent claim 1 (Compl. ¶69). Note: Claim 1 of the ’829 Patent was cancelled in IPR2019-00499.
- Essential elements of Claim 1 include:
- Applying ASR to an audio signal to produce a structured document.
- Determining if the document complies with a plurality of best practices to produce a conclusion.
- Inserting content into the document based on the conclusion.
- Generating a first indication for the user to provide input to conform the document to a first best practice.
- Generating a second indication for the user to provide input to conform the document to a second best practice.
U.S. Patent No. 8,412,524 - “Replacing Text Representing a Concept with an Alternative Written Form of the Concept,” issued April 2, 2013
Technology Synopsis
The patent addresses inefficiencies in editing dictated transcripts, where a transcriptionist might need to replace an expanded phrase (e.g., "cerebrovascular accident") with an abbreviation ("CVA") or vice-versa (Compl. ¶¶85-90). The invention provides a method for identifying a phrase in a document, displaying alternative written forms of the underlying concept to the user, and replacing the original phrase with a user-selected alternative upon receiving a simple instruction (e.g., a keystroke or voice command) (Compl. ¶¶82, 91).
Asserted Claims
At least Claim 1 (Compl. ¶98). Note: Claim 1 of the ’524 Patent was cancelled in IPR2019-00496.
Accused Features
The complaint accuses features in Nuance's Dragon products, such as the "correction menu," which allow a user to select a word or phrase and be presented with a list of alternatives for replacement via voice command or other input (Compl. ¶¶104, 107).
U.S. Patent No. 7,716,040 - “Verification of Extracted Data,” issued May 11, 2010
Technology Synopsis
The patent addresses the challenge of verifying the accuracy of structured data (e.g., allergies, diagnoses) that is automatically extracted and coded from unstructured, free-form dictation (Compl. ¶¶125-128). Instead of requiring a user to understand and verify complex underlying data codes (e.g., XML tags), the invention renders the corresponding text with a visual characteristic (e.g., highlighting or bolding) based on the code. The user can then confirm or deny the accuracy of the rendering, and the system modifies the underlying code based on that feedback, creating a more intuitive verification workflow ('040 Patent, Abstract; Compl. ¶129).
Asserted Claims
At least Claim 1 (Compl. ¶133).
Accused Features
The complaint accuses Nuance's CAPD solutions, which analyze physician notes, extract and highlight potential diagnoses or medications, and then prompt the user to confirm ("add") or reject ("remove") these findings, thereby allowing the user to verify the accuracy of the extracted data (Compl. ¶¶136, 139-140).
III. The Accused Instrumentality
Product Identification
The primary accused instrumentalities are Defendant’s cloud-based speech recognition and clinical documentation solutions, including Dragon Medical One, Dragon Medical Advisor, Dragon Anywhere, and Dragon Anywhere Group (Compl. ¶¶15, 18-19, 34, 47, 69, 98, 133).
Functionality and Market Context
- The complaint describes Dragon Medical One as a cloud-based speech recognition platform that allows clinicians to dictate patient narratives directly into Electronic Health Records (EHRs). It can be used with accessories like the PowerMic Mobile application, which turns a smartphone into a wireless microphone for the system (Compl. ¶¶15, 16, 37).
- Dragon Medical Advisor is alleged to be a Computer-Assisted Physician Documentation (CAPD) solution that integrates with Dragon Medical One. It is described as analyzing physician notes in real-time to identify and prompt for areas where additional specificity is needed to comply with billing codes or clinical quality metrics (Compl. ¶¶18, 72). A screenshot in the complaint shows Dragon Medical Advisor alerting a clinician that a diagnosis of "altered level of consciousness" requires additional specificity. (Compl. p. 51).
- Dragon Anywhere and Dragon Anywhere Group are described as cloud-based mobile dictation applications for iOS and Android devices, targeted at general-purpose professional documentation (Compl. ¶¶19, 48).
- The complaint positions these products as market-leading, alleging their use by hundreds of thousands of clinicians and a majority of U.S. hospitals (Compl. ¶¶34, 69).
