AI“看懂”影像、“理解”病历：“京医千询”大模型推动医疗普惠

Core Insights - The "Jingyi Qianxun" medical AI model by JD Health showcases significant advancements in the healthcare AI sector, particularly in its core capabilities and ecosystem development [1][2]. Group 1: Core Capabilities - "Jingyi Qianxun" is built on three core capabilities: multimodal perception, human-like dialogue, and reliable reasoning, which form the technical foundation for the model [2]. - The upgraded "Jingyi Qianxun 2.0" integrates information from various modalities, including text and imaging, enabling comprehensive health record construction and advanced pixel-level image analysis [2]. - The model's reasoning ability simulates human diagnostic logic, allowing it to analyze complex medical data and provide reliable second opinions, even from distorted medical reports [2]. Group 2: Dialogue and Interaction - The human-like dialogue capability allows "Jingyi Qianxun" to understand vague patient descriptions and provide tailored responses based on individual patient characteristics [3]. - A patient simulator has been developed to enhance dialogue realism, enabling the model to train on diverse user expressions and disease severities, thereby improving interaction quality [3]. Group 3: Performance Metrics - "Jingyi Qianxun" has demonstrated outstanding performance in authoritative evaluations, achieving 99% in medical calculation, 91% in medical Q&A, and 91% in text summarization, positioning it as a leader in the industry [3][4]. - The model has outperformed competitors in 21 evaluation metrics across five major capabilities, confirming its superior medical AI capabilities [4]. Group 4: Specialized Development - JD Health has focused on specialized models by leveraging high-quality online diagnosis data from its internet hospital, ensuring alignment with clinical tasks [5]. - The evolution from a general model to specialized models is supported by collaborations with top medical institutions and the integration of diverse clinical data for training [5]. Group 5: Future Prospects - The ongoing development of "Jingyi Qianxun" aims to bridge the gap between AI capabilities and human clinical cognition, potentially providing high-quality medical services to users in remote areas [6].