浙江发布胃癌影像筛查AI模型 用“平扫CT+AI”识别早期胃癌

Core Viewpoint - The introduction of the world's first AI model for gastric cancer screening, DAMO GRAPE, represents a significant advancement in early detection and intervention, potentially increasing the five-year survival rate from 35% to over 90% if cancer is detected early [3][5]. Group 1: AI Model Development and Application - The DAMO GRAPE model utilizes plain CT imaging to identify early gastric cancer lesions and has undergone large-scale clinical research involving nearly 100,000 individuals across 20 centers [3]. - The model's initial screening significantly reduces the high-risk population for gastric cancer, achieving detection rates of 24.5% and 17.7% in two regional hospitals, with approximately 40% of detected cases being asymptomatic [4]. - The AI model demonstrates a sensitivity of 85.1% and specificity of 96.8%, surpassing radiologist diagnoses by 21.8% and 14.0%, respectively, making early detection of gastric cancer via plain CT feasible [4]. Group 2: Clinical Impact and Future Plans - The AI model has shown the ability to detect gastric cancer 2 to 10 months earlier than traditional methods, as evidenced by a case where a patient was diagnosed with advanced gastric cancer that could have been identified earlier with AI monitoring [4]. - The model is being deployed in high-incidence areas for gastric cancer across the country, with plans for large-scale promotion both domestically and internationally [5].