Google Research、NHS 与 Imperial College 合作发表于 Nature Cancer 的研究表明,AI 系统在乳腺癌筛查中可检出 25% 传统方法遗漏的 interval cancers。该技术在显著提高敏感性(真阳性检出率)的同时,未对特异性(假阳性)产生明显影响,还能减少医护人员筛查工作量,并加速向医生和患者反馈诊断结果。
Excited to see this joint collaboration between @GoogleResearch, @NHSuk and @imperialcollege showing AI's potential to detect 25% of the interval cancers previously missed by conventional methods. Additionally, the research found AI can reduce screening workloads, and give results back to clinicians and patients faster.
This first figure from the @NatureCancer article shows how the study was set up, and the second figure shows that the AI system dramatically increases sensitivity (detecting true positives) without significantly affecting specificity (false positives).
Learn more ⬇️
Blog: https://blog.google/innovation-and-ai/technology/health/google-ai-breast-cancer-detection/
Nature Cancer paper: https://www.nature.com/articles/s43018-026-01127-0