Research Associate
Adrian Serapio is interested in the development, evaluation, and translation of large language models (LLMs) and vision-language models (VLMs) for radiological applications. Previously, he has fine-tuned an LLM to summarize radiology reports to generate radiology report impressions. Currently, he is evaluating the performance of different LLMs in summarizing clinical notes for more comprehensive clinical histories and evaluating the reliability of vision-language models for generating radiology reports given clinical histories.