How to approve medical artificial intelligence (AI) products? "Many AI technology companies and investment institutions are struggling to ask."
Recently, in the 2018 China Biomedical Industry High-Quality Development Forum held in Beijing, the “Mechanical Experts, Enterprise Representatives and Regulatory Sub-Forums”, representatives of industry experts, business representatives and regulatory authorities attended the new era of medical device industry. New challenges and progress in the review and approval of innovative medical device evaluations were discussed.
The reporter learned from the forum that the medical device technology review department is exploring, researching and predicting the landing of AI products, and preparing for the new generation of AI medical devices to enter the review process.
AI medical device encounters bottleneck
“At present, AI has performed very well in academic papers, technical patents, social certifications, capital investment, etc., but it is difficult to achieve substantive breakthroughs in product application.” Fan Xiaodong, vice president of China Medical Device Industry Association, said bluntly.
"As of June 2018, there were 4,925 AI companies in the world, and China had 1,011, second only to the United States, ranking second. From the perspective of specific cities, Beijing topped the list with 395, and 287 ranked second in San Francisco. Fan Xiaodong said that the data of China's industrial information "Analysis of the future development trend of China's AI+ medical imaging industry in 2017" shows that from 2014 to 2017, the financing situation of domestic medical AI has increased year by year, and in 2017 it has hit an amazing 1.842 billion yuan. Value. However, this year it is obviously cold, especially after March and April, it can be described as bleak.
In this regard, Fan Xiaodong believes that the development of medical AI technology innovation enterprises must experience three major gaps between talents and technology, scenarios and products, and markets. In the past two years, with the increase of big data and computing power, medical AI has crossed the talent and technology gap, but it is currently blocked in the application-driven scenarios and mature products.
He said that the founders of Shuangchuang (innovation and entrepreneurship) who are now involved in AI have not had any relevant learning and experience in the medical field, and lacked the basic professional knowledge and standardization of the medical device industry. This is very likely Leading technology and products out of industry regulations and standards, unable to meet clinical actual needs, thus away from users and the market.
Create a platform to solve the problem
How to break through the shackles and let AI cutting-edge technology be widely used in clinical practice as soon as possible to serve the majority of doctors and patients? Recently, the Medical Device Technology Evaluation Center (hereinafter referred to as the “Approval Center”) is also trying to give an answer. ""Made in China 2025" clearly needs to develop and support the artificial intelligence industry, and the evaluation agency must play a role in this." Sun Lei, director of the Center for Inspection and Evaluation, said.
“So far, no medical AI product has been submitted to the review center and entered the review process.” Sun Lei previously pointed out the prosperous appearance of the medical AI field at the 3rd China Drug Regulatory Science Conference Medical Device Science Supervision Forum. The real situation underneath.
Only by relying on the concept to attract a large number of private placements, funds to save money, and finally completely evolved into capital market operations, resulting in the current medical AI industry is booming. The cold financing in the first half of this year shows that the capital side began to rationally look at product conversion and business transformation in this area.
It is reported that the previous review center organized research on more than 100 technology companies and research institutions that are conducting AI product research and development. "According to the general statement of enterprises, everyone is in the research and development stage. Most of the technical researchers and product developers of these AI products do not know how to transform these technologies into medical products to provide medical services." Sun Lei said Transforming AI technology into a health care product is not a simple matter.
To this end, the Laboratory Center and the China Biomedical Engineering Society have established an open AI medical device research platform, which includes well-known research and development institutions, enterprise representatives and experts involved in research and development.
Under the premise of clarifying the characteristics of medical products, technical research and development will be submitted to the review on the premise of clearing the technical evaluation points of medical devices, and become the two major visions for the research platform to play an early guiding role. Sun Lei bluntly said, "First of all, we want to help companies solve the problem of how AI products should be designed and developed to meet the basic requirements of medical and health products through this research platform. Secondly, through the joint research of all parties, the review requirements for AI products are formed. In this case, when a medical device product enters the review and approval process, the company and the reviewer are very clear about all the guiding principles and review points, which not only greatly enhances the efficiency of the review process, but also saves the enterprise. R&D costs."
He Weijun, the minister of the second review of the review center, also said: "Based on this research platform, we can integrate the superior resources of all parties to carry out related research work."
"If we can achieve the above two points, the AI medical device research platform will make a huge contribution to the application of domestic AI industry technology to the health field." Sun Lei has high hopes for the platform.
Benign interaction and win-win cooperation
"It is undeniable that AI technology has a far-reaching impact on the medical device industry and is also the development direction of everyone's attention. Therefore, we must have a corresponding pre-judgment, keep abreast of the development of medical device core technology, and do a good job of pre-marketing. Technical preparation work." He Weijun said.
It is reported that at the end of 2017, the Detective Center led the establishment of the AI Research Working Group. At present, the working group has determined the key research directions of 7 medical AIs such as data set, algorithm evaluation, clinical evaluation, verification confirmation, analysis and research of overseas related AI product listing, data security, and registration change. On September 12th, the working group discussed the key points of medical software review for deep learning decision-making and formed a general draft review element.
AI medical devices are different from traditional medical device products. If you want to rely on an established evaluation standard to verify the risks of all AI medical products, you may ignore the product's own characteristics and technical points. In this regard, He Weizhen said that the relatively ideal state is that the reviewers began to pay attention to the technical points and product characteristics of the products from the product development status. The AI medical device research platform will help to further enhance the technical level of the review.
According to reports, the current laboratory has established cooperation with the Institute of Automation of the Chinese Academy of Sciences, Zhejiang University, Tsinghua University, and the Chinese Information and Communication Research Institute of the Ministry of Industry and Information Technology. At the same time, with the advantage of the AI medical device research platform docking enterprises, further research is carried out for specific products, and then specific review points and technical requirements are formed. More importantly, in-depth research on the technical difficulties of different products.
"This is a benign interaction between the two sides and win-win cooperation." He Weijun said that the so-called full life cycle supervision is to require the company to carry out research and development design, verification and confirmation with the relevant technical and regulatory requirements of medical devices to ensure its safety and effectiveness. The establishment of the AI medical device research platform enables the review department to provide targeted review opinions for enterprises to help enterprises improve the efficiency of technology conversion. The reviewers can deepen their understanding of product technology through communication with enterprises. Help regulators improve the level of supervision. (Reporter Zhang Xiaoyu)