In recent years, the rapid development of China’s medical device industry, at present there are about 77,000 valid registration certificates of medical devices and more than 37,000 records of medical devices. With the continuous emergence of new technologies and new products, the current medical device classification system has been unable to fully meet the needs of the medical device industry development and regulatory work. The shortcomings of the “Medical Device Category Directory” promulgated in 2002 in China have become increasingly prominent.
On August 31, 2017, the State Food and Drug Administration (CFDA) released a new version of the “Medical Device Classification Catalog”, which will come into effect on August 1, 2018. The original 43 subdirectories of the “Catalog” are reduced to 22.
Although the new version of the directory has been streamlined, but the content is more refined, from the original 15 pages to more than 150 pages, and more than two thousand products intended use and product attributes were refined description; products The number of name examples also increased six-fold to 6609, of which the proportion of medical imaging equipment increased significantly.
Among them, it is especially noticeable that it has added a category corresponding to artificial intelligence-assisted diagnosis, which is embodied in the catalog of analysis and processing of medical images and pathological images.
With the rapid development of medical informatization, more and more enterprises are beginning to devote themselves to the medical big data industry. Intelligent medical products are beginning to pay more and more attention. For the new edition of Medical Device Catalog, some experts conclude that the new version is more convenient for timely correction, refinement and updating. The “update” is mainly due to the ever-changing and endless stream of software and equipment for artificial intelligence-assisted diagnosis , The intervention of medical treatment also increased gradually. In the high-tech industry where machine learning and in-depth learning are becoming more and more popular, using artificial intelligence to perform imaging three-dimensional segmentation, pathological image analysis and processing, and personalized precision medical treatment to assist doctors in diagnosis and treatment planning has become increasingly popular.
With development, there will be limited development. AI assisted medical diagnosis of doctors have a certain guiding role, of course, to assume a certain degree of risk, with the intervention of intelligent products increased, the greater its safety responsibility, the laws and regulations will also be more stringent control. If you want to profit, it is necessary to cooperate with the hospital, to achieve legal fees profit, CFDA certification is unavoidable. Public information shows that in the current medical artificial intelligence business, in addition to two CFDA certified companies – Wuhan Landing and EDDA technology, almost all businesses are still free to provide trial phase.
According to the latest classification provisions, if the diagnostic software through the algorithm to provide diagnostic advice, only the auxiliary diagnosis function, does not directly give the diagnosis, then declare the second class of medical equipment, if the lesion site automatically identified and provide a clear diagnosis, In accordance with the third category of medical device management.
It is noteworthy that the third category of medical devices is the need to do clinical trials, the second category of equipment exempt from clinical trials directory, diagnostic software to declare whether the exemption can be exempted, CFDA has not yet made specific norms.
Relative to artificial intelligence in the field of medical imaging, artificial intelligence in voice input, data structure and the secondary application of structured, drug development and other fields of application threshold is relatively low. In these scenarios, artificial intelligence is only a tool, voice input and data structure does not require CFDA certification, and in the field of new drugs, a mature reporting process, artificial intelligence is only to speed up drug discovery and clinical trials, so in In these areas, the arrival of artificial intelligence is relatively easy.
It is understood that the artificial intelligence technology used in these areas of the company, most payers are B-side customers, because they do not require CFDA certification, but also significantly improve efficiency, so their profitability model is also easy to achieve.
Public information shows that, of the current medical artificial intelligence enterprises, only Wuhan Landing and EDDA Technology obtained CFDA certification, and therefore, the two companies have bid farewell to the free trial phase, all customers need to pay them. They can either sell the system directly to a healthcare facility or set up a cloud platform to deliver services to more hospitals and deliver benefits while delivering services.
Most companies currently do not get the CFDA certification, including IBM’s Watson has not yet received the US FDA certification, but its services are within the framework of the law allowed. Companies are now accumulating large amounts of clinical data by putting their products into hospitals for free trials to improve the accuracy of clinical applications, provide a solid foundation for reporting CFDA certifications, and narrow the difference between clinical and laboratory results.
The specification will be implemented on August 1, 2018. If all medical artificial intelligence companies want to go to the hospital to purchase this road, then obtain CFDA certification is the only way to certification if the three types of medical equipment, or diagnostic software Without the exemption, a large amount of real clinical data will provide tremendous help for the company’s application.
With the advent of classification catalogs, various medical artificial intelligence companies at the present stage should speed up the CFDA certification process while researching and developing products so that they can reach cooperation with device companies and medical institutions under relatively equal conditions in the process of marketization. In the premise of the possibility of profitability, to protect their own brand.