The picture shows the scene of the 89th China International Medical Equipment (Spring) Expo Courtesy of the organizer
Medical devices are one of the important areas of AI application, and innovative products are constantly emerging. "Flexible" surgical robots, "smart" large models, and intelligent medical imaging equipment...... Walking into the scene of the 89th China International Medical Equipment (Spring) Expo (CMEF), which kicked off in Shanghai on April 11, the reporter seemed to enter the hospital scene in a science fiction novel.
As the practical application of AI in the field of medical devices gradually moves from scenario exploration to commercialization, the "scissors gap" between AI R&D investment and commercialization has become a thorn that enterprises have to face. Industry insiders said that in the research and development process of AI applications in the medical field, enterprises need to understand the actual needs of doctors and solve clinical pain points. From the perspective of the "AI + healthcare" industry chain, data collection, processing and analysis in the upstream, algorithm model development and optimization in the midstream, and medical service provision and marketing in the downstream need to work closely together to form an effective business model.
"AI + medical" products have become the focus of attention
From April 11th to April 14th, the CMEF was held in Shanghai with the theme of "Innovation and Technology, Smart Future". It is reported that nearly 5,000 brand companies from more than 30 countries and regions around the world brought tens of thousands of products to the show.
On April 11, the reporter interviewed a number of exhibitors at the scene, and "AI + medical" products became the focus of attention on the spot. "I am a traditional device maker, and I don't know much about AI applications, so I mainly want to see cutting-edge technologies at this exhibition. A lady who was visiting the United Imaging Group's booth told reporters. "I'm more interested in the application of AI in CT products, and as far as I know, this kind of product is selling relatively well at the moment. A medical device industry practitioner told reporters.
It is reported that at the CMEF, United Imaging Intelligence unveiled innovative achievements such as large medical models, more than 100 AI applications and more than 10 AI solutions.
In terms of medical large models, "'multimodality' is the 'highlight' of the development of large model technology. In the medical scenario, the cross-fusion of multi-modal information can train a large model that 'truly understands the medical scene'. Shen Dinggang, co-CEO of United Imaging Intelligence, said in an interview with a reporter from China Securities Journal that United Imaging Intelligence has launched a large model base in the vertical medical field "uAI Shadow Intelligence Large Model". Different from most of the existing medical models, the uAI image intelligence model is used as a base model, which can help in the development of text, image, and mixed-modal products.
At this exhibition, Neusoft Medical launched a variety of AI-driven intelligent high-end medical equipment, including the world's first 0.235-second ultra-wide-body CT, the industry's first dual-energy 3.0T magnetic resonance, the world's first one-stop smart catheterization laboratory, China's first 180-picosecond PET/CT, etc.
"Large-scale, high-quality data is one of the core elements of 'AI + Healthcare' product development. Ma Ruibing, director of Neusoft Medical Smart Imaging Software R&D Center, said in an interview with a reporter from China Securities Journal, "A large amount of data will be generated in the process of daily diagnosis of patients in hospitals, and these data will be precipitated in medical institutions, and we can help hospitals give full play to the value of these data, such as the medical data management platform we have done, which can help hospitals establish multi-modal data management capabilities for a certain disease." ”
In addition, the relevant staff of Mindray Medical told reporters that based on clinical scenarios, the company has launched the MC series automatic blood cell morphology analyzer (referred to as "AI film reader") and ultrasound full-stack intelligent solutions.
The application space is vast
In recent years, with the rapid iterative upgrading of artificial intelligence technology, "AI + healthcare" has entered the fast lane of development, and drug discovery and medical imaging are the two most important areas of AI application.
Xu Chao, project manager of Sullivan's life science division in Greater China, told reporters that on the whole, the current investment in the research and development of "AI + medical" products is mainly focused on auxiliary diagnosis and image recognition, intelligent medical equipment and personalized medicine. From the perspective of subdivisions, AI medical imaging is a popular application field in recent years, and pulmonary nodules and fundus screening are two disease areas that are widely deployed by enterprises.
