Star Metro - China Explores a New Pathway for Medical AI: Bridging R&D Innovation and Clinical Adoption

HANGZHOU, CN / ACCESS Newswire / November 10, 2026 / For enterprises advancing medical AI commercialisation, hospitals are the ultimate testing ground for real-world validation. Algorithms must target specific clinical pain points, adapt to local imaging datasets, and integrate seamlessly with frontline operational workflows. However, replicating customised development and adaptation work across individual healthcare institutions hinders scalable expansion and drives up costs.

Now, a newly launched partnership in southern China presents a model for restructuring medical AI application workflows. On 6 November, Hangzhou Diagens Biotechnology Co., Ltd. (DIAGENS, HKEX:02526.HK), a Hong Kong-listed medical imaging AI player, entered into a three-year strategic cooperation agreement with Yongjiang Laboratory in Guangxi. The cooperation covers model training, multi-centre clinical validation and the implementation of demonstration projects, with cross-border cooperation in Southeast Asia as a long-term objective.

This partnership aligns with China's broader strategy of integrating AI innovation with clinical practice. Public healthcare institutions, clinical teams and technology enterprises are jointly building a process to accelerate the integration of specialty AI models into mainstream healthcare workflows.

Translating Clinical Requirements into AI-enabled Solutions

Diagens anchors its technological system on iMedImage^®, its reusable foundational medical imaging model. Its proprietary iMedLoop^™ medical imaging AI R&D and production acceleration platform integrates key stages including data processing, intelligent annotation, quality control, model training, unified evaluation, release, deployment and application feedback. Clinical teams define clinical requirements and task specifications based on their professional expertise, while Diagens offers a technical foundation and tailored tools to develop task-specific AI models.

This paradigm delivers distinct advantages. Teams developing CT diagnostic models for renal diseases and X-ray models for breast conditions operate in different scenarios, yet they can share underlying R&D infrastructure. The reuse of foundational capabilities optimises cost allocation across multiple scenarios and enables hospitals to launch projects without assembling large technical teams.

Guangxi has already witnessed practical achievements. Leveraging the core technologies of Diagens, local hospital teams have developed CT diagnostic models for renal cancers and X-ray diagnostic models for breast cancers, securing awards in a national medical imaging AI competition held by China's National Healthcare Security Administration. These successful cases demonstrate how clinicians participate in scenario-oriented AI model development. The new laboratory cooperation in Guangxi aims to scale up such collaborative models.

The business model can scale along two dimensions. Individual healthcare institutions can develop multiple AI-enabled diagnostic tools, while mature solutions can be iterated and adapted for deployment across diverse healthcare facilities. The competitive advantage derives from the transfer of technical experience and clinical insights across projects.

Transformative Impacts of Industry Policy Support

China's policies have paved a clear pathway for the clinical implementation of this collaborative model. A national healthcare plan released in November 2025 sets clear development targets: specialty models and AI agents are to be fully established by 2027, and AI-assisted medical imaging tools shall achieve universal adoption in secondary-level and higher hospitals nationwide by 2030.

These targets deliver tangible value by integrating all essential resources, including participating hospitals, clinical specialists, professional evaluation systems and technical capabilities. The systematic coordination of these core elements constitutes a critical step in accelerating the implementation of medical AI technologies.

Diagens' cooperation framework agreement with Zhejiang Anzhen'er Medical Artificial Intelligence Technology Co., Ltd. ("Anzhen'er"), signed in September 2026, serves as a typical case. As the operating entity of the National Artificial Intelligence Application Pilot Base (Medical) in Zhejiang, Anzhen'er plans to explore an integrated service model with Diagens, combining on-site medical examinations, cloud-based AI analysis and specialist manual review. This framework standardises the role of AI technologies in clinical workflows and enables grassroots healthcare institutions to access high-quality shared resources.

Guangxi presents a different healthcare ecosystem. By the end of 2025, this province hosted 940 hospitals and 1,258 township healthcare centres. Clearly, cross-institutional collaboration requires adaptive optimisation for differentiated equipment configurations, staffing structures and computing resources. Successful implementation in Guangxi will generate replicable experience for other regions grappling with similar resource gaps between large hospitals and grassroots healthcare facilities.

