YiduTech Unveils Proprietary Medical Large Language Model and Signs Strategic Cooperation Agreement with Huawei

On December 21, 2023, the “2023 Medical Intelligence Conference,” hosted by Yidu Cloud and co-organized by Huawei Technologies Co., Ltd. (hereinafter referred to as “Huawei”), was held in Beijing. Themed “Smartly Mapping Healthcare, Innovating for the Future,” the conference conducted in-depth discussions on the opportunities and transformations brought to the healthcare industry by new technologies such as artificial intelligence. At the event, Yidu Cloud officially launched its self-developed large model specialized for the medical vertical domain, marking China’s first professional large language model designed for multi-scenario applications within the medical vertical sector. This launch also heralded a comprehensive upgrade of Yidu Cloud’s healthcare scenario solutions. Meanwhile, Yidu Cloud and Huawei signed an agreement to deepen their strategic cooperation at the conference and jointly introduced smart healthcare solutions, accelerating the intelligent transformation of the healthcare industry.

Xu Jiming, CEO and Co-founder of Yidu Cloud, stated that unlike previous AI waves, a collective consensus has formed more rapidly during the current large model technology wave: large models can only demonstrate their true value by transitioning from general-purpose to vertical applications and swiftly achieving industrial implementation. Yidu Cloud firmly believes in the immense potential of medical large models, continuously promotes their innovative applications, strives to make medical technologies more accessible, and works hand in hand with partners to usher in a new chapter in intelligent healthcare.

 Address by Xu Jiming, CEO and Co-Founder of Yidu Cloud

Lu Yongping, President of Huawei’s Government and Enterprise Education and Healthcare System Department in China, stated in his address that Huawei continues to uphold the open and collaborative philosophy of “Partners + Huawei,” insisting on serving as the industry’s “black soil” by fulfilling its roles as an enabler and empowerer to support a diverse array of models and foster prosperity across all industries. By providing robust ICT infrastructure encompassing computing, storage, networking, cloud, and security, and by integrating and converging with industry partners at the technical and solution levels, Huawei supports software development partners in more efficiently and conveniently developing algorithms, software, applications, and tools, thereby facilitating the intelligent transformation and upgrading of the healthcare industry.

 Address by Lu Yongping, President of the Government and Enterprise Education & Healthcare System Department, Huawei China

Today, as the strategic intersection of “Digital China” and “Healthy China,” digital health has made the development of the digital health industry an inevitable requirement for the deep integration of the digital economy and the healthcare sector. Against this backdrop, artificial intelligence and big data technologies, particularly emerging large language model (LLM) technologies, are accelerating their deep integration with the medical industry.

At the conference, Mei Jianping, former Deputy Director-General of the Department of High-Tech Industrialization at the Ministry of Science and Technology and a First-Level Inspector, pointed out that the intelligent development of the healthcare sector faces challenges unique to this field, such as the protection of medical health data privacy, complex application scenarios, and difficulties in implementing practical models. However, he expressed his delight at witnessing continuous innovation in China’s AI-driven healthcare sector, which has given rise to outstanding innovative enterprises like Yidu Cloud. Mr. Mei highly praised the strategic partnership between Yidu Cloud and Huawei Technologies Co., Ltd., believing that it would contribute significantly to the formation and development of China’s independent technology ecosystem.

 Address by Mei Jianping, Former Deputy Director-General of the Department of High-Tech Industrialization and First-Class Inspector at the Ministry of Science and Technology

Wang Caiyou, Chairman of the Information Professional Committee of the Chinese Hospital Association, delivered a keynote presentation titled “Integration and Innovation in the Digital Era to Promote High-Quality Hospital Development.” He noted that digital technologies are increasingly permeating all aspects and processes of healthcare. This digital transformation poses new challenges for collaboration and innovation among governments, healthcare institutions, and technology enterprises. Policy frameworks require vertical integration and horizontal coordination, while healthcare institutions and technology companies, as well as businesses among themselves, must work synergistically to ensure that medical intelligence brings tangible benefits to both healthcare professionals and patients.

