On April 24th, coinciding with the official release and open-sourcing of DeepSeek's next-generation flagship model DeepSeek-V4 preview, YIDU TECH (02158) announced that its clinical evidence-based intelligent agent "Yidu Smart Pathway" has completed medical scenario adaptation with a dual-base architecture incorporating both DeepSeek-V4 and Yidu's proprietary large model. Three key upgrades featuring a million-token context window, enhanced reasoning capabilities, and improved complex problem-solving have been simultaneously implemented across Yidu Smart Pathway's three core scenarios: clinical Q&A, medical record analysis, and evidence-based retrieval.
In practical use, the system intelligently allocates different model capabilities based on question complexity, with dedicated solutions for both simple queries and complex tasks, including multi-model collaboration when necessary. Previously, a patient's complete medical record often contained dozens of pages including admission notes, daily progress reports, test results, and imaging descriptions. Traditional AI systems constrained by limited context windows couldn't process entire records at once, forcing doctors to input information segmentally and potentially miss critical clues.
DeepSeek-V4's one-million-token context window, equivalent to approximately 400,000 Chinese characters, can now accommodate complete hospitalization records or even entire clinical guidelines. Doctors can import full medical records at once and ask comprehensive questions like "What are the potential causes of septic shock in this patient?" with the AI analyzing the entire clinical course from admission to present.
For evidence-based medicine, traditional AI retrieval typically returned fragmented literature excerpts, requiring manual synthesis across documents by physicians. Leveraging the extended context and advanced reasoning capabilities, Yidu Smart Pathway can now process multiple clinical guidelines, randomized controlled trials, and case reports simultaneously, automatically integrating evidence from disparate sources to generate complete evidence pathways with properly cited conclusions - essentially performing preliminary literature review for clinicians.
This upgrade brings three substantive improvements: medical record analysis transitions from segmented input and fragmented responses to complete input with holistic answers; evidence retrieval evolves from finding literature fragments for manual synthesis to cross-document integration with direct evidence pathways; complex clinical questions shift from superficial responses with long waits to more comprehensive reasoning with intelligent responses. Yidu Smart Pathway is transforming from a data retrieval tool to a clinical workflow assistant that supports clinical thinking.
The technological advancement coincides with promising规模化 deployment prospects, closely tied to YIDU TECH's established hospital network. Public information shows that as of September 30, 2025, the company has established deep collaborations with 127 top-tier Chinese hospitals and 44 regulatory agencies, with its hospital network covering over 10,000 medical institutions powered by the core "AI Medical Brain" YiduCore algorithm engine.
Solutions built on the Yidu Smart Pathway evidence engine have been deployed in more than 40 tertiary hospitals, participating in over 500,000 clinical decisions. With accumulated expertise covering thousands of diseases and knowledge graphs in specialized areas including oncology, hematology, and cardiovascular diseases, YIDU TECH has developed a complete product matrix ranging from "AI Middle Platform" to "Doctor Copilot" and "Yidu Smart Pathway."
The company's collaboration with leading medical institutions further strengthens its competitive moat and commercial potential. For instance, Xiangya Hospital of Central South University, as the first medical institution deeply integrated with DeepSeek's AI Middle Platform, has successfully completed localized deployment of YIDU TECH's domestically developed AI medical platform, creating AI assistants covering the entire treatment process. At Sun Yat-sen University Cancer Center, YIDU TECH achieved private deployment of its AI middle platform, combining YiduCore and DeepSeek dual engines with hospital-wide big data platforms to launch AI diagnostic assistants covering full treatment scenarios.
Analysis indicates that with deep coverage across over 10,000 hospitals and demonstration effects from top-tier hospital projects, YIDU TECH's integration of DeepSeek-V4 into Yidu Smart Pathway's dual-model architecture not only achieves generational improvement in core clinical evidence-based experience but also opens significant commercial potential. The business flywheel comprising top technology, premium hospital partnerships, and regional project capabilities is accelerating. The upgrade is now fully available for physicians through the Yidu Smart Pathway app or WeChat mini-program.