Dysion Intelligence Files IPO Prospectus After Winning ISIC Skin AI Challenge

The first motivation stems from the exploration of business models. Judging by the current trajectory of medical AI development, it is difficult to rapidly achieve success by relying solely on algorithmic breakthroughs in AI products and seeking monetization through such means. For most artificial intelligence companies targeting radiology departments, cash flow remains a challenging issue to resolve. On one hand, the prolonged delay in obtaining Class III medical device approvals continuously erodes the patience of both hospitals and investors; on the other hand, radiology departments at Tier 3 Grade A hospitals may not be ideal payers.

Therefore, some AI companies are beginning to experiment with either continuing to reduce the costs of their existing services or exploring new revenue streams to seek fresh opportunities across various B2B and C2C segments.

Founded in 2017, Dingshi Intelligence has explored various business models. Focusing on the field of dermatology, Dingshi Intelligence leverages proprietary artificial intelligence algorithms to calculate skin lesion severity and the fractal dimension of surrounding skin tissue, thereby constructing structural maps of different growth patterns of the involved tissues. This has led to the launch of Wending Dermatology Smart Cloud.

The software is deployed within large tertiary Grade A hospitals via a private cloud, enabling physicians to search and query existing cases. It serves primary care hospitals through a public cloud to assist grassroots doctors in diagnosis, and provides auxiliary diagnostic services to third-party testing institutions. Furthermore, it guides patients on monitoring skin lesions and delivers risk assessments within 30 seconds based on image analysis; patients can receive diagnostic results through the Dingzhi Smart mini-program. This business model generates revenue by charging B-side clients for information system access and annual service fees, while also deriving income directly from C-end users.

Why Did Dingshi Intelligence Choose the Dermatology Field? What Are Its Unique Features in Core Technologies, Data, and Algorithms? To Address These Questions, VCBeat (WeChat ID: vcbeat) Conducted an Exclusive Interview with Ma Lei, Co-Founder and COO of Dingshi Intelligence, and Dr. Zhuang Yixin, CTO.

Ma Lei told reporters that before transitioning to AI-driven healthcare, his team had spent approximately three years developing facial recognition systems, accumulating substantial expertise in computer vision and image recognition. Coupled with a technical team composed of overseas returnees, the shift toward dermatological diagnosis was a natural progression.

According to the VCBeat article “A Review of 11 Global AI Projects in Dermatology: 63% Target Clinicians, Chinese Companies Predominate, All Partnering with Top-Tier Hospitals,” dermatology is a discipline that relies heavily on morphological features, and skin imaging is a crucial tool for diagnosing skin diseases. The diagnosis of skin conditions through imaging has evolved from initial visual inspection to magnifying glass- and microscope-assisted diagnosis, and more recently to digital imaging technologies and intelligent analysis.

Currently, the integration of AI with dermatological imaging is advancing rapidly, with numerous academic papers suggesting that AI technology can surpass dermatologists. The foundation of AI lies in big data, specifically dermatological imaging data, which includes clinical skin photographs, dermoscopy, reflectance confocal microscopy (skin CT), and dermatopathology. Standardizing and annotating this big data transforms it into training material for AI machine learning, a critical step essential for the future development of AI.

Stanford University published an article in Nature exploring the use of artificial intelligence for the automated diagnosis of skin diseases, leveraging a database of 130,000 dermatological images. The image database comprised dermoscopic images, smartphone photographs, and standardized clinical photos. The study evaluated the AI diagnostic system’s ability to differentiate between benign skin tumors, malignant tumors, and other non-neoplastic skin conditions. The results demonstrated a very high concordance between the AI system’s diagnoses and those made by dermatology experts, with diagnostic performance reaching parity.

However, the wide variety of dermatological conditions, coupled with the lack of unified criteria for differential and definitive diagnosis, makes it challenging to train robots to recognize and diagnose diseases. This represents one of the bottlenecks in AI-based dermatological diagnosis. Currently, automatic recognition and diagnosis based on dermatological imaging still fall short of achieving the accuracy of histopathological images. Furthermore, rare skin diseases have very few documented cases, resulting in an insufficient sample size to meet the data requirements for machine learning training. Consequently, achieving optimal efficiency in automated recognition and diagnosis remains difficult.

Human skin serves as a mirror, reflecting an individual's vitality and health status. People with different health conditions exhibit variations in skin tone, while distinct racial groups show pronounced differences in pigmentation. These variations pose significant challenges for AI image recognition; therefore, AI companies must accumulate images of patients with diverse skin tones to effectively serve corresponding demographic groups.

