SYNFAC-EDIT: Synthetic Imitation Edit Feedback for Factual Alignment in Clinical Summarization

TL;DR

This study introduces an innovative pipeline that utilizes >100B parameter GPT variants to act as synthetic feedback experts offering expert-level edit feedback, that is used to reduce hallucinations and align weaker LLMs with medical facts using two distinct alignment algorithms (DPO & SALT).

Abstract

Large Language Models (LLMs) such as GPT & Llama have demonstrated significant achievements in summarization tasks but struggle with factual inaccuracies, a critical issue in clinical NLP applications where errors could lead to serious consequences. To counter the high costs and limited availability of expert-annotated data for factual alignment, this study introduces an innovative pipeline that utilizes >100B parameter GPT variants like GPT-3.5 & GPT-4 to act as synthetic experts to generate high-quality synthetics feedback aimed at enhancing factual consistency in clinical note summarization. Our research primarily focuses on edit feedback generated by these synthetic feedback experts without additional human annotations, mirroring and optimizing the practical scenario in which medical professionals refine AI system outputs. Although such 100B+ parameter GPT variants have proven to demonstrate expertise in various clinical NLP tasks, such as the Medical Licensing Examination, there is scant research on their capacity to act as synthetic feedback experts and deliver expert-level edit feedback for improving the generation quality of weaker (<10B parameter) LLMs like GPT-2 (1.5B) & Llama 2 (7B) in clinical domain. So in this work, we leverage 100B+ GPT variants to act as synthetic feedback experts offering expert-level edit feedback, that is used to reduce hallucinations and align weaker (<10B parameter) LLMs with medical facts using two distinct alignment algorithms (DPO & SALT), endeavoring to narrow the divide between AI-generated content and factual accuracy. This highlights the substantial potential of LLM-based synthetic edits in enhancing the alignment of clinical factuality.

Figures (2)

• Figure 1: (a): The illustration of a standard simple fine-tuning pipeline. (b & c): The illustration of our proposed synthetic edit feedback generation & alignment training pipeline. In Step 1 of our synthetic edit feedback generation pipeline we generate preference data in two directions: (1) Low$\rightarrow$High[b], where we generate a factually improved summary given an unaligned existing summary system generated summary (Section \ref{['sec:L2HData']}); (2) High$\rightarrow$Low[c], where we generate a hallucinated summary given a clinical not article and a ground truth reference summary (Section \ref{['sec:H2LData']}). In Step 2 we align the Weaker LLM model using the Step 1 generated preference data using two alignment algorithms namely DPO & SALT (Section \ref{['sec:alignment_training']}).
• Figure 2: Top: % of edits made in the $D_{eval}$ for each edit type listed in Table \ref{['tab:edit_types']} of Appendix \ref{['appendixc']}. Bottom: % of edits that resulted in hallucinations/factuality aid according to the annotators. Plot legend format: <Synthetic Expert> <Data Generation Setting>