Medical AI Ecosystem Innovation Forum Highlights Data-Driven Transformation in Healthcare and Drug Discovery

On July 4, the Medical AI Ecosystem Innovation Forum and iMedLoop Global Medical Imaging Data Platform Launch was held in Beijing, jointly organized by Liaowang Finance and Diagens Technology (搜索)搜索该词条. Guided by the theme of "AI for Science," the event brought together more than 100 representatives spanning government, industry, academia, research, and healthcare, including experts from the Chinese Academy of Sciences, the Chinese Academy of Engineering, the China National Health Association, and leading hospitals such as Zhejiang Cancer Hospital and Ruijin Hospital Affiliated to Shanghai Jiao Tong University School of Medicine.

During the event, Diagens Technology (搜索)搜索该词条 officially launched the iMedLoop Global Medical Imaging Data Platform, a proprietary platform developed specifically for the medical AI industry. More than 30 strategic cooperation agreements were also signed, establishing a practical foundation for cross-sector collaboration to advance medical AI development.

The Critical Role of Data Quality in Medical AI

Professor Chen Runsheng, bioinformatician and researcher at the Institute of Biophysics, Chinese Academy of Sciences, delivered a keynote on the technical principles and future challenges of large AI models. He noted that AI has become deeply integrated into medical imaging and is increasingly serving as an essential analytical tool. "AI can integrate the knowledge and expertise of medical imaging specialists, bringing together multiple analytical approaches to deliver high-quality imaging analysis capabilities," Chen stated, emphasizing that AI's greatest strength lies in consolidating the knowledge and experience of multiple experts to overcome the limitations of individual interpretation.

Academician Cai Xiujun, President of Sir Run Run Shaw Hospital, reinforced this perspective through practical examples including remote robotic surgery, remote ultrasound diagnosis, and AI-assisted medical imaging diagnosis. He identified data quality, data scale, and data security as the three critical factors determining the success of AI applications in healthcare. "Poorly standardized or low-quality data directly reduces AI performance and ultimately limits its clinical value and broader adoption," Cai warned, calling on the industry to prioritize standardized medical data governance.

Academician Dong Jiahong of the Chinese Academy of Engineering and President of Beijing Tsinghua Changgung Hospital stated that the engineering foundations for AI hospitals are now in place, driven by the simultaneous maturation of three pillars: the commercialization of AI-powered medical devices, the advancement of large medical AI models to near-specialist levels of clinical reasoning, and the engineering development of AI agents. Unlike smart hospitals or internet hospitals, AI hospitals are "built upon digital twins and powered by AI-native operational logic," fundamentally reshaping healthcare workflows from perception and cognition to decision-making and service delivery.

iMedLoop: Building a Trusted Industrial Foundation

Dr. Song Ning, Chairman and CEO of Diagens Technology (搜索)搜索该词条, unveiled the iMedLoop platform, highlighting the scale of the challenge facing the industry. He noted that there are more than 3,000 medical imaging indications worldwide, while traditional AI model training typically requires hundreds of thousands of annotated images, with each annotation taking approximately one hour. "Even if hundreds of thousands of imaging, pathology, and laboratory professionals across China devoted one hour per day to annotation work, completing the annotations for all projects would still take more than a thousand years," Dr. Song explained.

To address this bottleneck, Diagens launched iMedImage, described as the world's largest medical imaging foundation model by parameter scale, in May 2025. According to Dr. Song, the foundation model reduces the amount of annotated data required for disease-specific model training to one two-hundredth of traditional levels, shortens development cycles to one-twelfth, and reduces both development costs and computing expenses to one-tenth. Over the past 12 months, Diagens has participated in six national and provincial-level major projects and collaborated with 87 leading hospitals to train 145 vertical AI models.

The iMedLoop platform integrates the iMedImage foundation model, the iMedStudio intelligent annotation tool, and the iMedMaaS online model training and deployment platform. The platform now hosts more than 3,000 professional annotators, 28.95 million high-quality data records, and over 100 deployed medical AI models.

AI in Oncology and Drug Discovery

Academician Zhan Qimin of the Chinese Academy of Engineering and Director of the National Institute of Health Data Science at Peking University highlighted AI's transformative potential in oncology. "In the past, treatments were often broad and one-size-fits-all, without sufficient consideration for individual differences and precision. Such approaches could lead to significant side effects and limited efficacy," Zhan said. "Today, by combining multi-omics data with AI analysis and applying the insights to pathology slides, it is becoming possible to provide each cancer patient with a truly tailored treatment plan." He also underscored AI's potential in drug discovery, including shorter development cycles, lower costs, and higher success rates.

In a related forum held on July 22 during the 2026 World Internet Conference Digital Silk Road Development Forum in Xi'an, Yang Yang, director of group office and head of group AI strategy at Hansoh Pharma (搜索)搜索该公司, examined the current landscape of AI-enabled innovative drug research. Yang noted that innovative drug research has become a frontier for AI applications, with the pharmaceutical industry's digital and intelligent transformation reaching a critical turning point.

Yang emphasized that high-quality data is the key to overcoming current limitations. "While publicly available medical literature has become increasingly saturated, vast amounts of unstructured clinical data stored by healthcare institutions remain underutilized," he said. Unlocking these fragmented datasets could enable much closer alignment between AI models and real-world clinical needs, improving both research efficiency and reliability. He called for building a full-industry data ecosystem supported by favorable policies, making better use of healthcare data to create "digital twins" of patients and support more precise analysis of drug targets and patient characteristics.

Ecosystem Collaboration and the Path Forward

Zhang Hong, Deputy Party Secretary and Executive President of Zhejiang Cancer Hospital, emphasized that for AI to be adopted in hospitals, it must meet three requirements: improved efficiency, ease of use, and data security. "All three standards are indispensable," he stated, noting that clinical practice is both the ultimate testing ground for AI value and the source of feedback driving technological iteration.

Ren Jiuxuan, Deputy Director of the Digital Health Department at CAICT, revealed that China is building a dual-track evaluation system covering both laboratory testing and clinical validation, including testing for AI agents capable of multi-turn dialogue. He noted that the diversity of domestic AI products already exceeds that of the United States and predicted that high-quality medical datasets will experience explosive industry growth within the next one to two years.

Dr. Song Ning concluded: "The greatest challenge remains technological breakthroughs. Once the underlying technology advances, regulation and commercialization will gradually follow. What is required is long-term commitment." Yang Yang echoed this sentiment, stating that future competition will increasingly depend on data governance, biological validation, and cross-sector collaboration rather than algorithms alone.