IV. Analysis of Infringement Allegations
’786 Patent Infringement Allegations
| Claim Element (from Independent Claim 1) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| an audio capture component... comprising means for capturing a first audio signal... | The Dragon Medical One application running on a computer with a microphone, or the PowerMic Mobile application running on a smartphone. | ¶36 | col. 4:10-15 |
| a speech recognition processing component comprising means for performing automatic speech recognition... | The cloud-based speech platform of Dragon Medical One that receives audio and performs recognition. | ¶37 | col. 4:16-20 |
| a first result processing component... a second result processing component... | A processing device running an EHR on a desktop (first component) and mobile productivity applications running on a smartphone or in the cloud (second component). | ¶¶38, 39 | col. 4:21-27 |
| a context sharing component comprising means for identifying a first one of the first and second result processing components as being associated with a first context of the user... | The Dragon Medical One system uses a "unique Nuance Healthcare ID" to track the user, the device being used, and session information. A screenshot shows the Dragon Medical One login screen where user context is established. (Compl. p. 21). | ¶40 | col. 4:28-35 |
| means for identifying a list of at least one result processing component authorized for use on behalf of the user... | The Dragon Medical One platform is alleged to include a list of EHRs, applications, and other components associated with the user's Nuance Healthcare ID to track activity. A screenshot shows a "My Nuance Healthcare ID" website listing available mobile apps. (Compl. p. 23). | ¶¶42, 43 | col. 11:3-7 |
| means for determining that the at least one result processing component in the list is associated with the context of the user... | The system determines which applications (e.g., EHRs) are associated with the user's Nuance Healthcare ID upon login, thereby associating the active component with the user's current context. | ¶¶44, 45 | col. 11:8-11 |
| speech recognition result provision means for providing the first speech recognition results to the identified first one... | The Dragon Medical One system securely communicates the recognized speech data to the appropriate application, such as an EHR, using TLS encryption protocols. | ¶46 | col. 11:12-16 |
- Identified Points of Contention:
- Scope Questions: Claim 1 recites multiple elements in "means-plus-function" format (e.g., "means for identifying a list"). The scope of these elements is limited to the corresponding structures described in the patent specification and their equivalents. A key dispute may arise over whether Nuance’s "Healthcare ID" and associated infrastructure constitute equivalents to the specific server components and data structures disclosed in the ’786 patent.
- Technical Questions: A factual question may be whether the accused system's user authentication and session management performs the distinct, multi-step process recited for the "context sharing component"—specifically, first "identifying a list" of authorized components and then separately "determining" that one component from that list is associated with the current context.
’829 Patent Infringement Allegations
| Claim Element (from Independent Claim 1) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| (A) applying automatic speech recognition to an audio signal to produce a structured document... | Dragon Medical Advisor, as part of a cloud-based speech platform, receives and transcribes speech to create structured documents like physicians' notes. | ¶71 | col. 2:56-59 |
| (B) determining whether the structured document includes an indication of compliance for each of a plurality of best practices to produce a conclusion; | The system analyzes notes to identify where additional specificity is needed to comply with best practices for diagnosis and payment models. A screenshot shows an alert for an "altered level of consciousness." (Compl. p. 51). | ¶72 | col. 2:59-64 |
| (C) inserting content into the structured document, based on the conclusion, to produce a modified structured document; | Dragon Medical Advisor prompts the user with suggested text or provides an option to insert text to address a documentation deficiency. A screenshot shows a prompt to "Add the following sentence(s)..." (Compl. p. 53). | ¶73 | col. 3:1-3 |
| (D) generating a first indication that a user should provide additional input of a first type to conform the structured document to a first best practice...; and | The system generates alerts that flag potential missed diagnoses or details required to meet a first best practice, such as supporting a specific payment model. | ¶74 | col. 3:4-8 |
| (E) generating a second indication that the user should provide additional input... to conform the structured document to a second best practice... | The system is alleged to generate a second, distinct prompt to flag other missing details required to comply with a second best practice. A screenshot shows the system identifying two missing concepts: the type of pneumonia and sepsis. (Compl. p. 55). | ¶75 | col. 3:9-13 |
- Identified Points of Contention:
- Legal Questions: A dispositive issue is that the asserted independent claim 1 was cancelled during an inter partes review. This raises the question of whether this entire count is moot and subject to dismissal.
