AI Pathology Test Outperforms Genomic Gold Standard in Breast Cancer

An AI pathology test based on a 300M-parameter vision foundation model analyzes standard H&E slides without sequencing. Outperforming the genomic gold standard Oncotype DX across multiple cohorts, it demonstrates strong independent prognostic value, especially in underserved subgroups like Black patients.

2026-07-06 ~ 2026-07-06 · 11 related posts

• 多模态AI从病理切片预测乳腺癌复发风险，论文发表于Nature Communications — kjgeras · 2026-07-06
• AI病理模型从H&E切片预测乳腺癌复发风险 — kjgeras · 2026-07-06
• AI病理测试在5个独立队列3502例患者中C指数达0.71 — kjgeras · 2026-07-06
• AI病理测试在HR+/HER2-患者中超越Oncotype DX金标准 — kjgeras · 2026-07-06
• AI病理测试在多变量模型中仍保持独立强预后价值 — kjgeras · 2026-07-06
• AI病理模型在三阴性和HER2+乳腺癌亚型中均有效 — kjgeras · 2026-07-06
• AI病理测试在黑人患者群体和晚期复发预测中同样有效 — kjgeras · 2026-07-06
• AI病理测试无需额外组织处理可直接用于活检标本 — kjgeras · 2026-07-06
• Kestrel：3亿参数病理视觉基础模型在4亿切片块上自监督预训练 — kjgeras · 2026-07-06
• 注意力多示例学习将巨像素切片压缩为单一病理风险评分 — kjgeras · 2026-07-06
• Ataraxis AI将医学基础模型扩展至Falcon并拓展癌症临床应用 — kjgeras · 2026-07-06