This paper presents our contribution to the RadSum23 shared task organized as part of the BioNLP 2023. We compared state-of-the-art generative language models in generating high-quality summaries from radiology reports. A two-stage fine-tuning approach was introduced for utilizing knowledge learnt from different datasets. We evaluated the performance of our method using a variety of metrics, including BLEU, ROUGE, bertscore, CheXbert, and RadGraph. Our results revealed the potentials of different models in summarizing radiology reports and demonstrated the effectiveness of the two-stage fine-tuning approach. We also discussed the limitations and future directions of our work, highlighting the need for better understanding the architecture design’s effect and optimal way of fine-tuning accordingly in automatic clinical summarizations.

This paper reports on the performance of Edinburgh_UCL_Health’s models in the Social Media Mining for Health (SMM4H) 2022 shared tasks. Our team participated in the tasks related to the Identification of Adverse Drug Events (ADEs), the classification of change in medication (change-med) and the classification of self-report of vaccination (self-vaccine). Our best performing models are based on DeepADEMiner (with respective F1= 0.64, 0.62 and 0.39 for ADE identification), on a GloVe model trained on Twitter (with F1=0.11 for the change-med) and finally on a stack embedding including a layer of Glove embedding and two layers of Flair embedding (with F1= 0.77 for self-report).