A number of large models have been rolled out in the field of traditional Chinese medicine, and the "AI old Chinese medicine" is coming?

With the advent of the era of AI inclusiveness, large models have been "rolled" to various industries. According to incomplete statistics, since 2023, more than 20 related large models have been released in the traditional Chinese medicine industry.

Behind the rush of the large model of traditional Chinese medicine is the urgent need for traditional Chinese medicine to enter the modernization in terms of inheritance and innovation. Although the large model of traditional Chinese medicine has been explored in imaging diagnosis and intelligent diagnosis, due to the complexity of traditional Chinese medicine theory and practice, the exploration of digital intelligence of traditional Chinese medicine is still in its infancy.

A number of interviewed experts believe that in terms of drug research and development, there are still many challenges in the accumulation of basic clinical data and talent training, and more clinical research units and front-line doctors need to be involved, and the real data of large model manufacturers and clinical research can be deeply integrated in order to truly exert value.

The large model is rolled into the field of traditional Chinese medicine

In many traditional Chinese medicine hospitals, AI intelligent TCM diagnosis and treatment instruments have been collecting patients' faces and tongues, and after AI intelligent recognition, combined with the knowledge in the database, they can give physical fitness analysis reports and customized health suggestions. There are even tea shops in Nanjing that have invited "Cyber Old Chinese Medicine" to launch AI tongue and face diagnosis and pulse diagnosis to match and recommend tea to customers.

"The technology of artificial intelligence provides us with a more effective methodology." Professor Ge Guangbo, executive vice president of the Institute of Interdisciplinary Sciences of Shanghai University of Traditional Chinese Medicine, told the first financial reporter that at present, Shanghai University of Traditional Chinese Medicine has developed a modern traditional Chinese medicine four-diagnosis instrument in combination with artificial intelligence, which is to combine the picture information collected from the clinic, and then combine it with artificial intelligence to classify the symptom type.

According to incomplete statistics, since 2023, more than 20 related large models have been released one after another. Behind these large models of traditional Chinese medicine, there are a large number of universities and scientific research institutes, and there are also large manufacturers such as Huawei and Baidu.

In terms of universities, Fudan University and Tongji University jointly developed the Zhongjing University of Traditional Chinese Medicine language model. East China Normal University, Shanghai University of Traditional Chinese Medicine, East China University of Science and Technology, Naval Medical University, Lingang Laboratory, and China Resources Jiangzhong National Key Laboratory of Modern Chinese Medicine have jointly developed the "Digital Intelligence Qihuang" traditional Chinese medicine model. The National Supercomputing Center in Tianjin, together with the Haihe Laboratory of Modern Chinese Medicine, Tianjin University of Traditional Chinese Medicine, Tianjin University, and Xinchuang Haihe Laboratory, has developed the "Tianhe Lingshu Model".

In terms of large manufacturers, Huawei and Tasly have jointly built a "Digital Intelligence Materia Medica Model". Chengdu University of Traditional Chinese Medicine, Baidu and Taiji Group jointly released the "Materia Medica Think Tank", a large model of the whole industry chain of traditional Chinese medicine.

A number of interviewed experts said that China has a wealth of ancient classics of traditional Chinese medicine and the accumulation of doctors' experience, which provides a good foundation for "AI+". "Artificial intelligence can also generate some new applications in the discovery of pharmacodynamic substances and mechanisms of action in traditional Chinese medicine, in addition to discovering new indications in modern pharmaceuticals and in the research and development of innovative drugs." Ge Guangbo said.

Auxiliary diagnostic applications are easier to implement

Behind the rush of the large model of traditional Chinese medicine is the urgent need for traditional Chinese medicine to enter the modernization in terms of inheritance and innovation.

The "Opinions on Promoting the Inheritance, Innovation and Development of Traditional Chinese Medicine" issued in October 2019 clearly put forward the need to support the construction of the traditional Chinese medicine service system with informatization, implement the action of "Internet + Chinese medicine health services", "take the pulse" and "prescribe" for the development of traditional Chinese medicine, improve the level of informatization of traditional Chinese medicine, vigorously promote the digitization and informatization of the traditional Chinese medicine system, and promote the interconnection and information sharing between traditional Chinese medicine and the national health information platform.

