Prior art

To identify the most relevant prior art for US patent 12142371, I will examine the "References Cited" section of the patent itself, as listed on Google Patents and corroborated by the USPTO's public search tool (if available, which it is). Prior art is typically identified in a patent's "References Cited" section, which includes patents, publications, and other materials that the examiner and applicants considered relevant to the patentability of the claimed invention.

A recent news release from Hippocratic AI on October 1, 2025, and November 27, 2024, mentions that Patent 12,142,371 covers key components of their Polaris system, including its constellation architecture where a main model drives the conversation while other models provide support for accuracy and safety in healthcare. This context will be helpful when evaluating the descriptions of prior art.

Here's an analysis of the prior art cited in US Patent 12142371 (US12142371B1):

Non-Patent Literature Cited:

The patent lists several "OTHER PUBLICATIONS" as references. These are often particularly relevant for rapidly evolving fields like AI.

1. Gao et al. "Retrieval-Augmented Generation for Large Language Models: A Survey."

   • Full Citation: Gao et al. "Retrieval-Augmented Generation for Large Language Models: A Survey," Shanghai Research Institute for Intelligent Autonomous Systems, Tongji University, Mar. 27, 2024, 21 pages. [Url: arXiv:2312.10997].
   • Publication/Filing Date: March 27, 2024.
   • Brief Description: This survey covers Retrieval-Augmented Generation (RAG) for large language models (LLMs). The patent itself mentions RAG as an AI framework used to retrieve facts from an external knowledge base to provide accurate and up-to-date information to the LLM, grounding it on external sources to supplement the LLM and allow output to be checked for accuracy.
   • Potential Anticipated Claim(s) (35 U.S.C. § 102): This reference is highly relevant to the concept of augmenting LLMs with external knowledge for improved accuracy and up-to-dateness. This could potentially anticipate aspects of claims related to the conversation interface utilizing external knowledge or a knowledge graph for generating responses, as described in the abstract and the general description of the system's function (e.g., how the LLM accesses a knowledge graph to reduce latency). The discussion of RAG in the patent (specifically in relation to the RAG module 234 in FIG. 2) directly relates to this publication.

2. Karan et al. "Large Language Models Encode Clinical Knowledge."

   • Full Citation: Karan et al, "Large Language Models Encode Clinical Knowledge," Google Research, Dec. 26, 2022, 44 pages. [arXiv:2212.13138].
   • Publication/Filing Date: December 26, 2022.
   • Brief Description: This publication discusses how large language models (LLMs) encode clinical knowledge. The patent extensively describes using LLMs for specialized healthcare-related functions, such as preventative screenings, intake, scheduling, pre-op, discharge, and chronic care, and states that the AI-based virtual assistant may perform tasks normally handled by medical professionals.
   • Potential Anticipated Claim(s) (35 U.S.C. § 102): This reference suggests that the concept of LLMs possessing and utilizing clinical knowledge was known prior to the patent's priority date (May 15, 2023). This could potentially anticipate claims, particularly Claim 1, that describe an LLM-based conversation interface emulating a human healthcare professional and engaging in healthcare-related conversations.

3. Peter et al. "Benefits, Limits, and Risks of GPT-4 as an AI Chatbot for Medicine."

   • Full Citation: Peter et al, "Benefits, Limits, and Risks of GPT-4 as an AI Chatbot for Medicine. The New England Journal of Medicine, Mar. 30. 2023, 7 pages [N Engl Med 388;13].
   • Publication/Filing Date: March 30, 2023.
   • Brief Description: This article evaluates the benefits, limitations, and risks of using GPT-4 as an AI chatbot in medicine. This directly addresses the application of AI, specifically large language models, in a healthcare context, including potential challenges. The patent itself highlights issues with current AI virtual assistants in healthcare, such as incapability of following a checklist, verbosity, and lack of compliance with medical safety laws.
   • Potential Anticipated Claim(s) (35 U.S.C. § 102): This reference demonstrates prior awareness and discussion of AI chatbots (like GPT-4) in medical applications, including their shortcomings. This could potentially anticipate broad aspects of claims related to conversational AI in healthcare, particularly the recognition of problems that the claimed invention aims to solve.

