New AI method uses patient labels to train cancer registry models

Researchers have developed a novel framework using Attention-Based Multiple Instance Learning (ABMIL) to train deep learning models for cancer registry tasks without requiring individual report annotations. This method leverages existing patient-level labels, which are typically not directly linked to individual pathology reports, to create a high-quality training dataset. The ABMIL approach demonstrated effectiveness in tumor group classification at the BC Cancer Registry, achieving a macro F1 score of 0.83 and outperforming other methods. AI

IMPACT This approach could significantly improve the efficiency and accuracy of cancer registry workflows by enabling better use of existing data.

RANK_REASON The cluster contains an academic paper detailing a new machine learning method for a specific domain. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• ABMIL
• alphaXiv
• arXiv
• Attention-Based Multiple Instance Learning
• BC Cancer Registry
• CatalyzeX
• DagsHub
• Gotit.pub
• Hugging Face
• ScienceCast

• paper
• model release
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →