AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Towards Semi-Structured Automatic ICD Coding via Tree-based Contrastive Learning

Chang Lu, Chandan K. Reddy, Ping Wang, Yue Ning

NeurIPS 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Partial and multi-label learning · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add contrastive pre-training and masked section training (CM)MSMN · MIMIC-50Macro F1: 68.1 -> 69.1Improved
add CMMultiResCNN · MIMIC-50Macro F1: 60.8 -> 62.2Improved
add CMPLM-ICD · MIMIC-50Macro F1: 64.5 -> 65.2Improved
add CMMultiResCNN · MIMIC-rare-50Macro F1: 11.2 -> 22.8Improved
add CMHierarchical · MIMIC-rare-50Macro F1: 23.1 -> 32.0Improved
add CMMSMN · MIMIC-rare-50Macro F1: 23.7 -> 31.2Improved
add CMMSMN · MIMIC-fullMacro F1: 10.3 -> 11.4Improved
add CMPLM-ICD · MIMIC-fullMacro F1: 10.4 -> 10.6No clear effect
add CMPLM-ICD · MIMIC-fullMicro F1: 59.8 -> 60.0No clear effect
use only contrastive pre-training with Jaccard similarity instead of tree edit distance (ablation)JointLAAT, EffectiveCAN, MSMN · MIMIC-50performanceHurt
use tree edit distance instead of Jaccard similarity for contrastive pre-training (ablation)JointLAAT, EffectiveCAN, MSMN · MIMIC-50performanceImproved
add only masked section training versus only contrastive pre-training (ablation)backbone ICD models · MIMIC-50performanceMixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy