Research on Third-Class AI Medical Devices: Case Studies

Number of Registrations

Research on Third-Class AI Medical Devices: Case Studies

AI medical devices include independent software that is itself a medical device (AI SaMD) and medical devices containing AI software components (AI SiMD). Third-class medical devices are directly approved by the National Medical Products Administration, and the registration and regulation of third-class deep learning independent software have become an important benchmark affecting the development of the AI medical industry in China.

The fifth issue of China Digital Medicine in 2023 published a special article titled Analysis of Registration Status of Third-Class Deep Learning Independent Software Products by scholars including Peng Liang, summarizing the registration status of third-class deep learning independent software products in China over the past three years.

1. Registration Status of Third-Class Deep Learning Independent Software Products

Since the registration of the first product in January 2020, a total of 39 third-class deep learning independent software products have been registered (as of the end of December 2022), as shown in Table 1. Among them, 38 products are registered domestically, accounting for 97.4%; 1 imported product, accounting for 2.6%. The registrations were 9, 11, and 19 for the years 2020, 2021, and 2022 respectively, with only 1 imported product registered in 2020.

Table 1 Registration Status of Third-Class Deep Learning Independent Software Products (Items)

Registration Type	2020	2021	2022	Total
Domestic Products	8	11	19	38
Imported Products	1	0	0	1
Total	9	11	19	39

2. Types of Third-Class Deep Learning Independent Software Products

Based on intended use (see Figure 1), there are 16 items of software for auxiliary triage and assessment, accounting for 41.0%; 12 items of software for auxiliary detection, accounting for 30.8%; and 11 items of software for auxiliary diagnosis, accounting for 28.2%.

Based on the location of lesions (see Figure 2), there are 16 items of software for lung-related conditions, accounting for 41.0%; 7 items for heart-related conditions, accounting for 17.9%; 6 items for skeletal conditions, accounting for 15.4%; 5 items for eye-related conditions, accounting for 12.8%; 4 items for head and neck conditions, accounting for 10.3%; and 1 item for blood-related conditions, accounting for 2.6%.

Based on data types (see Figure 3), there are 36 items of image-based software, accounting for 92.3%; and 3 items of waveform data-based software, accounting for 7.7%. Among the image-based software (see Figure 4), there are 24 items of CT image-based software, accounting for 66.7%; 5 items each of X-ray image-based software and fundus image-based software, accounting for 13.9% each; and 1 item each of MR image-based software and microscope image-based software, accounting for 2.8% each. The waveform data-based software is all ECG-related.

Currently, the product types can be classified into 15 categories based on the three dimensions mentioned above, with an average of 2.6 products per category. If we distinguish between CT images and CT angiography images, X-ray images and X-ray contrast images, resting ECG data and dynamic ECG data, the product types can be further subdivided into 21 categories, with 10 types of CT image-based software, 4 types of X-ray image-based software, 3 types of fundus image-based software, 2 types of ECG-related software, and 1 type each of MR image-based software and microscope image-based software, with an average of 1.9 products per type.

Typical products include software for auxiliary detection of lung nodules in CT images, software for auxiliary triage and assessment of pneumonia in CT images, software for auxiliary diagnosis of diabetic retinopathy in fundus images, software for auxiliary triage of intracranial hemorrhage in CT images, software for auxiliary assessment of bone age in children’s hand X-ray images, and ECG analysis software, especially the first two products which have a high concentration, totaling 15 products.

3. Registrants of Third-Class Deep Learning Independent Software

A total of 23 enterprises have registered third-class deep learning independent software, among which 22 are domestic enterprises, with an average of 1.7 products registered per enterprise, as shown in Table 2; 1 imported enterprise has registered only 1 product.

From the perspective of geographic distribution, there are 8 enterprises in Beijing, registering 15 products; 6 enterprises in Shanghai, registering 10 products; 3 enterprises in Shenzhen, registering 5 products; 2 enterprises in Hangzhou, registering 5 products; and 3 enterprises in other regions, each registering 1 product.

Table 2 Registrant Information of Third-Class Deep Learning Independent Software

Currently, the AI medical device industry in China has entered a rapid development stage, showing a good development trend. The mainstream products of independent software are domestic products; however, the variety and quantity of products are relatively small, and some products have a high concentration, with an unreasonable distribution of products. It is likely that new AI medical device products and technologies will continue to emerge. In response to this situation, China needs to accelerate the formulation of guidelines for key products and strengthen the construction of information systems and product identification to improve regulatory requirements.

According to the query method mentioned in the article, the list of domestic products of third-class deep learning independent software in China is as follows (as of June 25, 2023, see Table 3).

