On May 22, DIAGENS-B (02526) announced that its self-developed AI AutoVision® Chromosome Karyotype Assisted Diagnostic Software has officially obtained the Class III Medical Device Registration Certificate from the National Medical Products Administration (NMPA). This represents the fastest-approved medical imaging AI product globally in fields such as birth defect prevention and assisted reproduction, having passed China's highest level of medical device registration verification. The Class III certificate is China's most stringent medical device market access level, with the strictest approval standards and highest clinical validation requirements, often referred to as the "highest pass" for AI medical devices.
AI AutoVision® had previously entered the Innovative Medical Device Special Review Process (green channel), recognized as an innovative product by the NMPA. Its formal approval signifies that the product has completed the full validation process from "innovation recognition" to "compliant market launch" at the regulatory level. For investors, the value of this certificate lies in its triple rarity: global first-of-its-kind, highest-level market access, and foundational large model passing regulatory validation for the first time. The combination of these three factors is unprecedented among Hong Kong-listed companies.
"Global First-of-its-Kind": An Unpriced Market Entry Point The chromosome karyotype analysis field where AI AutoVision® operates is a fundamental testing link in birth defect prevention and assisted reproduction, directly related to prenatal diagnosis and infertility treatment. However, this area has long faced supply bottlenecks. Under traditional methods, doctors must identify, count, and arrange 46 chromosomes under a microscope, taking an average of about 34 minutes per case, with a report cycle extending to approximately 30 days. Training a skilled karyotype analyst requires 3 to 5 years, leading to a severe shortage of professional talent.
Leveraging DIAGENS-B's iMedImage® medical imaging foundational large model and proprietary AI algorithms, AI AutoVision® enables one-click chromosome segmentation, counting, arrangement, and case-level anomaly detection, upgrading the analysis process from "manual interpretation" to "AI automated analysis + doctor review." A multicenter prospective clinical trial involving 1,734 cases demonstrated: 100% sensitivity and specificity for numerical anomaly detection; 94.31% sensitivity and 100% specificity for structural anomaly detection; average analysis time reduced from 34.42 minutes to 11.19 minutes; report cycle shortened from about 30 days to 4-7 days; and automated analysis accuracy exceeding 99%.
This represents the largest multicenter clinical trial for chromosome karyotype analysis to date globally, with its extensive coverage and large sample size being rare in approvals for similar AI medical devices. The dual 100% results in sensitivity and specificity indicate that, for the core metric of numerical anomaly detection, AI has reached or even surpassed the level of experienced analysts. This is not a gradual improvement but a fundamental replacement of the supply model.
"Highest Level Access": Regulatory Barriers as Competitive Barriers The difference between Class III and Class II certificates is not merely a matter of approval hierarchy but a significant gap in market entry thresholds. Class III certification requires products to undergo rigorous multicenter clinical trials to prove safety and efficacy, with approval cycles typically measured in years, and imposes high demands on algorithm interpretability, data compliance, and network security. In contrast, Class II certificates have relatively lenient clinical validation requirements and shorter approval cycles.
This means that products with Class III certification possess a first-mover advantage in market access, hospital procurement, and medical insurance inclusion that Class II products cannot match. The approval of AI AutoVision® directly opens channels for large-scale hospital adoption and lays the groundwork for subsequent inclusion in medical insurance payment systems. For competitors, completing the full process from "R&D → clinical trials → Class III certification approval" from scratch would take at least 2 to 3 years, creating a tangible time barrier for DIAGENS-B in the chromosome karyotype analysis field.
"Foundational Model Validation": A Certificate's Value Extends Beyond a Single Product The core significance of AI AutoVision®'s Class III approval lies not only in the capabilities of a single product but also in its underlying iMedImage® medical imaging foundational large model—one of the world's most broadly covering medical imaging foundational models, with 104 billion parameters, supporting 19 medical imaging modalities, and covering over 90% of clinical imaging detection scenarios.
The value logic of a foundational model is: validate once, replicate everywhere. The approval of AI AutoVision® demonstrates that products derived from the iMedImage® technological pathway can pass the highest level of regulatory scrutiny. The foundational model is no longer just a "trainable" research tool but a "regulatorily approvable" commercial infrastructure.
Over the past six months, DIAGENS-B has collaborated with 65 top-tier hospitals to incubate 92 advanced imaging-specific models, covering 32 human organs and 64 disease areas. With the first Class III certificate secured, the approval pathway for subsequent products has been established, with approval cycles expected to shorten further. For over 3,200 medical imaging testing items in China and approximately 5,000 globally, this signifies that DIAGENS-B has evolved from "making products" to "building a platform"—a fundamental shift in valuation logic.
Investment Logic: From "Selling Equipment" to "Selling Capabilities" DIAGENS-B's 2025 annual report data has already shown early signs: annual revenue reached approximately RMB 164 million, a year-on-year increase of about 134%; technology licensing revenue surged to RMB 84.3 million, a dramatic 331.7% increase year-on-year, becoming the largest revenue source for the first time, accounting for over 50% of total revenue.
This structural change indicates that DIAGENS-B's growth engine has upgraded from single-point delivery of "equipment + software" to a composite, high-margin model of "equipment entry + model platform + technology licensing." Technology licensing revenue exceeding 50% essentially represents the commercial realization of the foundational large model's "industrialized mass production" capability.
Currently, DIAGENS-B's products cover over 400 medical institutions nationwide, with its market share in chromosome karyotype analysis firmly ranking first in China. Following the Class III approval of AI AutoVision®, the opening of large-scale hospital adoption channels is expected to further accelerate revenue growth driven by core products.
According to Frost & Sullivan forecasts, the global AI medical imaging market is expected to grow from approximately $1.6 billion in 2024 to about $9.3 billion by 2030; China's core AI medical imaging market is also projected to surge from around RMB 2.4 billion to about RMB 40.1 billion. As one of the first companies globally to successfully navigate the entire chain from "foundational model → Class III certified product → technology licensing revenue," DIAGENS-B stands at the starting point of this explosive growth.
Comments