Preventing Disease Before It Strikes, Detecting Subtle Signs with Keen Eyes: Shanggong Yixin's Advancement in AI-Powered Retinal Image Analysis

“Superior physicians treat disease before it arises” originates from the Su Wen: Treatise on Regulating the Spirit in Accordance with the Four Seasons. Its fundamental meaning is to prioritize prevention as a therapeutic strategy. The concept of “superior physicians treating disease before it arises” encompasses multiple aspects, including preventing disease before its onset, preventing progression once disease has occurred, and halting further deterioration when changes have already taken place. This requires individuals not only to treat existing illnesses but also to prevent them; furthermore, beyond mere prevention, one must strive to arrest the trend toward pathological changes and prepare emergency interventions before such changes manifest. Only in this way can one seize the initiative in disease management and achieve the “perfect cure” characteristic of the “art of superior physicians.”

According to VCBeat (WeChat Official Account: vcbeat), Beijing ShangGong YiXin Technology Co., Ltd. has established disease prevention and screening as the cornerstone of its corporate development and service implementation since its inception. Throughout this process, the company has continuously integrated cutting-edge AI technologies into the early prevention and examination of patient diseases. Ultimately, it became the first intelligent medical technology company in China to successfully implement the combination of AI technology with fundus image analysis, and it is currently a leader in AI-powered fundus analysis services.

The key reason Shangong Yixin has endured and evolved into an AI healthcare technology company that truly achieves clinical application lies in its products’ genuine alignment with real-world medical needs. Its technologies are purpose-built for healthcare, and its products are developed specifically to address medical requirements.

Currently, Shanggong Yixin has applied AI technology for automated fundus image analysis to the clinical screening of diabetic retinopathy, glaucoma, and cataracts. It has built the AutoEye Intelligent Fundus Image Analysis Platform and launched a series of products, including Huiyan Diabetic Retinopathy, Huiyan Glaucoma, and Huiyan Cataract. These products are already in use at nearly 400 hospitals across China. The platform also provides online channels for automated fundus analysis and triage referral services for hospitals, doctors, and patients, ultimately integrating with the Big Data Fundus Analysis Platform to form Shanggong Yixin’s business ecosystem.

In the course of providing fundus AI services to major hospitals across China, Shangong Yixin’s AutoEye cloud platform has aggregated a vast volume of authentic, large-scale clinical fundus image data. As of July 2018, the platform had accumulated more than 750,000 fundus images, with the database growing at a rate of 60,000 images per month. After undergoing data cleaning, these datasets are utilized for the training and iterative optimization of Shangong Yixin’s AI products.

Since its inception, ShangGong YiXin has been dedicated to providing meticulous services for fundus screening across China, while continuously integrating the latest research advancements in the field of artificial intelligence—particularly machine learning represented by deep learning—into clinical screening services. The company has consistently launched deep learning products featuring independent intellectual property rights owned by ShangGong YiXin.

For example, in the first half of 2017, the company launched RetinalNet, a deep learning network designed for the automated staging and classification of diabetic retinopathy (DR), which was integrated into its Huiyan DR product. In actual clinical practice, this product achieved a diagnostic concordance rate of 95%. In the second half of 2017, the company introduced S-DSN, a short-connection-based deep learning network for the identification and segmentation of grade I, II, and III blood vessels. Its segmentation accuracy ranks among the leading levels reported in current international literature. This technology is slated for application in subsequent chronic disease screening products, such as Huiyan Stroke and Huiyan Hypertension. In the first half of 2018, the company unveiled SOSNet, a deep learning network targeted at the identification and segmentation of fundus lesions. To date, SOSNet is the world’s first deep learning network capable of simultaneously identifying and segmenting hard exudates, soft exudates, microaneurysms, and hemorrhages within a single analytical process. The network also innovatively employs multi-scale and multi-lesion feature fusion techniques to achieve the identification and segmentation of multiple lesion types.

Shanggong Yixin has not only made continuous strides in its own AI technology but also actively engaged globally, conducting technical exchanges and collaborations with leading international research institutions in related fields, such as the National University of Singapore and the Bioinformatics Institute of Singapore. Furthermore, it actively participates in top-tier academic exchanges and competitions in the medical field.

In March 2018, the Shangong Yixin team achieved outstanding results in the IDRiD Fundus Image Analysis Competition held at ISBI, the premier international conference on medical imaging, securing first place in optic disc segmentation, third place in optic disc detection, and fourth place in macula detection.

In July 2018, Shangong Yixin secured impressive results in the preliminary round of the REFUGE Glaucoma Fundus Image Analysis Challenge, held at MICCAI, another top-tier international conference on medical imaging. The company ranked second in optic disc segmentation, third in optic cup segmentation, and second in cup-to-disc ratio estimation. The Shangong team will compete in the finals scheduled for September 20 in Spain, where it will engage in in-depth exchanges with international research teams and industry peers. We look forward to achieving even better results in the final competition.

Additionally, on August 22, ShangGong YiXin announced the completion of its RMB 40 million Series A financing round. The round was led by Zhuhai Esen Technology, with co-investment from Haida Investment and GF Qianhe. Following this Series A financing, ShangGong YiXin will continue to allocate funds toward product refinement, promotion of market implementation, and the development of external collaborative platforms.

"Superior physicians prevent disease before it arises and detect subtle ailments with keen insight." Shangong Yixin will steadfastly uphold this service mission and philosophy, driving corporate growth by addressing clinical needs and leveraging real-world data, while propelling continuous leadership through AI technology as we forge ahead on the path of AI innovation.