Table 3 List of Domestic Products of Third-Class Deep Learning Independent Software

Registration Certificate Number	Product Name	Registrant Name
国械注准20203210035	Coronary Blood Flow Reserve Calculation Software	Keya Medical Technology Co., Ltd.
国械注准20203210080	ECG Analysis Software	Shenzhen Kaiwuer Electronics Co., Ltd.
国械注准20203210687	Diabetic Retinopathy Fundus Image Auxiliary Diagnosis Software	Shenzhen Silicon-based Intelligent Technology Co., Ltd.
国械注准20203210686	Diabetic Retinopathy Fundus Image Auxiliary Diagnosis Software	Shanghai Eagle Eye Medical Technology Co., Ltd.
国械注准20203210844	Coronary CT Angiography Image Vessel Narrowing Auxiliary Triage Software	Yukun (Beijing) Network Technology Co., Ltd.
国械注准20203210862	Fracture CT Image Auxiliary Detection Software	Shanghai United Imaging Healthcare Technology Co., Ltd.
国械注准20203210920	Lung Nodule CT Image Auxiliary Detection Software	Hangzhou Deep Insight Technology Co., Ltd.
国械注准20203210839	Lung Nodule CT Image Auxiliary Detection Software	Tuixiang Medical Technology Co., Ltd.
国械注准20213210177	Children’s Hand X-ray Image Bone Age Auxiliary Assessment Software	Hangzhou Yitu Medical Technology Co., Ltd.
国械注准20213210210	Pneumonia CT Image Auxiliary Triage and Assessment Software	Tuixiang Medical Technology Co., Ltd.
国械注准20213210211	Pneumonia CT Image Auxiliary Triage and Assessment Software	Hangzhou Deep Insight Technology Co., Ltd.
国械注准20213210308	Fracture X-ray Image Auxiliary Detection Software	Huiying Medical Technology (Beijing) Co., Ltd.
国械注准20213210471	Lung Nodule CT Image Auxiliary Detection Software	Shanghai United Imaging Healthcare Technology Co., Ltd.
国械注准20213210607	Pneumonia CT Image Auxiliary Triage and Assessment Software	Shanghai United Imaging Healthcare Technology Co., Ltd.
国械注准20213210612	Pneumonia CT Image Auxiliary Triage and Assessment Software	Tencent Medical Health (Shenzhen) Co., Ltd.
国械注准20213210911	Pneumonia CT Image Auxiliary Triage and Assessment Software	Beijing Ande Medical Technology Co., Ltd.
国械注准20213211007	Pneumonia CT Image Auxiliary Triage and Assessment Software	Yukun (Beijing) Network Technology Co., Ltd.
国械注准20213211094	Lung Nodule CT Image Auxiliary Detection Software	Hangzhou Yitu Medical Technology Co., Ltd.
国械注准20213211132	Quantitative Blood Flow Fraction Measurement Software	Shanghai Bodong Medical Technology Co., Ltd.
国械注准20223210295	Children’s Hand X-ray Image Bone Age Auxiliary Assessment Software	Hangzhou Deep Insight Technology Co., Ltd.
国械注准20223210309	Intracranial Hemorrhage CT Image Auxiliary Triage Software	Shanghai United Imaging Healthcare Technology Co., Ltd.
国械注准20223210482	Head and Neck CT Angiography Image Auxiliary Assessment Software	Yukun (Beijing) Network Technology Co., Ltd.
国械注准20223210570	Lung Nodule CT Image Auxiliary Detection Software	Yukun (Beijing) Network Technology Co., Ltd.
国械注准20223210575	Lung Nodule CT Image Auxiliary Detection Software	Huiying Medical Technology (Beijing) Co., Ltd.
国械注准20223210572	Chest Fracture CT Image Auxiliary Triage Software	Tuixiang Medical Technology Co., Ltd.
国械注准20223210625	Lung Nodule CT Image Auxiliary Detection Software	Shanghai Xingmai Information Technology Co., Ltd.
国械注准20223210687	Lung Nodule CT Image Auxiliary Detection Software	Guangxi Yizhun Intelligent Technology Co., Ltd.
国械注准20223210775	Intracranial Hemorrhage CT Image Auxiliary Triage Software	Tuixiang Medical Technology Co., Ltd.
国械注准20223211032	Lung Nodule CT Image Auxiliary Detection Software	Shanghai SenseTime Intelligent Technology Co., Ltd.
国械注准20223211098	Thoracic Vertebra CT Image Auxiliary Assessment Software	Siemens Medical Devices (Shanghai) Co., Ltd.
国械注准20223211102	Diabetic Retinopathy Fundus Image Auxiliary Diagnosis Software	Beijing Zhizhen Internet Technology Co., Ltd.
国械注准20223211081	Fundus Disease Fundus Image Auxiliary Diagnosis Software	Beijing Kangfuzi Health Technology Co., Ltd.
国械注准20223211099	Cardiovascular CT Image Auxiliary Assessment Software	Siemens Medical Devices (Shanghai) Co., Ltd.
国械注准20223211140	Chronic Glaucoma-like Optic Nerve Disease Fundus Image Auxiliary Diagnosis Software	Tencent Medical Health (Shenzhen) Co., Ltd.
国械注准20223211142	Dynamic ECG Analysis Software	Shenzhen Kaiwuer Electronics Co., Ltd.
国械注准20223211374	Pulmonary Tuberculosis X-ray Image Auxiliary Assessment Software	Jiangxi Zhongke Jiufeng Smart Medical Technology Co., Ltd.
国械注准20223211426	ECG Analysis Software	Xiamen Nalong Health Technology Co., Ltd.
国械注准20223211473	Peripheral Blood Cell Image White Blood Cell Auxiliary Recognition Software	Beijing Xiaoying Technology Co., Ltd.
国械注准20233210006	Breast X-ray Image Auxiliary Detection Software	Hangzhou Deep Insight Technology Co., Ltd.
国械注准20233210146	Coronary CT Blood Flow Reserve Calculation Software	Yukun (Beijing) Network Technology Co., Ltd.
国械注准20233210186	Coronary CT Angiography Image Vessel Narrowing Auxiliary Assessment Software	Yukun (Beijing) Network Technology Co., Ltd.
国械注准20233210216	Coronary CT Angiography Image Vessel Narrowing Auxiliary Assessment Software	Hangzhou Deep Insight Technology Co., Ltd.
国械注准20233210249	ECG Atrial Fibrillation Analysis Software	Shenzhen Libang Precision Instruments Co., Ltd.
国械注准20233210272	Cervical Cytology Digital Pathology Image Computer-Aided Analysis Software	Jiu Yi San Liu Medical Technology Nanjing Co., Ltd.
国械注准20233210405	Intracranial Hemorrhage CT Image Auxiliary Triage Software	Yukun (Beijing) Network Technology Co., Ltd.
国械注准20233210530	Lung Nodule CT Image Auxiliary Detection Software	Anhui Xunfei Medical Co., Ltd.
国械注准20233210568	Rib Fracture CT Image Auxiliary Detection Software	Hangzhou Deep Insight Technology Co., Ltd.
国械注准20233210695	Rib Fracture CT Image Auxiliary Detection Software	Yukun (Beijing) Network Technology Co., Ltd.
国械注准20233210707	Colonic Polyp Electronic Endoscope Image Auxiliary Detection Software	Tencent Medical Health (Shenzhen) Co., Ltd.

