China News Service, Guangzhou, October 10 (Reporter Cai Minjie) The lung cancer pathological gene multimodal large model DeepGEM2.0 was released in Guangzhou on the 10th. It can predict all genes of 12 mainland-approved targeted drugs for lung cancer at once in as little as one minute, and quickly pre-screen people who may have opportunities for targeted drugs. The data shows that the prediction accuracy can reach up to 90%. Zhong Nanshan, academician of the Chinese Academy of Engineering, attended the launch event.
The AI Empowered Lung Cancer Precision Diagnosis and Treatment Summit Forum and the National Center for Respiratory Medicine’s Lung Cancer Precision Diagnosis and Treatment Standardization Project Launch Meeting were held in Guangzhou. Lung cancer is a malignant tumor, and genetic testing is the prerequisite for targeted therapy. However, due to factors such as testing accessibility and economic conditions, a considerable proportion of patients fail to complete necessary genetic testing and miss opportunities for targeted and precise treatment.
In 2025, the First Affiliated Hospital of Guangzhou Medical University joined hands with KingMed and Tencent to release DeepGEM, a large multi-modal pathological gene model that can "identify genes by looking at pictures". It uses pathological slices as input to predict the probability of lung cancer-related gene mutations and their spatial distribution within tissues.
After nearly a year of multi-center data verification and clinical polishing, and relying on iterative optimization of data from nearly 8,000 digital pathology slides from 31 provinces across the country, DeepGEM2.0 can complete "identifying genes by looking at pictures" in as little as 1 minute. It will be lightweight and deployed in more than 500 medical institutions in the future. This technology is also expected to reduce the cost of lung cancer genetic testing to several hundred yuan.
Zhong Nanshan, an academician of the Chinese Academy of Engineering, said that in the new era of AI, scientific research is about using new technologies such as AI to make them affordable, easy, cheap, effective and safe for ordinary people to use. The significance of DeepGEM2.0 lies not only in how to apply the most advanced technology to the grassroots level, but also in that it embodies the integration of industry, academia, research, medicine, teaching and research.
Liang Wenhua, deputy director of the First Affiliated Hospital of Guangzhou Medical University and deputy director of the National Center for Respiratory Medicine, said that with DeepGEM2.0, grassroots hospitals will be able to carry out preliminary stratified triage of patients based on AI pre-screening results: patients with high probability of specific genes predicted by AI can directly complete the widely popularized targeted PCR (polymerase chain reaction) and immunohistochemistry verification locally or nearby, without having to travel long distances for medical treatment.
A multi-center scientific research project jointly participated by 50 medical institutions in the mainland was also launched on the same day, which will provide high-level evidence-based basis for the application of AI-assisted pathological diagnosis in precise treatment of lung cancer.
