AI 風險評估模型
La Trobe University 推出其 BowelRec AI 工具
La Trobe University 研究人員開發出 SÉMIL，一款 AI 模型，可分析數碼病理切片，識別出復發風險較高的第二期大腸癌病人。系統評估腫瘤在浸潤前緣的生長模式——這項預後特徵一向難以由病理學家作出一致分類。
SÉMIL 使用 388 宗第三期大腸癌病例進行訓練，並於超過 1,220 名第二期病人身上進行驗證。模型識別出一個高風險群組，其五年內復發風險約為一般病人的兩倍，且不受既有臨床風險因素影響。研究人員亦發現，將 SÉMIL 的評估結果與病理學家的判斷結合，可提升風險分類的準確度。
對臨床醫生而言，這款模型最終有望在毋須額外組織樣本或昂貴檢測的情況下，支援更個人化的跟進及治療決策。是項研究反映市場對運用 AI 輔助病理學進行癌症風險分層的興趣日益增加。
英文原文
La Trobe University researchers developed SÉMIL, an AI model that analyzes digital pathology slides to identify stage 2 bowel cancer patients at greater risk of relapse. The system evaluates tumor growth patterns at the invasive front, a prognostic feature that can be difficult for pathologists to classify consistently.
SÉMIL was trained using 388 stage 3 bowel cancer cases and validated across more than 1,220 stage 2 patients. The model identified a higher-risk group with roughly twice the risk of relapse within five years, independent of established clinical risk factors. Researchers also found that combining SÉMIL’s assessment with a pathologist’s evaluation improved risk classification.
For clinicians, the model could eventually support more individualized follow-up and treatment decisions without requiring additional tissue samples or costly tests. The research reflects growing interest in AI-assisted pathology for cancer risk stratification.
- 來源
- Trend Hunter
- 發布
- 2026-08-21
- 品類
- Ai
- 出處
- mobihealthnews