4. Tao et al. "Towards Conversational Diagnostic AI."

   • Full Citation: Tao et al, "Towards Conversational Diagnostic AI," Google Research, Jan. 11, 2024, 46 pages. [arXiv:2401.05654].
   • Publication/Filing Date: January 11, 2024.
   • Brief Description: This publication focuses on conversational diagnostic AI. The patent's focus on a multi-turn conversational system in healthcare, particularly one that can identify symptoms and probe medical conditions, aligns with the concept of diagnostic AI.
   • Potential Anticipated Claim(s) (35 U.S.C. § 102): Given the publication date is after the priority date of US12142371B1 (May 15, 2023), this document cannot be prior art for the current patent application under 35 U.S.C. § 102 (unless it claims an earlier priority date that predates May 15, 2023, which is not indicated here). However, if it were considered prior art, it would potentially anticipate claims related to conversational AI systems for medical diagnosis, particularly those involving symptom identification and medical condition queries, which are aspects of Claim 1.

Patent Literature Cited (as mentioned in the PGR2025-00075 filing):

The Post-Grant Review (PGR) case PGR2025-00075 against US Patent 12142371 explicitly cites other U.S. Patents as prior art. This is highly significant.

1. US Patent 9,824,188 B2

   • Full Citation: U.S. Patent No. 9,824,188 B2.
   • Publication/Filing Date: Not explicitly stated in the provided text for the filing date, but the grant date for a B2 patent would precede its citation in a PGR.
   • Brief Description: The PGR petition states that "U.S. Patent No. 9,824,188 B2 teaches the control logic comprising a trigger detection logic, a question insertion logic, and an answer classification logic (see Fig.1 (102,104), Fig.3 and Col.9, Line 21-31, receiving a query, determining . . .)". This directly corresponds to components of Claim 1.
   • Potential Anticipated Claim(s) (35 U.S.C. § 102): This patent directly anticipates elements of Claim 1, specifically the "control logic" including "trigger detection logic, a question insertion logic, and an answer classification logic." The PGR filing suggests this reference makes Claim 1 obvious when combined with other prior art.

2. US Patent 11,843,565 B2

   • Full Citation: U.S. Patent No. 11,843,565 B2.
   • Publication/Filing Date: Not explicitly stated in the provided text for the filing date.
   • Brief Description: The PGR petition cites this patent in combination with US Patent 9,824,188 B2 to argue obviousness for claims 1-20 of US12142371. Without a detailed description of US Patent 11,843,565 B2, its specific teachings cannot be fully ascertained from the provided text.
   • Potential Anticipated Claim(s) (35 U.S.C. § 102): The PGR filing indicates that this patent, in combination with US Patent 9,824,188 B2, is used to argue obviousness for claims 1-20. This suggests it addresses some aspects of conversational AI systems, potentially relating to multi-turn interactions, control mechanisms, or other general AI conversational elements.

3. U.S. Patent Publication No. 2024/0185968 A1

   • Full Citation: U.S. Patent Publication No. 2024/0185968 A1.
   • Publication/Filing Date: Not explicitly stated in the provided text for the filing date.
   • Brief Description: This publication is also cited in the PGR petition in combination with US Patent 9,824,188 B2 to argue obviousness.
   • Potential Anticipated Claim(s) (35 U.S.C. § 102): Similar to US Patent 11,843,565 B2, this publication is cited in combination to argue obviousness for claims 1-20. It would likely touch upon aspects of conversational AI or related technologies that, when combined with 9,824,188 B2, would render the claims of 12,142,371 obvious.

Summary of Most Relevant Prior Art:

Based on the information available, the most relevant prior art appears to be:

• US Patent 9,824,188 B2: This patent is directly cited in the PGR for teaching specific elements of the "control logic" in Claim 1 of US12142371B1, namely trigger detection, question insertion, and answer classification logic. This indicates a strong potential for anticipating or rendering obvious key aspects of Claim 1.
• Karan et al. "Large Language Models Encode Clinical Knowledge" (2022): This non-patent literature directly addresses the core concept of LLMs possessing clinical knowledge for healthcare applications, which is fundamental to the described invention.
• Peter et al. "Benefits, Limits, and Risks of GPT-4 as an AI Chatbot for Medicine" (2023): This publication demonstrates the existing landscape and challenges of applying advanced AI chatbots like GPT-4 in medicine, providing a context of known problems that US12142371 aims to address.

The other cited patent references (US 11,843,565 B2 and US 2024/0185968 A1) are mentioned in the context of being combined with US 9,824,188 B2 to argue obviousness, suggesting their relevance as part of a combination. The Gao et al. paper on RAG is relevant to the technical implementation details of how the LLM accesses external knowledge, which can be seen in the RAG module (234) of the AI architecture. The Tao et al. publication is not considered prior art due to its publication date being after the priority date of US12142371B1.