As one of the four major races in the world, East Asians rank second in population. For companies starting with the domestic market, enriching effective medical case images of East Asians is the core of their technological development.

Dingshi Intelligence is rooted in the domestic market, collaborating with the dermatology departments of dozens of leading hospitals in China. By annotating tens of thousands of learning cases, it has established a high-quality, standardized dermoscopic image database specifically tailored for individuals of Asian descent.

“Dingshi Intelligence leverages its proprietary ‘deep learning + transfer learning + reinforcement learning’ technology, integrating top-tier expert experience to continuously push the boundaries of algorithmic precision,” introduced Zhuang Yixin. In clinical trials at authoritative hospitals, it surpassed expert-level performance. The accuracy of AI-assisted diagnosis in dermatological imaging increased from 71.6% to 91%.

In August 2019, Dingshi Zhihui secured first place in the public dataset category and second place in the non-public dataset category at the International Skin Imaging Collaboration (ISIC) AI Algorithm Challenge. “This marks the first time a Chinese company or institution has won the championship in this competition,” said Zhuang Yixin. In the same competition, Tencent AI Lab ranked 12th in the public dataset category, while Lenovo AI Lab ranked 25th in the public dataset category.

In terms of specific product implementation, Dingshi Wisdom has developed the “Wending Skin AI Cloud” product suite, which includes the Wending Skin Multimodal Imaging Management Cloud, the Wending Zhifu Doctor App, the Wending Zhifu Patient Mini Program, and the Wending Skin Smart Box.

WenDing Dermatology Multimodal Imaging Management Cloud: WenDing Dermatology Multimodal Imaging Management Cloud is a cloud-based platform designed to assist dermatologists in the centralized entry, diagnosis, management, retrieval, and disease tracking of all types of dermatological images, organized around the patient. Its powerful AI-assisted diagnostic function achieves an accuracy rate exceeding 90%.

Wending Zhifu Doctor App: Enables real-time synchronization with Wending Skin Cloud to perform functions such as image acquisition, diagnostic entry, record-keeping, and management. Additionally, its robust social features facilitate remote consultations within dermatology medical consortia, as well as patient follow-up and chronic disease management.

WenDing ZhiFu Patient Mini Program: Built on authentic doctor-patient relationships, it integrates online and offline services. Patients can view their diagnostic reports and initiate new consultations to communicate with doctors through the mini program, truly enabling remote follow-up visits and chronic disease management.

WenDing Skin Smart Box: A compact, plug-and-play hardware-software integrated device that connects data acquisition equipment to the cloud platform.

In terms of practical implementation, the AI-powered intelligent dermatological diagnostics within Wending Skin Smart Cloud enable patients to perform self-screening for skin tumors via AI. The system can also serve as an AI research platform for dermatology and facilitate AI-based skin screening for health examination centers and insurance companies.

Wending Skin Wisdom Cloud’s multimodal dermatological image management system facilitates the development of dermatology imaging centers. Serving as a universal operating system for dermatological imaging devices and a comprehensive informatization solution for dermatology medical consortia, it further enhances doctor-patient communication.

It is reported that Dingshi Intelligence has established strategic partnerships with multiple medical device manufacturers. Ma Lei introduced that these manufacturers have adopted the Wending Skin System as the standard operating system for the dermatology industry, leveraging their existing distributor networks to rapidly promote its adoption. Additionally, through a strategic partnership with Inspur Group’s Medical Big Data Company, Dingshi Intelligence provides software and technical support to Inspur in areas such as dermatological data applications.

To date, the Wending Dermatology AI Cloud has been deployed in multiple leading hospitals. Its PACS system integrates AI-assisted diagnosis to support clinical practice and scientific research, facilitates remote consultations within dermatology medical consortia, bridges online and offline services, and enables patient follow-up and chronic disease management.

In terms of future planning, Dingshi Zhihui will continuously enhance its SaaS services, achieve breakthroughs in AI technology, realize comprehensive coverage of AI-assisted diagnosis for skin tumors, and improve algorithm accuracy. In the market sector, it aims to capture a 50% share of dermatology departments across China. Additionally, the company will develop intelligent dermoscopy hardware and expand into overseas markets serving Asian populations.

It is reported that,Dingshi Smart is currently raising its pre-A round of financing., expanding from dermoscopy to pathology, rapidly advancing market penetration, with a plan to reach 200 client hospitals by next year.Interested institutions, please contact VCBeat's financing assistant, Xiao Yun., WeChat:DongMai_Investent