#财经新势力#
Author / Braised under the stars
Editor/Spinach Starfield
Typesetting/sandwiches under the stars
ChatGPT is definitely the most explosive science and technology news in recent years, this landmark technology is expected to comprehensively reshape the status quo of human society, artificial intelligence is also known as the fourth scientific and technological revolution after steam, electricity, information technology. As a high-profile frontier scientific and technological front, the medical field is naturally widely and deeply connected with "AI+", and is expected to usher in a moment of change by riding the east wind of artificial intelligence.
In fact, "medical + AI" has long been not limited to the concept stage, but has some practical implementation in multiple aspects of medical treatment. Specific to A-shares, not only Runda Medical (603108) and Yidu Technology (02158) and other manufacturers providing information solutions, Jinyu Medical (603882), Anbiping (688393) and Dean Diagnostics (300244) and other manufacturers have achieved business layout in terms of AI-assisted pathological diagnosis and hospital informatization, WuXi AppTec (603259), CXOs such as Hengrui Pharmaceutical (600276) and original pharmaceutical companies have also said that they have been paying attention to and developing the application scenarios of AI technology in new drug research and development.
1. Empower the entire medical industry chain
First of all, we need to answer the question of what "medical + AI" is.
In the author's opinion, as long as artificial intelligence technology can actually reach and realize the medical industry link that empowers or improves efficiency, it can be said that it has achieved "medical + AI" in the true sense.
For example, on the R&D side, AI can greatly improve R&D efficiency and reduce R&D costs through interventional drug screening and assisted design. For example, on the manufacturing side, AI can optimize the production management process and accelerate product iteration; On the medical side, AI-assisted imaging diagnosis has already achieved maturity in cervical cancer and other diseases.
"Medical + AI" scenario
At present, under the big concept of medical AI, there are several directions that can have clear commercialization prospects.
Second, policy promotion, market expansion
In fact, the promotion of artificial intelligence in the medical field has been shouted at the policy level for six or seven years, from the "New Generation of Artificial Intelligence Development Plan" in 2017 to last year's "14th Five-Year Plan" National Health Plan, the State Council and other issuing agencies have proposed to combine AI and medical equipment and other industries, and even clearly stated that they will promote qualified artificial intelligence products to enter the clinical trial stage.
Driven by this, the market size of domestic medical AI is indeed continuing to expand, with a total plate of 6.625 billion yuan in 2020. The CAGR will be 39.4% from 2020 to 2025 and will exceed 30 billion yuan in 2025. With a compound growth rate of close to 40%, "medical + AI" is definitely a fertile ground for imagination.
2019-2025E China's main medical AI application market scale (100 million yuan) Source: iResearch, China Business Industry Research Institute, Huaan Securities Research Institute
Among these subdivisions, the author believes that pathological diagnosis may be one of the first "bonanzas" in the field of medical AI.
As the slogan of graded diagnosis and treatment becomes louder and louder, the construction of grassroots pathology departments has also been put on the agenda. At present, primary medical institutions often lack pathological ability, and most of them complete pathological diagnosis in the form of sending for examination. However, the delivery mode is too limited, and can only solve temporary short-term clinical needs, and it is difficult to meet the needs such as intraoperative testing. So what about this? Perhaps AI-assisted remote pathological diagnosis will be a feasible path.
Pathological diagnosis is a diagnostic method based on image information, and in recent years, with the promotion of many diagnostic technologies and the investment of high-end equipment, the number of diagnoses has increased exponentially. In this context, the current situation of a serious shortage of registered pathologists in China has become more and more prominent. According to the figures of the Health Statistics Yearbook, if calculated according to the bed allocation requirements in 2022, 9.5-190,000 pathologists are needed in China, but how many registered pathologists are there in China? There are only 20,400 people, which is several times the gap between supply and demand.
Growth of practicing (assistant) pathologists in China Source: China Health Statistics Yearbook, Soochow Securities Research Institute
Because pathologists are difficult to train, the number will definitely not be significantly improved in the short term, and artificial intelligence-assisted reading technology is particularly important for domestic medical institutions. In fact, because the area occupied by pathology is generally only less than 1%, a lot of energy needs to be focused on "sieve yin" under the traditional pathological reading. Today, AI-assisted pathological diagnosis can initially realize the demand, helping doctors save time greatly. Even a study by Tongji Hospital showed that in the identification of benign and malignant lesions, the judgment of AI models and pathologists has not been much different, and the diagnosis time will be significantly shorter than that of pure manual judgment.
Third, auxiliary case diagnosis manufacturers contend
Since the application scenarios are relatively clear, many domestic companies are vying to deploy AI-assisted pathological diagnosis. In recent years, more than a dozen auxiliary diagnostic software in the fields of lung, cardiovascular and fundus have been successfully certified, and more products are also in the clinical and application stage, and the competitive landscape has been very lively.
Status of obtaining certificates for three types of AI medical imaging products (as of March 2023) Source: National Medical Products Administration website, Soochow Securities Research Institute
For example, the AI-assisted cervical cancer screening model jointly developed by Jinyu Medical and Huawei has a reading accuracy rate of 99%, and the artificial intelligence-assisted diagnosis products for cervical cytology jointly developed by Anbiping and Tencent are also entering high-end channels such as third-class hospitals.
In addition, pathological diagnosis itself also has a trend of automation, standardization and streamlining, which is also conducive to the landing of AI. It can reduce manual intervention in the process of film production and reading, improve sample consistency, and improve the accuracy of pathological artificial intelligence diagnosis.
But is the market really as good as it seems? Not really.
At present, the functions of AI-assisted pathological diagnosis related products are still relatively limited and homogeneous. While accuracy has improved considerably, it has yet to meet more expectations from clinicians. For example, patients who find nodules are often anxious, how can AI help doctors more accurately help patients understand their own situation? How are follow-up requirements determined? At present, AI still can't give answers.
In addition to listed companies, SenseTime (HK 0020) and other manufacturers in the field of artificial intelligence are testing the water for pathological diagnosis. After all, the financing environment in the past two years has not been ideal, and these companies need real money to give investors an explanation.
In the author's opinion, AI-assisted pathological diagnosis still needs to break through information silos and achieve more complete functions with more powerful computing power and abundant data volume, which is not something that a single enterprise can do. But after all, this track has achieved the first exploration of commercialization, and we have reason to expect that "medical + AI" can bring more surprises to mankind in the near future.
Note: This article does not constitute any investment advice. The stock market is risky, and you need to be cautious when entering the market. There is no harm without buying and selling.