Speech by Wang Caiyou, Chairman of the Information Professional Committee of the Chinese Hospital Association

Feng Tianyi, Deputy Director (Presiding) of the Digital Health Department at the Cloud and Big Data Research Institute of the China Academy of Information and Communications Technology (CAICT), affirmed the potential of large AI models in the healthcare sector. He believes that large AI models will not only reshape the production and consumption patterns of digital content but also drive a significant leap in productivity. Currently, exploratory applications of large AI models have been implemented in nine major areas, including medical services, patient services, operational management, and clinical research. However, there are substantial differences across these fields in terms of application opportunities, impact, and maturity rates. In the future, standardization will be a key focus; for instance, high-quality healthcare datasets will significantly enhance the capabilities of large AI models.

 Speech by Feng Tianyi, Deputy Director (Presiding) of the Digital Health Department, Cloud and Big Data Institute, China Academy of Information and Communications Technology

Seizing the market opportunities presented by large model technology, Yidu Cloud is actively exploring the research, development, and application of healthcare-specific large models. At the conference, Yidu Cloud’s large model made its public debut.

 Yidu Cloud CTO and Chief AI Scientist Yan Jun Unveils Yidu Cloud’s Large Language Model

As a leading enterprise in China’s healthcare intelligence industry, Yidu Cloud has been deeply engaged in the field for nearly a decade. Its “Healthcare Intelligence Brain,” YiduCore, has been authorized to process and analyze over 4 billion medical records, accumulating a vast amount of multi-dimensional, quantifiable knowledge graphs. Leveraging its proprietary data generation technology, the company applies these knowledge graphs to the training of large language models. “The accumulation and consolidation over the past decade have established the core competitive barrier for Yidu Cloud’s large models,” said Yan Jun, CTO and Chief AI Scientist at Yidu Cloud.

Yidu Cloud positions its large language model as a new generation “Healthcare Super Brain,” providing easy-to-use, high-quality, and customizable robust support for various application scenarios in the healthcare industry.

Targeting consumer-facing users, Yidu Cloud’s large language model delivers professional, medical-grade personalized services. At the launch event, live comparative demonstrations showcased how Yidu Cloud’s large language model provides more precise and practical medical advice and treatment plans tailored to individual users’ needs and conditions. When responding to user inquiries, unlike general-purpose large language models that simply list possible options, Yidu Cloud’s large language model proactively asks follow-up questions, collecting sufficient decision-making factors through multi-round “consultations” before delivering more rigorous and well-founded responses. According to introductions, Yidu Cloud’s large language model has already been piloted in the “Huimin Bao” (city-specific supplemental medical insurance) sector. Jiangsu Provincial Huimin Bao, “Jiangsu Yihuibao No. 1,” is the first to integrate Yidu Cloud’s large language model, which will support its 24/7 health management services.

 Yidu Cloud Large Model Comparison Demonstration

Targeting B-end clients, Yidu Cloud’s large language model (LLM) not only leverages robust professional capabilities to enhance quality and efficiency across all stages of medical care, education, research, and management, but also enables every healthcare professional to easily build their own applications. Yidu Cloud’s LLM is the first large model in the healthcare sector to achieve full compatibility with Ascend AI. Leveraging Ascend AI hardware, the two parties have jointly developed an integrated training-and-inference appliance solution for B-end users. This solution incorporates a foundation model pre-trained on hundreds of billions of refined tokens and provides a complete toolchain covering data management, model management, training management, evaluation management, and application management. Users can rapidly customize their exclusive models and, after fine-tuning, empower their business systems with a single click.

According to reports, Yidu Cloud’s large language model has surpassed GPT-3.5 in performance evaluations across multiple well-defined medical tasks, including triage, basic medicine, general practice, clinical internal medicine, clinical surgery, and professional licensing examinations. Its accuracy and recall rates in data structuring tasks also significantly exceed those of large models with comparable parameter sizes. Currently, Yidu Cloud’s large language model has been deployed in collaboration with several top-tier hospitals ranked among the top 20 nationwide, empowering diverse scenarios such as medical research, clinical decision support, and data governance.

 Yidu Cloud Large Model Comparative Evaluation

Yan Jun stated that the training and implementation of large language models constitute a complex systemic engineering endeavor. Yidu Cloud aims to reduce this complexity and accelerate the deployment and innovation of large language models through comprehensive collaboration and solutions.

Leveraging its leading large language model (LLM) technology, Yidu Cloud has upgraded its solutions across multiple scenarios—including data governance, hospital management, clinical research, and clinical diagnosis and treatment—to accelerate the adoption of LLMs.
 
