Self-Prompting Small Language Models for Privacy-Sensitive Clinical Information Extraction

TL;DR

This paper presents a framework enabling small language models to self-generate and optimize prompts for extracting clinical entities from dental notes, improving privacy-sensitive clinical information extraction.

Contribution

It introduces a locally deployable, self-prompting approach with fine-tuning and preference optimization for small models in clinical NLP tasks.

Findings

Qwen2.5-14B-Instruct achieved the highest baseline performance.

Post-training with DPO improved model F1 scores significantly.

Task-specific evaluation revealed substantial performance variation among models.

Abstract

Clinical named entity recognition from dental progress notes is challenging because documentation is highly unstructured, domain-specific, and often privacy-sensitive. We developed a locally deployable framework that enables small language models to self-generate, verify, refine, and evaluate entity-specific prompts for extracting multiple clinical entities from dental notes. Using 1,200 annotated notes, we evaluated candidate open-weight models with multi-prompt ensemble inference and further adapted selected models using QLoRA-based supervised fine-tuning and direct preference optimization. Model performance varied substantially, highlighting the need for task-specific evaluation rather than reliance on generic benchmarks. Qwen2.5-14B-Instruct achieved the strongest baseline performance. After DPO, Qwen2.5-14B-Instruct and Llama-3.1-8B-Instruct achieved micro/macro F1 scores of 0.864/0.837 and 0.806/0.797, respectively. These findings suggest that automated prompt optimization combined with lightweight preference-based post-training can support scalable clinical information extraction using locally deployed small language models.