Guideline: Retrospective data provide support for clinical, launch standard database, post-marketing clinical, technical black box, major product change declaration, Sino-US approval difference, China's new generation medical AI product approval and long-term. The good news is that the standard test database is being established, and the standard data set (the fundus part) has been completed in 3 months. Some enterprises will soon get the registration test results issued by the Chinese inspectorate. A standard database of pulmonary nodules has also been launched. The specific criteria that the arterial network knows from the enterprise are described below.
In 2018, the development of the medical artificial intelligence industry is still fast. In the first quarter, companies such as Huiyi Huiying, Pushu Technology, Shenrui Medical, Airdoc, and Vision Medical announced that they have completed a new round of financing, almost all at the level of 100 million yuan. Lian Ying also released its own open intelligent AI platform, which fully penetrates the medical AI field.
According to the arterial network, although the industry is developing rapidly, there is a problem that has always restricted the development of the industry. That is, there is no new generation of medical AI products from a company that has obtained a medical device registration certificate. This aspect is due to the fact that medical AI has a certain level of intelligence as a brand new product, but there has not been a database of experience and standards for the approval of such products. On the other hand, the products of domestic medical AI companies are still in the grinding stage, and medical care is a rigorous industry, which is related to the safety of the people.
It is worth noting that most of the products that have been certified before are obtained based on the previous CAD product approval method, which is somewhat different from the recent popular new generation medical AI products (this article explores a new generation of medical artificial intelligence products).
Without certification, there is no market access qualification. Although each startup has its own legitimate and legal revenue channels and high financing, AI company's data, talent, computing power, and operating costs are expensive. If medical AI products are the main business. The company has not been able to obtain market access qualifications, and the long-term development of the company will face challenges.
Why medical AI products developed with deep learning technology can't get certification, where is the problem, and what problems need to be solved after obtaining certification, what preparations are made by various medical AI companies, regulatory agencies, etc. What is the difference between China and the United States? The arterial network tries to solve these puzzles.
1. Why can't China issue certifications like the US?
In early April 2018, the FDA approved the first IDx-DR software program developed by IDx for autonomous artificial intelligence diagnostic equipment for first-line medical care. It can view retinal photos without the involvement of a professional doctor. Diagnosis of diabetic retinopathy.
The arterial network learned that this seemingly simple diagnosis, IDx has gone for 21 years. IDx is a company that specializes in the development of clinically autonomous diagnostic algorithms. It took seven years for IDx to communicate with the FDA on how to evaluate the system and ensure its accuracy and safety.
Dr. Zhou Shaohua, the chief expert of medical imaging analysis in Siemens, told reporters that the US regulatory system is like this.
The FDA regulates medical devices in three categories based on the possible use of the device and the risks it poses to the patient. The first class I is a low-risk device such as medical gloves; the second class II is a medium-risk device such as a CT machine; and the third class III is the highest-risk device such as a bracket.
There are two types of Al imaging systems: computer-aided detection (CADe) and computer-assisted diagnostic (CADx). CADe is used to detect abnormalities in abnormal conditions, while CADx assesses the presence or absence of disease, such as the severity, classification, or prediction of a disease.
The FDA has a lot of experience in supervising CADe software and provides 510(k) indicative criteria for how to perform clinical performance evaluations. However, the FDA has traditionally used the CADx system as class III.
(The 510(k) document is a pre-market application file submitted to the FDA to demonstrate that the device being applied for is as safe and effective as a legally listed device that is not affected by pre-market approval (PMA), ie, substantially equivalent The applicant must compare the application for listing with one or more similar devices on the US market to arrive at and support the conclusion of the equivalent device.)
However, according to the arterial network, it is known that the medical AI products currently approved by the FDA are mostly class II. That is to say, the approval has been passed by lowering the threshold. From the current situation, China is positioning the next-generation medical AI as three types of medical devices, so the approval will be different.
2. It is recommended that each company consider certification of three types of medical devices.
Guo Na, co-founder of Huiyi Huiying, said that several AI products recently approved by the United States are all undergoing Class II certification process, which proves safety and effectiveness by comparing with traditional CDSS (Clinical Decision Support System). The regulations of the China Food and Drug Administration are relatively stricter, and the path control of clinical evaluation is very strict. According to the current regulations, basically domestic AI products have to follow the evaluation route of clinical trials, which will take a long time. At present, the Food and Drug Administration has not issued such a signal to reduce the approval criteria. According to the new classification catalogue, the assistant decision-making software that directly gives the diagnosis and treatment recommendations belongs to three categories. If only quantitative values, such as bone density, are given, For medical devices, most of the AI ​​products on the market fall into three categories.
Yasen Technology told the Arterial Network that although it has obtained a second-class certificate after nearly two years of work, it is still in the process of preparing for the license of the three types of certificates according to the latest regulations. On September 4, 2017, the CFDA released a new version of the "Medical Devices Classification" medical software included in the auxiliary diagnosis, mainly for the auxiliary diagnostic products that have appeared before, such as CAD systems. However, this type does not include an automatic diagnostic system, and the definition of the recently popular artificial intelligence software products is not perfect.
This catalogue is re-established by the relevant departments based on the previous experience and issues of classification, approval, and certification of auxiliary diagnostic products. For the new generation of medical artificial intelligence products, there is no precedent to find. At this stage, although artificial intelligence companies have applied for certification by the Food and Drug Administration, no company has received three types of medical devices.
The arterial network has learned from various companies that several major companies such as Yasen, Huiyi Huiying, Imagination, Shenrui, Airdoc, and Yitu Medical are actively applying for the declaration of three types of medical devices. According to the company, their product matrix is ​​doing three types of certification. In addition, Airdoc sent the first server in China to load the artificial intelligence AI software to be tested.
A key box is a special box for storing keys. It is usually made of cold rolled steel. It is a kind of steel office furniture. The biggest difference between it and the ordinary box is that there are many hooks on the key box for hanging the key. Using a key box to store keys can effectively store and classify keys. So that enterprises can be high quality, efficient management.
The wall mounted key box is a secure storage solution designed to hold and organize keys. It typically features a durable metal construction with a lock or combination lock mechanism to prevent unauthorized access. Key boxes come in various sizes and styles, ranging from small boxes that can hold a few keys to larger cabinets that can store hundreds of keys. They are commonly used in commercial buildings, hotels, car rental agencies, and other establishments where multiple keys need to be securely stored and easily accessible. Key boxes help to prevent key loss, theft, and unauthorized access, ensuring the safety and security of the premises.
Steel Key Box ,Steel Cash Safe,Small Tin Key Box,Steel Cash Coin Box,Wall Mounted Key box
Hebei Hupai Aodi Cabinet Industry Co.,Ltd. , https://www.hupaiaodi.com