In terms of data governance, traditional methods rely primarily on manual labor and typically require 8–12 months. Yidu Cloud’s new-generation data center, YiduEywa 2.0, built on NLP and knowledge graph technologies, shortens this process to 4–6 months. At the launch event, YiduEywa was upgraded to version 3.0, leveraging large language models to achieve more efficient data understanding, structured normalization, and data quality control, thereby boosting data governance efficiency by over 50% once again.
 
Built on the YiduEywa 3.0 foundation, Yidu Cloud has comprehensively upgraded its products and solutions for application scenarios such as clinical research, clinical decision support, medical record quality control, and hospital management. The new-generation research data platform, empowered by large language models, provides comprehensive support to clinical researchers—from AI-powered literature reading and summarization and natural language-based medical record search to intelligent data processing, automated statistical analysis, and intelligent generation of manuscript drafts—accelerating the paper publication cycle from 6–12 months to 1–2 months. The hospital lean management platform, also empowered by Yidu Cloud’s large language models, helps hospitals comprehensively improve management efficiency through enhanced tool usability and professional intelligent advisory capabilities based on Copilot technology.
 
In empowering clinical diagnosis and treatment, Yidu Cloud is also leveraging large language models to enhance various aspects, including medical record documentation, extraction of disease characteristics, generation of assisted diagnostic and treatment plans, and post-treatment follow-up recommendations. These efforts aim to improve physicians’ work efficiency and the quality of medical care, thereby further enhancing patients’ healthcare experience.

Speech by Xu Jiming, CEO and Co-Founder of Yidu Cloud

Xu Jiming, CEO and Co-founder of Yidu Cloud, stated that the company’s large language model (LLM) will further empower multiple scenarios, including city-level comprehensive supervision of medical resources, infectious disease monitoring, early warning, and emergency response, as well as consumer-facing health services. The company will restructure its existing products and solutions using LLMs, while continuously innovating and exploring new applications to drive the upgrade and advancement of the entire healthcare ecosystem.

Regarding the deployment of large language models, a clear trend in the industry is that an increasing number of companies are embracing collaboration and interoperability to achieve win-win outcomes through joint ecosystem development.

At the conference, Yidu Cloud and Huawei Technologies Co., Ltd. signed an agreement to deepen their strategic partnership. This collaboration marks the launch of a comprehensive strategic ecosystem partnership between the two parties in the healthcare sector. Leveraging Yidu Cloud’s technological expertise and practical experience in medical intelligence, along with Huawei’s strengths in core areas such as computing power, software and hardware, and distribution channels, the partners will engage in joint technology research and development, market expansion, and industry co-creation. Together, they aim to drive the innovation and application of large healthcare models, develop industry-leading joint healthcare solutions, and accelerate the intelligent transformation and upgrading of the healthcare industry.

 Yidu Cloud Signs Cooperation Agreement with Huawei

Yidu Cloud’s collaboration with Huawei dates back to 2020. This year, the two parties officially launched joint innovation in medical large language models (LLMs). Leveraging the powerful computing capabilities of Huawei’s Ascend AI and Yidu Cloud’s professional expertise in medical intelligence, they have developed an integrated training and inference appliance solution tailored specifically for the healthcare sector.

At the press conference, Sun Pengfei, Vice President of Huawei’s Public Sector System Department, delivered a keynote address, sharing insights into the trends of intelligent healthcare and unveiling joint solutions focused on three key scenarios: medical large language models, smart hospitals, and smart healthy cities. These solutions include the Yidu Cloud Large Model Training and Inference All-in-One Machine, the Scientific Research Big Data Platform, and the Smart Healthy City Big Data Platform.

 Yidu Cloud and Huawei Release Joint Smart Healthcare Solution

In his speech, he stated that the release of this joint solution marks a new beginning for the deepened cooperation between the two parties. Moving forward, both sides will develop more competitive solutions through close collaboration and joint innovation, delivering greater value to industry clients and driving the sustainable development of the healthcare ecosystem.

 Speech by Sun Pengfei, Vice President of Huawei's Public Sector System Department

“We look forward to witnessing the continuous innovation and evolution of intelligent healthcare technologies, bringing greater transformation and convenience to the medical industry,” said Xu Jiming. Yidu Cloud will continue to dedicate itself to the research, development, and innovative application of large healthcare models, working jointly with partners to drive the expansion of medical technology into deeper and broader domains, thereby contributing to the building of a better future for digital health.