MModal LLC C O'Duane Morris LLP v. Nuance Communications Inc

1. Case Identification

• Case #: IPR2018-01355
• Patent #: 7,379,946
• Filed: July 5, 2018
• Petitioner(s): MModal LLC
• Patent Owner(s): Nuance Communications, Inc.
• Challenged Claims: 1-6

2. Patent Overview

• Title: Categorization of Information Using Natural Language Processing and Predefined Templates
• Brief Description: The ’946 patent relates to a computer-implemented method for generating a report from an input data stream, typically from medical dictation. The system uses natural language processing (NLP) techniques to identify and extract "latent information" from the data stream to automatically activate and populate a relevant report template.

3. Grounds for Unpatentability

Ground 1: Claims 1-6 are obvious over Taira in view of Buchanan

• Prior Art Relied Upon: Taira ("Automatic Structuring Of Radiology Free-Text Reports," a 2001 article) and Buchanan (Patent 5,267,155).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that Taira taught a method for generating structured reports from unstructured, free-text radiology reports, which inherently contain "latent information" (e.g., the existence, location, and properties of a medical finding). Taira's NLP system was alleged to perform all steps of independent claim 1: receiving an input data stream (the radiology report), processing it to identify latent information, activating a template ("structured frame"), populating it, and generating a report (the formalized frame output). Petitioner specifically mapped Taira's "structural analyzer," "lexical analyzer," "parser," and "semantic interpreter" to the claimed NLP steps of identifying relevant sections, bounding sentences, classifying information into topics (e.g., "abnormal findings"), and normalizing the text (e.g., standardizing terminology). Dependent claims were argued to be obvious as Taira's reports inherently involved medical data (claim 2), predetermined classes of information (claim 3), and medical problems (claim 4).
  • Motivation to Combine: To the extent Taira's "structured frames" were not considered formal "reports," Petitioner contended a person of ordinary skill in the art (POSITA) would combine Taira with Buchanan. Buchanan disclosed a document generation system that used templates with "holes" to create formal reports, including medical billing reports, from stored information. A POSITA would be motivated to use the structured data extracted by Taira's system to populate the formal report templates of Buchanan. This combination would address the shared problem of converting unstructured physician input into structured documents suitable for insurance companies and regulatory agencies. For claim 5, this combination would allow for the generation of a medical billing report.
  • Expectation of Success: A POSITA would have a high expectation of success as both references addressed automated medical report generation using conventional techniques, making the integration of Taira's data extraction with Buchanan's report formatting a predictable implementation.

Ground 2: Claims 1 and 5 are obvious over Heinze, alone or in view of Taira

• Prior Art Relied Upon: Heinze (Patent 6,915,254) and Taira.
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner asserted that Heinze, which was cited during prosecution, taught a system for automatically assigning medical codes from transcribed physician notes that rendered the challenged claims obvious. Heinze's NLP engine was argued to disclose the key processing limitations added to claim 1 during prosecution to achieve allowance. Specifically, Heinze’s system performed "note segmentation" (identifying and bounding relevant portions), "parsing" to tag phrases into categories like "Disposition" (classifying), and "morphological and lexical processing" to standardize word forms (normalizing). Heinze also taught generating reports (e.g., billing reports, claim forms) containing latent information (e.g., billing codes, demographic data) extracted from the free-text notes and activating templates based on this extracted information.
  • Motivation to Combine: If Heinze was deemed not to disclose the specific NLP processing steps, a POSITA would have been motivated to substitute the well-described NLP engine from Taira into Heinze's system. Both NLP engines performed the same function of extracting latent medical information from unstructured text. Given Taira's touted benefits of accuracy and precision, this would have been a predictable substitution of one known component for another to improve a known system. This combination would result in the method claimed in the ’946 patent.
  • Expectation of Success: As the combination involved substituting one known NLP method for another to perform the identical function in the same technical field, a POSITA would have been confident of success.

4. Arguments Regarding Discretionary Denial

• Petitioner argued that discretionary denial under §325(d) for Ground 2, which relied on the previously cited Heinze reference, was unwarranted. The core argument was that the Patent Office committed "clear error" during prosecution by failing to properly apply Heinze to the limitations of then-pending claim 6. Petitioner asserted that these limitations, which were critical for allowance and incorporated verbatim into the issued independent claim 1, were expressly disclosed by Heinze, yet the Examiner allowed the claims without providing any substantive analysis of these features in view of the reference.

5. Relief Requested

• Petitioner requested the institution of an inter partes review (IPR) and the cancellation of claims 1-6 of the ’946 patent as unpatentable.