Development and Evaluation of a Digital Scribe: Conversation Summarization Pipeline for Emergency Department Counseling Sessions towards Reducing Documentation Burden

Abstract

Objective: We present a proof-of-concept digital scribe system as an ED clinical conversation summarization pipeline and report its performance.

Materials and methods: We use four pre-trained large language models to establish the digital scribe system: T5-small, T5-base, PEGASUS-PubMed, and BART-Large-CNN via zero-shot and fine-tuning approaches. Our dataset includes 100 referral conversations among ED clinicians and medical records. We report the ROUGE-1, ROUGE-2, and ROUGE-L to compare model performance. In addition, we annotated transcriptions to assess the quality of generated summaries.

Results: The fine-tuned BART-Large-CNN model demonstrates greater performance in summarization tasks with the highest ROUGE scores (F1_ROUGE-1=0.49, F1_ROUGE-2=0.23, F1_ROUGE-L=0.35) scores. In contrast, PEGASUS-PubMed lags notably (F1_ROUGE-1=0.28, F1_ROUGE-2=0.11, F1_ROUGE-L=0.22). BART-Large-CNN's performance decreases by more than 50% with the zero-shot approach. Annotations show that BART-Large-CNN performs 71.4% recall in identifying key information and a 67.7% accuracy rate.

Discussion: The BART-Large-CNN model demonstrates a high level of understanding of clinical dialogue structure, indicated by its performance with and without fine-tuning. Despite some instances of high recall, there is variability in the model's performance, particularly in achieving consistent correctness, suggesting room for refinement. The model's recall ability varies across different information categories.

Conclusion: The study provides evidence towards the potential of AI-assisted tools in reducing clinical documentation burden. Future work is suggested on expanding the research scope with larger language models, and comparative analysis to measure documentation efforts and time.

Keywords: Emergency department; Natural Language Processing; Text Summarization; clinical conversation; documentation burden; pre-trained language model.

Figure 1.

Study design

Figure 2.

A histogram of the number of tokens per transcript. The tokens were generated for this graph using the BART tokenizer.[40] The vertical line represents the maximum input length of the models, 1024 tokens, and 82% of transcripts clusters to the left of this line.

Figure 3.

Histograms showing information recalled (without consideration of correctness) [right] and correctly recalled information [left] by a generated summary that appeared in the ground truth summary.

Figure 4.

Example generated and nurse note samples with LINK and CORRECT annotations