In August this year, the State Administration of Traditional Chinese Medicine and the National Data Bureau jointly issued the "Several Opinions on Promoting the Development of Digital Traditional Chinese Medicine" (hereinafter referred to as the "Opinions"), which put forward 20 measures in terms of traditional Chinese medicine diagnosis and treatment services, talent training, and scientific and technological innovation.

At the recent academic conference on network pharmacology and artificial intelligence, Chen Kaixian, academician of the Chinese Academy of Sciences and researcher of the Shanghai Institute of Materia Medica, Chinese Academy of Sciences, said in his keynote speech that traditional Chinese medicine has accumulated a large number of complex data in the long-term development process, and it is completely possible to apply the latest modern science and technology to the inheritance, innovation and development of traditional Chinese medicine, and artificial intelligence will promote the wave of technological change in the innovation and development of traditional Chinese medicine.

Judging from the large models that have been released so far, the specific application scenarios of large models are still mainly used for auxiliary consultation, and there are also certain explorations in the originality and research and development of traditional Chinese medicine. Some experts have pointed out that at present, due to the development of wearable devices and the accumulation of imaging, artificial intelligence is relatively easy to break through in auxiliary diagnosis and treatment, and can also play a relatively large role. However, the industry's current expectation for large models is that AI can play a greater role in the research and development of traditional Chinese medicine.

According to Academician Chen Kaixian, Nanjing University of Information Science and Technology and Beijing Jiaotong University have carried out multidisciplinary and multi-unit cooperation research, using network medicine to reveal some common mechanisms of the therapeutic effect of TCM syndrome differentiation and treatment, providing new ideas for the scientific connotation of TCM syndrome differentiation and treatment, and exploring some studies applied to the field of TCM and solving TCM problems by artificial intelligence and big data.

Ge Guangbo said that compared with traditional chemical drugs, traditional Chinese medicine usually acts on multiple drug targets through multi-component and multi-signaling pathways, and then produces an integrated and synergistic effect. This kind of integration and synergy effect of traditional Chinese medicine makes it difficult to screen out a single component in the research of new traditional Chinese medicine, which can replace the pharmacological activity of the entire traditional Chinese medicine compound, and some innovation in methodology and research strategy is needed at this time.

High-quality clinical data in TCM is key

On November 24 this year, HUAWEI CLOUD and Tasly, together with experts and scholars from Shanghai University of Traditional Chinese Medicine, Guangzhou University of Traditional Chinese Medicine, Zhejiang University of Traditional Chinese Medicine, China Pharmaceutical University, and other universities and research institutes, launched the Digital Intelligence TCM Innovation Alliance, with a view to promoting the application and development of modern technology in the field of TCM innovation.

As early as 2021, HUAWEI CLOUD created a large model of Pangu drug molecules. Li Yin, General Manager of HUAWEI CLOUD's Scientific Computing Domain, mentioned in a recent speech that HUAWEI CLOUD hopes to integrate a large platform to attract users and developers around the world to focus on the platform to do things more simply. The "Digital Intelligence Materia Medica Model" jointly built by Tasly and HUAWEI CLOUD aims to make efforts in the field of traditional Chinese medicine research and development.

In the field of traditional Chinese medicine, thousands of years of experience and literature have been accumulated, but the role of traditional Chinese medicine on the human body is still in a relatively vague state. At present, the large model of traditional Chinese medicine proposed in China generally uses the traditional experience knowledge graph of traditional Chinese medicine as the training data, but how to use modern scientific methods to analyze the principle of action of traditional Chinese medicine, so that traditional Chinese medicine and target database can be corresponded one-to-one, so as to optimize and improve the clinical efficacy of traditional Chinese medicine on a scientific basis, is an urgent problem to be solved.

