跳到正文
arXiv cs.CL· Ilia Kopanichuk, Daniil Kartashev, Yuri Bolkonsky, Petr Anokhin, Vladimir Makharev, Iaroslav Bespalov, Dmitry V. Dylov·· 9 小时前AI 评分31

基于监督候选生成与选择性 LLM 调用的高效 ICD-10 编码预测

Call-Efficient ICD-10 Code Prediction via Supervised Candidate Generation and Selective LLM Invocation

AI 导读

一种结合 Transformer 嵌入、树模型监督候选生成与门控 LLM 辅助的 ICD-10 编码预测流程,最终系统 Pre-LLM fusion 在锁定测试集上达到 83.19% Top-1 准确率,仅对 31.97% 的病例调用 LLM。

正文 · 原文

View PDF HTML (experimental)

Abstract:Automatic assignment of ICD-10 codes from short clinical diagnosis text is a practical medical text classification and normalization task. The task is difficult because diagnosis phrases are short and noisy, the ICD-10 label space is hierarchical, and the class distribution is strongly imbalanced. This paper presents a call-efficient ICD-10 code prediction pipeline that combines transformer embeddings, tree-based supervised candidate generation, candidate-level reranking, and a gated LLM assistance. The evaluation uses a leakage-safe validation protocol: model and threshold selection are based on out-of-fold cross-validation predictions, while final reporting is performed on a locked test set with no shared source-file groups. The supervised conditional parent-code candidate generator reaches 74.32% Top-1 accuracy on the locked test set and 93.05% Top-10 candidate coverage. A direct zero-shot GPT-5.6-Sol baseline reaches 81.01% Top-1 accuracy, but requires an invocation of this large model for every diagnosis. The final deployable system, Pre-LLM fusion, reaches 81.24% OOF-CV accuracy and 83.19% locked-test accuracy while calling the LLM for only 31.97% of cases. This improves the locked-test error rate from 25.68% for the supervised parent-code baseline to 16.81%, corresponding to a 34.5% relative error reduction. The results show that the most effective role of the LLM is controlled reranking on uncertain examples rather than unrestricted coding or full-dataset invocation.
Comments: 8 pages, 8 figures, 4 tables
Subjects: Computation and Language (cs.CL)
Cite as: arXiv:2611.21891 [cs.CL]
  (or arXiv:2611.21891v1 [cs.CL] for this version)
  https://doi.org/10.48550/arXiv.2611.21891

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Vladimir Makharev [view email]
[v1] Mon, 16 Nov 2026 21:05:59 UTC (129 KB)

来源:arXiv cs.CL · arxiv.org