Vimig Socrates, MS

Project Title: Evaluation of Large Language Models as Automated Deprescribing Tools in Older Patients

Vimig Socrates is a 5th year PhD Student in Computational Biology and Bioinformatics under the Section for Biomedical Informatics & Data Science. He is a member of the HarmonAI Health Lab (HAIL) lab (led by Dr. Andrew Taylor, MD, MHS). With a background in Computer Science, he is primarily interested in gaining a deeper understanding of the intersection between natural language processing methods and cognition in clinical reasoning. His previous work includes the development of language models for emergency medicine applications, including UTI symptoms and incarceration status detection. His CERSI scholar project aims to use large language models to interpret and evaluate geriatric deprescribing criteria (STOPP/Beers) on emergency room patients. The study has the potential to reduce the rate of adverse drug events to the ED, improve geriatric quality of life, and reduce the health system's overall medication burden.