Evaluating ChatGPT on Medical Information Extraction Tasks: Performance, Explainability and Beyond
Wei Zhu
TL;DR
This study systematically evaluates ChatGPT on medical information extraction (MedIE) tasks across six benchmarks, using five evaluation dimensions: performance, explainability, confidence, faithfulness, and uncertainty. It employs carefully designed prompts and expert annotations to compare ChatGPT against fine-tuned baselines and SOTA methods. The findings show that ChatGPT underperforms relative to task-specific fine-tuning, though it yields high-quality explanations and generally faithful outputs, with notable over-confidence and generation-induced uncertainty. The work highlights practical implications for deploying LLMs in MedIE and points to future directions such as improved prompting, calibration, and retrieval-enhanced approaches.
Abstract
Large Language Models (LLMs) like ChatGPT have demonstrated amazing capabilities in comprehending user intents and generate reasonable and useful responses. Beside their ability to chat, their capabilities in various natural language processing (NLP) tasks are of interest to the research community. In this paper, we focus on assessing the overall ability of ChatGPT in 4 different medical information extraction (MedIE) tasks across 6 benchmark datasets. We present the systematically analysis by measuring ChatGPT's performance, explainability, confidence, faithfulness, and uncertainty. Our experiments reveal that: (a) ChatGPT's performance scores on MedIE tasks fall behind those of the fine-tuned baseline models. (b) ChatGPT can provide high-quality explanations for its decisions, however, ChatGPT is over-confident in its predcitions. (c) ChatGPT demonstrates a high level of faithfulness to the original text in the majority of cases. (d) The uncertainty in generation causes uncertainty in information extraction results, thus may hinder its applications in MedIE tasks.