Chinese medical AI model wins top state science award as sectoral "ChatGPT moment" arrives

Results have recently been released for the final of the National Healthcare Security AI Image Recognition Contest. The contest, organised by China's National Healthcare Security Administration, was built around real-world clinical imaging requirements. A total of 1,332 teams across China participated. Drawing on Hangzhou Diagens Biotechnology Co., Ltd. (DIAGENS)’s proprietary iMedImage®, the world’s first cross-modal foundation model, and its iMedLoop™ medical imaging AI R&D and production acceleration platform, a team from the Second Affiliated Hospital of Guangxi University of Chinese Medicine developed a CT diagnostic model for kidney cancer to claim first prize in the contest, while a breast cancer X-ray diagnostic model built by the First Affiliated Hospital of Guangxi Medical University secured second prize.

These remarkable results further validate DIAGENS' technological leadership in the global medical AI sector and the broad applicability of its solutions.

In China, clinicians have taken center stage in the field of medical AI, with a domestic foundational technology stack enabling cross-modal applications to claim top awards. DIAGENS translated sophisticated R&D capabilities into foundations and tooling accessible to clinical teams, enabling clinicians to evolve from AI end-users into AI creators. The "ChatGPT moment" for medical imaging has arrived.

Foundation model: resetting the starting point for AI-driven medical innovation

Clinical specialists understand disease pathology best and grasp most clearly the pain points to resolve in real-world clinical workflows. For years, technical barriers spanning data processing, algorithm development, model training, evaluation and deployment have prevented vast pools of clinical expertise from translating directly into AI systems.

DIAGENS has rewritten this R&D pathway. Founded in 2016, the Hong Kong-listed artificial intelligence firm is recognised as the world's first publicly listed company focused on medical imaging foundation models. It specialises in artificial intelligence and biomedicine, and aims to build a global industrialization platform for medical imaging AI. Centered on its self-developed iMedImage® medical imaging foundation model, the company provides a shared development backbone for specialised disease models. With 104 billion parameters, the model was pre-trained on 80 million clinical imaging studies. It delivers multi-modal compatibility across CT, X-ray, MRI, and ultrasound. Model support was available across 26 clinical specialties and over 4,000 disease scenarios. New clinical tasks can be adapted atop this existing foundation, reducing redundant effort associated with development from scratch.

On this basis, DIAGENS links data generation, model training, unified evaluation, deployment and application feedback via its iMedLoop™ medical imaging AI R&D and production acceleration platform, forming an integrated closed-loop R&D and production pipeline. Powered by iMedImage® and iMedLoop™, clinicians define tasks, set benchmarks and participate in development around real clinical challenges. The mature foundation and platform provide the capacity to tackle intricate technical and engineering challenges.

This framework packages DIAGENS' long-accumulated technical expertise into an accessible platform, returns control of innovation to clinical teams, and supports disease specialists in leading innovation in medical AI.

National contest validation: proving the industrial rollout pathway for medical AI

The strong performance of the kidney cancer CT-based and mammography X-ray detection projects led by frontline clinicians at this contest stands as real-world proof of DIAGENS' multimodal foundation capabilities and industrial innovation framework, demonstrated on a national professional stage.

Beyond the contest, this underlying technology stack has already undergone scaled rollout across clinical practice. Built on the iMedImage® medical imaging foundation model, DIAGENS has extended the capability frontier of medical imaging AI across diverse imaging modalities. CT, X-ray, MRI and ultrasound, alongside karyotyping and cytomorphology studies, all draw on the same foundational backbone and R&D-to-production pipeline. To date, DIAGENS has delivered a cumulative total of 158 model projects in partnership with 99 hospitals — 65 of which are Grade III Class A hospitals — spanning 43 anatomical sites and 61 disease categories.

This unified underlying capability spans multiple imaging modalities to support an ongoing stream of distinct diseases and clinical tasks. The industrialized framework enables batch development of high-performance specialist AI tools. DIAGENS continues to roll out its leading capabilities across further niche medical imaging verticals.

China's new approach: ushering in the "ChatGPT moment" for global medical AI

ChatGPT redefined how people interact with AI, unlocking access to powerful AI capabilities for broader user groups. DIAGENS is reshaping how medical AI is developed. Its foundation model and R&D platform cut development hurdles for specialty-specific models, enabling more clinical specialists to shift from AI end-users to AI creators.

When clinicians can build AI models without writing code themselves, the scope for clinical innovation broadens substantially. Clinicians across diverse specialties can pursue research around their own clinical challenges, test hypotheses, translate specialist experience into a rich ecosystem of AI models, and distil clinical wisdom into reusable, deployable medical AI capabilities. This technical pathway rooted in Chinese practice widens the community of medical AI builders and expands the pool of beneficiaries of these innovations.

Powering multi-modal innovation, this industrial framework delivers on the "one foundation for thousands of models" paradigm — marking the arrival of the "ChatGPT moment" in medical imaging. DIAGENS leverages its homegrown proprietary technology to lead a global shift in the production of medical imaging AI.