Page 129 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

2 Use Case Description

2�1 Description

• Objectives: Yunnan Baiyao is committed to optimizing the entire industry chain through 4.1-Healthcare
AI technology, achieving digitalization, visualization, and traceability throughout the entire
value chain, from medicinal herb cultivation to R&D, production, and service. Utilizing
the multi-dimensional capabilities of large models, such as knowledge retrieval, agile
computing, and interactive experiences, it is applied in various scenarios including
traditional Chinese medicine (TCM) R&D, decision support, knowledge popularization,
and digital marketing. It provides services for diverse groups: It enables health-conscious
individuals to access wellness guidance, patients to obtain disease management and
basic medication instructions, Chinese herbal medicine growers to enhance cultivation
techniques, sales professionals to master product knowledge, primary TCM practitioners
to receive diagnostic support, and students to acquire specialized knowledge in TCM.
• Pain point: The quality of TCM R&D data is uneven. The large models are prone to
forgetting during incremental training. Data is scattered and insufficiently used. Traditional
methods make it difficult to achieve in-depth data mining of the whole industry chain.
Lack of ability to handle complex knowledge and scenarios.
• Solution: Based on the high-quality data accumulated by Yunnan Baiyao, Leigong large
model learns tens of millions of high-quality TCM documents, millions of structured
prescriptions and TCM drug knowledge graphs, and builds a vector knowledge base.
With Huawei Pangu's AI computing power foundation, AI platform, and development
tools, it has brought about the sharing and integration of AI technologies around the
world.
• The Chinese herbal medicine industry has deployed AI applications, including intelligent
identification and sorting—achieving 30-50x efficiency gains and 300% accuracy
improvements over traditional methods. AI-driven planting guidance and smart cultivation
monitoring have reduced annual return costs for Poria cocos (‐‐) alone by over 50 million
CNY. These advancements elevate quality control and standardization, strengthening
global trust and acceptance of traditional Chinese medicines.
• Chinese medicine assists decision-making, completes model training (such as incremental
pre-training), data processing (ancient, semi-white and semi-ancient), data proportioning,
and data synthesis applications, and solves problems such as data quality and large model
incremental training forgetting in the research and development process of Chinese
medicine.
• The knowledge science service enables comprehensive querying and interpretation
of TCM knowledge (including composition of medicinal materials and symptom
explanations), improving knowledge acquisition efficiency by 3-5 times. With its traceable
knowledge system and rigorous verification mechanism, the service significantly reduces
hallucination issues in large language models, increases effective knowledge utilization
rate, and enhances patient satisfaction by 30-50%. This approach effectively addresses
international misunderstandings and biases towards TCM, facilitating the global
dissemination and cultural exchange of traditional Chinese medicine.

Use Case Status:

The use case is part of a larger research project. Currently, the intelligent TCM herb recognition
system, automated herb sorting system, and TCM knowledge science application have been
developed and deployed. We regularly analyse problem feedback for project iteration and
optimization and collaborate with institutions such as the Chinese Academy of Medical Sciences
and the University of Traditional Chinese Medicine to continuously improve and revise the
training data for large models.

93