Talking about the direction of "AI + medicine" with development potential in the future, Ma Ruibing said that the first is personalized medicine, that is, through the integration of patient medical history, examination data, medical imaging information, etc., it may be possible to obtain a more accurate diagnosis, so as to provide more effective treatment options.
For example, there is no relatively effective treatment for Alzheimer's disease, but the earlier it is detected, the earlier it can be intervened, and the time of Alzheimer's disease can be delayed through exercise, diet, drugs, etc.
The third is the fusion of multimodal data, with the rapid development of medical models, the combination of multimodal information, that is, images and texts, can bring greater imagination to AI-assisted diagnosis.
Zheng Wei, an analyst at Guolian Securities, said that in the field of AI imaging, it is expected to continue to expand coverage in the future, including ophthalmology, ultrasound, pathology, dermatology, electroencephalogram room, etc. It is estimated that the size of the AI medical imaging market in mainland China will reach 92.3 billion yuan in 2030.
Balance economic benefits and R&D investment
In the process of commercialization of "AI + medical" products, the "scissors gap" between AI R&D investment and commercialization has become a thorn that enterprises have to face. According to industry insiders, since January 2020, Keya Medical has obtained Class III medical device certificates, marking the commercialization of AI medical devices in China. As of June 30, 2023, 63 products have successfully obtained Class III certificates. More and more Class III certificates have been approved, which also means that the development focus of the industry is shifting from R&D to commercialization.
"At present, AI medical products are indeed facing the dilemma of high R&D investment and difficulty in commercialization, which is mainly due to various factors such as technology maturity, business model, and regulatory policies. First of all, from the perspective of business model, the R&D and commercialization of AI medical products need to build a complete industrial chain, including upstream data collection, processing and analysis, midstream algorithm model development and optimization, and downstream medical service provision and marketing. In this process, all parties need to work closely together to form an effective business model.
Secondly, technological breakthroughs are also the key to solving the problem of commercialization. At present, AI medical products need to be further improved in terms of accuracy, stability, and reliability of algorithms. In addition, it is necessary to strengthen the research and application of data governance and privacy protection technologies to ensure the security and compliance of patient data.
In addition, regulatory policies are also an important factor affecting the commercialization of AI medical products. With the rapid development of AI technology, regulatory authorities need to formulate and improve corresponding laws, regulations, and standards to provide strong institutional guarantees for the commercialization of AI medical products.
From the perspective of enterprises, how to solve the problem of the difficulty of landing the "AI + medical" scenario? Shen Dinggang said that on the one hand, it is necessary to work closely with doctors to understand the pain points of clinical needs. In practice, there may be situations where you think AI is a good application, but it is not actually needed in clinical practice. "We will send product managers to extensively investigate the situation of the hospital, and communicate with the directors of relevant departments to understand the actual needs, in addition, we work closely with industry, academia, research and medicine to jointly undertake major national projects, and find and solve a number of clinical pain points. ”
Ma Ruibing also said that in the field of auxiliary diagnosis, for example, the current AI-assisted diagnosis related products provide relatively simple capabilities, usually only one type of disease, but when doctors see a doctor, they judge the type of disease from the abnormality. In this field, many AI manufacturers are also looking for differentiated ways to survive, such as transforming to assist decision-making, and in a sense, they have also knocked on the door of another market, expanding from imaging to clinical departments in practice, and playing a greater value.
On the other hand, "the iteration of the 'AI + medical' field is very fast, and it cannot keep up with the iteration speed, and it is easy to be eliminated." In this regard, our R&D strategy is to make a base. The so-called base means that we have a lot of parts in it, and put it together, so that we can quickly get different applications, thereby reducing the cost of research and development. With the big model, our overall development speed has increased a lot. Shen Dinggang said.