In a significant milestone for clinical artificial intelligence in the Asia-Pacific region, Hong Kong-based medical technology firm DIAGENS has released benchmark results for its latest foundation model, detailing an unprecedented breadth of diagnostic capabilities.
On 12 November 2026, the publicly traded company announced the rollout of iMed-6.0, an iteration of its iMedImage architecture designed to handle highly complex, multimodal reasoning. According to the release, the updated system powers a multi-task Computed Tomography (CT) model capable of simultaneously analysing 509 distinct diseases and imaging findings across the chest and abdomen.
This wide-spectrum approach marks a critical departure from earlier generations of medical AI, which typically relied on narrow, single-disease algorithms to detect specific conditions like lung nodules or breast lesions. By extending a single model's analytical capacity to hundreds of potential targets, researchers are inching closer to the holistic reasoning processes employed by human specialists.
In recent evaluations using the global MedBench multimodal API framework, two variants of the DIAGENS system—dubbed Thoth and Aegis—secured the first and second positions respectively, validating the model's capacity to maintain accuracy despite a dramatically expanded diagnostic footprint.
For healthcare systems in Aotearoa New Zealand and Australia, the ongoing maturation of broad-spectrum medical AI presents a compelling practical opportunity. Both nations face entrenched shortages of specialist radiologists, leading to bottlenecks in scan reporting and delayed patient interventions. A foundation model capable of reliably triaging a vast array of abdominal and thoracic abnormalities could allow local health boards to redirect human expertise to the most complex or ambiguous cases.
Crucially, the iMed-6.0 ecosystem also includes "zero-code" development tools. This capability allows clinical researchers and speciality departments to build and refine custom AI diagnostic applications without requiring extensive software engineering resources—a feature that could empower regional hospitals to tailor the technology to local population health trends.
While extensive clinical validation across diverse patient demographics remains necessary before such systems can operate autonomously in local jurisdictions, the technical leap from narrow algorithms to multi-task, 509-target foundation models demonstrates that the underlying science of medical imaging AI is accelerating at a formidable pace.
Sources:
- PR Newswire: DIAGENS Leads Medical AI Benchmarks
- Accessnewswire: DIAGENS Leads in Two Major Global AI Benchmarks
AI disclosure: Automated systems assisted with research and drafting. Gemini independently peer-reviewed this article before publication. Sources are linked for verification.