- Technical Questions: Assuming the claim were valid, a potential dispute is whether the accused system's prompts constitute generating distinct indications for a "plurality of best practices." The defense could argue that various prompts for more detail all serve a single "best practice" of ensuring sufficient specificity for billing, rather than two or more distinct practices as required by the claim.
V. Key Claim Terms for Construction
For the ’786 Patent:
- The Term: "context sharing component"
- Context and Importance: This term is central to the patent's contribution of a "loosely coupled" system. The construction of this term will determine whether Nuance's "Nuance Healthcare ID" system, which links a user's profile and applications across different devices, falls within the scope of the claims. The term is drafted in means-plus-function format, so its construction will depend on identifying the corresponding structure in the specification.
- Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The specification describes the component's function as identifying a result processing component "associated with a first context of the user at a first time," which could be read broadly to cover any system that tracks a user's active application session (’786 Patent, col. 4:28-35).
- Evidence for a Narrower Interpretation: The detailed description and figures depict the "context sharing component" as a specific software module with defined inputs and outputs that performs discrete logical steps, such as maintaining an explicit list of authorized components ('786 Patent, Fig. 2A, element 250; col. 11:3-11). This could support a narrower construction limited to that specific architecture.
For the ’829 Patent:
- The Term: "a plurality of best practices"
- Context and Importance: The claim requires determining compliance with multiple best practices and generating separate indications for at least two of them. The definition of this term is critical because if all of Nuance's system prompts are found to relate to a single, overarching "best practice" (e.g., "increase documentation specificity"), then infringement cannot be established, even if the claim were still valid.
- Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The specification describes best practices in general terms as "guidelines for providing complete, accurate, and useful information" ('829 Patent, col. 2:59-64). This could support an argument that different rules for achieving one goal (e.g., different specificity requirements for different diagnoses) each constitute a separate "best practice."
- Evidence for a Narrower Interpretation: The patent may provide examples suggesting that "best practices" are distinct categories of rules, such as one set for clinical accuracy and a separate set for billing compliance. This could support a narrower construction requiring the accused system to check against qualitatively different types of rules.
VI. Other Allegations
Indirect Infringement
The complaint alleges both induced and contributory infringement for all four patents. For inducement, it alleges Defendant encourages infringement by providing customers with user manuals, training, data sheets, promotional videos, and customer support services that instruct on using the accused products in an infringing manner (Compl. ¶¶59-60, 77-78, 110-111, 142-143).
Willful Infringement
While the complaint does not use the word "willful," it alleges facts that could support such a claim. It alleges Defendant had pre-suit knowledge of Plaintiff's technology and intellectual property assets since at least 1998, citing three past proposals by Defendant to acquire Plaintiff or its predecessors (Compl. ¶¶23-24). It further alleges knowledge of the specific asserted patents since at least the filing of the lawsuit (Compl. ¶¶62, 80, 113, 145).
VII. Analyst’s Conclusion: Key Questions for the Case
This litigation presents several distinct legal and technical questions for the court's determination.
- A primary threshold issue is one of mootness: given that the asserted independent claims of the ’829 and ’524 patents were cancelled in post-filing inter partes review proceedings, can the infringement counts associated with those patents survive in any form, or must they be dismissed?
- For the ’786 patent, a core issue will be one of structural equivalence: as the claims rely on means-plus-function limitations, the analysis will focus on whether Nuance’s "Healthcare ID" and cloud architecture contain structures identical or equivalent to the specific "context sharing component" disclosed in the patent’s specification, or if there is a fundamental mismatch in their technical implementation.
- For the ’040 patent, a key evidentiary question will be one of functional mapping: does the user workflow in the accused CAPD systems—where a clinician is prompted to "add" or "remove" a highlighted, machine-identified concept—perform the specific sequence claimed, which requires rendering data based on a code, receiving user input on the rendering's accuracy, and then modifying a "verification status" of the code itself?