NLP4Health: Multilingual Clinical Dialogue Summarization and QA with mT5 and LoRA

Moutushi Roy, Dipankar Das


Abstract
In this work, we present NLP4Health, a unified and reproducible pipeline to accomplish the tasks of multilingual clinical dialogue summarization and question answering (QA). Our system fine-tunes the multilingual sequence-to-sequence model google/mt5-base along with parameter-efficient Low-Rank Adaptation (LoRA) modules to support ten Indian languages. For each clinical dialogue, the model produces (1) a free-text English summary, (2) an English structured key–value (KnV) JSON summary, and (3) QA responses in the dialogue’s original language. We conducted preprocessing, fine-tuning, and inference, and evaluated across QA, textual, and structured metrics, analyzing performance in low-resource settings. The adapter weights, tokenizer, and inference scripts are publicly released to promote transparency and reproducibility.
Anthology ID:
2025.nlpai4health-main.10
Volume:
NLP-AI4Health
Month:
December
Year:
2025
Address:
Mumbai, India
Editors:
Parameswari Krishnamurthy, Vandan Mujadia, Dipti Misra Sharma, Hannah Mary Thomas
Venues:
NLP-AI4Health | WS
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
93–97
Language:
URL:
https://aclanthology.org/2025.nlpai4health-main.10/
DOI:
Bibkey:
Cite (ACL):
Moutushi Roy and Dipankar Das. 2025. NLP4Health: Multilingual Clinical Dialogue Summarization and QA with mT5 and LoRA. In NLP-AI4Health, pages 93–97, Mumbai, India. Association for Computational Linguistics.
Cite (Informal):
NLP4Health: Multilingual Clinical Dialogue Summarization and QA with mT5 and LoRA (Roy & Das, NLP-AI4Health 2025)
Copy Citation:
PDF:
https://aclanthology.org/2025.nlpai4health-main.10.pdf