Policy incentives run parallel to rigorous product supervision. In May 2026, Diagens obtained Class III medical device registration for its AI AutoVision^™ chromosome analysis system in China. This milestone validates Diagens' capability to translate imaging AI technologies into fully compliant, commercially available clinical products via regulatory approval pathways.

Technical Capabilities Underpinning Clinical Deployment

As disclosed in Diagens' 2025 annual report, its iMedImage^® foundational model features 104 billion parameters, covering 19 imaging modalities and 26 clinical specialties. The extensive coverage aligns with the strategic goal of supporting diversified specialty-specific scenarios based on a unified foundational model. The model enables multi-level image analysis tasks, including classification, detection and segmentation.

Relevant technical reports demonstrate robust model performance. It achieves a chromosome classification accuracy rate of 99.96% based on 45,328 images. In end-to-end chromosomal abnormality analysis tasks leveraging data from 12 healthcare institutions, it delivers a sensitivity of 92.75% and a specificity of 91.50%. These task-specific results provide solid evidence for Diagens' technical strengths, as its registered products mark substantial progress in clinical commercialisation.

Medical imaging represents a core frontier for the global development of foundational AI models. Google's MedGemma 1.5 expands its capabilities to 3D CT, 3D MRI and whole-slide histopathology imaging scenarios, while Microsoft's MedImageParse and MedImageParse3D enable segmentation, detection and recognition for 3D medical image analysis. These technological advances collectively reflect a clear industry trend: reusable foundational models are becoming the core infrastructure for diversified imaging tasks and specialty-specific applications.

The breadth of imaging modality coverage serves as a key indicator of competitiveness. Google's MedGemma 1.5 covers six major imaging categories, including chest X-ray, CT, MRI, dermatological imaging, fundus imaging and histopathology. Microsoft's MedImageParse supports nine imaging modalities, whereas iMedImage® from Diagens covers 19 modalities. Compared with mainstream foundational models, iMedImage® boasts industry-leading coverage of medical imaging scenarios, laying a solid foundation for the development of specialty AI models for diverse clinical tasks.

By integrating its foundational imaging capabilities, hospital-led customised model development, registered clinical products and revenue-generating model services, Diagens has established a leading position in the commercialisation of medical AI. Its core competitiveness lies in the closed-loop R&D system it enables: clinical expertise drives the iteration of specialty AI models, which are integrated into clinical workflows after rigorous evaluation. This unique enabling role differentiates Diagens in the global drive for standardised, scalable development and deployment of medical AI technologies.

Reuse-centric Commercialisation Model

This paradigm is backed by robust commercial performance. In the first half of 2026, Diagens recorded RMB 94.5 million in Model-as-a-Service (MaaS) revenue, representing a year-on-year surge of 101.1%. The revenue stream covers technology licensing, cloud-based services and on-premises deployment. As of 30 June, Diagens had launched 158 customised model projects in collaboration with 99 partner hospitals, building a solid client and R&D foundation for sustained business expansion.

The business model's profitability hinges on reuse and iterative optimisation. Each project accumulates actionable experience in data processing, task definition, model evaluation and clinical deployment, which streamlines the delivery of subsequent projects, enabling the platform to support more scenarios without repetitive foundational R&D investment. Authorised clinical data and practitioner feedback further drive continuous model iteration and technological development.

As a Chinese province bordering Vietnam and with links to Southeast Asia, Guangxi unlocks new cross-border cooperation potential for Diagens. The experience accumulated across China's tiered healthcare system will serve as a valuable reference for regional market expansion in Southeast Asia, where local clinical partnerships and scenario adaptation remain critical.

As a real-world example of China's medical AI strategy, Diagens is committed to translating its reusable foundational AI model into an expanding portfolio of clinically validated tools. Its core growth opportunity lies in building a standardised and repeatable system, which significantly lowers the barriers to medical AI R&D and large-scale clinical implementation.

Contact
DUAN YU
DIAGENS Public Relations
[email protected]

SOURCE: Hangzhou Diagens Biotechnology

View the original press release on ACCESS Newswire

A.Lavoie--SMC