Many of the experts interviewed said that in the process of digital intelligence in the Chinese medicine industry, the underlying database and talent training are the two biggest difficulties, which require the joint efforts of the Chinese medicine industry, large model manufacturers and all aspects of society.

Fan Xiaohui, director of the Yangtze River Delta Smart Oasis Innovation Center of Zhejiang University and Changjiang Scholar Distinguished Professor, said bluntly, "In fact, it may be that our industry itself is not ready."

"First of all, large models need to have high-quality big data, which is still relatively lacking in our traditional Chinese medicine industry." Fan Xiaohui said that the text data of some classic ancient books is more abundant, but in terms of clinical trials and other data, it is still not well sorted out. At the same time, he said that many university teams, research institutes, including some teams organized by hospitals, are doing data accumulation, but these data are scattered in various places and lack standardization. "This also involves the difficulties of combining data and clinical practice, the personalization of data and other reasons, as well as the privacy and ethical aspects of patients in the data of the pharmaceutical industry."

As one of the initiators of the "Digital Intelligence Chinese Medicine Innovation Alliance", Professor Cao Yongbing, director of the Shanghai Institute of Integrated Traditional Chinese and Western Medicine Vascular Diseases, mentioned that the reason why he wants to promote the establishment and development of the Digital Intelligence Chinese Medicine Innovation Alliance and help large model manufacturers to connect with clinical units is that he hopes that large model manufacturers can closely integrate with Chinese medicine clinical units and front-line Chinese medicine, truly promote the collection and sorting of high-quality Chinese medicine clinical diagnosis and treatment data, and provide effective support for the large model of Chinese medicine. It plays an important role in the research and development of new drugs and the clinical diagnosis and treatment of traditional Chinese medicine.

Professor Cao Yongbing told Yicai that in 2019, he and clinicians began to explore how to establish a vascular disease database integrating doctors, scientists and patients to manage the diagnosis and treatment data of integrated traditional Chinese and Western medicine for vascular diseases, but due to various reasons, the project could not be implemented. In addition to the problem of funding and the formation of clinical research teams serving doctors, more important is the desensitization and sharing of clinical data, "which is more of a headache than scientific research funding." ”

At the same time, Cao Yongbing emphasized that it is very difficult to collect clinical data of traditional Chinese medicine, but what is more difficult is how to improve the quality of clinical data of traditional Chinese medicine, especially the data related to efficacy and safety, which is the most important problem to be solved in the development of traditional Chinese medicine models. "If the data obtained by the large model is based on the network, rather than real and valid data from the clinic, it is equivalent to mixing a lot of 'sand' in the large model, which will interfere with the output of the large model and make the errors continue to be amplified."

Cao Yongbing mentioned that clinical data must be real and effective, and many large models are out of touch with real scientific research without the participation of the clinical team when the underlying data is entered. "If there is no data, there are a bunch of large models empty, which is equivalent to building a lot of unfinished buildings." In addition to the large number of valid real data of clinical research, another difficulty lies in talent, and some universities have begun to explore the intersection of artificial intelligence and traditional Chinese medicine.

Yicai learned that Shanghai University of Traditional Chinese Medicine is preparing for China's first artificial intelligence research institute in the field of traditional Chinese medicine, and the pharmacy discipline of Zhejiang University has also been paying great attention to using interdisciplinary technology to promote the construction of the entire new drug research system from basic research to transformation and application to talent training.

Fan Xiaohui told reporters that to really do the digital intelligence of traditional Chinese medicine, it is necessary to have a group of people who not only understand the technology of artificial intelligence, but also have a relatively good understanding of scientific problems in the field of traditional Chinese medicine.

"The traditional training framework and system cannot keep up with the current demand, and if the talent training system cannot be reformed to meet this demand, it will be difficult to break through this bottleneck in the short term." Fan Xiaohui said that universities around the world also hope to train students in the direction of interdisciplinarity, "which requires not only the efforts of people in the traditional Chinese medicine industry, but also the efforts of the country and various industries." ”

(This article is from Yicai)