(Note: 1. Some domestic product structural composition information does not include “deep learning,” so it has not been included in the statistics, and related data may be underestimated. 2. Based on product registration information, it cannot timely reflect the impact of commercial mergers and acquisitions on the number of enterprises, registered products, etc.)

To gain a deeper understanding of the application status of AI medical device registrations in China, and to study the development status of the AI medical industry in a broader scope, we are conducting the “2023 Annual China Medical AI Practice Typical Case” collection work, with the following notification regarding the work.

Notice on the Collection of “2023 Annual China Medical AI Practice Typical Cases”National Health Commission Document No. 21To all relevant units:To comprehensively understand the new progress of AI technology applications in the medical field in China, showcase the new achievements in the development of China’s medical AI industry, and demonstrate the role of typical cases, our organization has decided to carry out the collection activity for the 2023 Annual China Medical AI Practice Typical Cases. The relevant matters are notified as follows:1. Organizing UnitsGuiding Unit: National Health Commission Hospital Management Research Institute; Organizing Unit: China Digital Medicine Magazine2. Scope of Collection(1) Typical applications of medical imaging AI;(2) Typical applications of medical voice AI;(3) Other typical applications (classified mainly by application scenarios, such as applications of AI in various specialized fields of hospitals, traditional Chinese medicine, chronic disease management, health management, etc.).3. Application Requirements(1) The main application units should be various levels of medical institutions, public health institutions, proactive health management institutions, etc., in collaboration with relevant enterprises for applications;(2) The submitted case content must be authentic and credible;(3) Please submit the relevant materials to the case submission email: [email protected] by July 15, 2023.4. Review and Results Application(1) After the case collection deadline, our organization will form an expert review committee to evaluate based on the principles of fairness, openness, and justice, selecting typical cases that are innovative, effective in application, and replicable and promotable, which will be made public;(2) The evaluation results will be published on the official websites and WeChat public accounts of the National Health Commission Hospital Management Research Institute and China Digital Medicine Magazine;(3) With the consent of the activities, the cases will be used for the “2023 China Medical AI Development Status” project research, and outstanding typical cases will be recommended for publication in the China Medical AI Development Report (2023) (Blue Book);(4) Organizing typical cases for presentation or report exchange at national academic conferences.5. Contact Persons and Contact InformationContact Persons: Jiang Shengliang, Song Jinjing, Zuo Han, Chen XiaoyunContact Number: 010-81138786, 81138783, 81138782, 81138710Mailing Address: Room 201, Building 3, No. 6, Shouti South Road, Haidian District, Beijing, China Digital Medicine MagazineChina Digital Medicine Magazine Co., Ltd.June 7, 2023Attachments:1. Application Form for Typical Cases of Medical AI2. Case Writing Requirements Research on Third-Class AI Medical Devices: Case Studies