Calvados at MEDIQA-Chat 2023: Improving Clinical Note Generation with Multi-Task Instruction Finetuning

Kirill Milintsevich, Navneet Agarwal


Abstract
This paper presents our system for the MEDIQA-Chat 2023 shared task on medical conversation summarization. Our approach involves finetuning a LongT5 model on multiple tasks simultaneously, which we demonstrate improves the model’s overall performance while reducing the number of factual errors and hallucinations in the generated summary. Furthermore, we investigated the effect of augmenting the data with in-text annotations from a clinical named entity recognition model, finding that this approach decreased summarization quality. Lastly, we explore using different text generation strategies for medical note generation based on the length of the note. Our findings suggest that the application of our proposed approach can be beneficial for improving the accuracy and effectiveness of medical conversation summarization.
Anthology ID:
2023.clinicalnlp-1.56
Volume:
Proceedings of the 5th Clinical Natural Language Processing Workshop
Month:
July
Year:
2023
Address:
Toronto, Canada
Editors:
Tristan Naumann, Asma Ben Abacha, Steven Bethard, Kirk Roberts, Anna Rumshisky
Venue:
ClinicalNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
529–535
Language:
URL:
https://aclanthology.org/2023.clinicalnlp-1.56
DOI:
10.18653/v1/2023.clinicalnlp-1.56
Bibkey:
Cite (ACL):
Kirill Milintsevich and Navneet Agarwal. 2023. Calvados at MEDIQA-Chat 2023: Improving Clinical Note Generation with Multi-Task Instruction Finetuning. In Proceedings of the 5th Clinical Natural Language Processing Workshop, pages 529–535, Toronto, Canada. Association for Computational Linguistics.
Cite (Informal):
Calvados at MEDIQA-Chat 2023: Improving Clinical Note Generation with Multi-Task Instruction Finetuning (Milintsevich & Agarwal, ClinicalNLP 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.clinicalnlp-1.56.pdf