Ongoing
Natural Language Processing for Triaging Incoming Referral Letters
Using different models to predict the inflammatory arthrisis disease.
Authors
Kiana Aghakasiri, Mohamed Abdalla, Carrie Ye
Current & Past Research
Recent Work
Ongoing
Not What the Doctor Ordered: Surveying LLM-based De-identification and Quantifying Clinical Information Loss
This survey identifies key limitations in current LLM-based de-identification literature, highlighting inconsistent reporting and the risks of altering clinically relevant information while removing patient identifiers.
Authors
Kiana Aghakasiri, Noopur Zambare, JoAnn Thai, Carrie Ye, Mayur Mehta, J Ross Mitchell, Mohamed Abdalla
Towards Fair and Efficient De-identification: Quantifying Multi-cultural Generalizability
Evaluating how transformer models generalize clinical de-identification across different formats, cultures, and genders, introducing a benchmark for Mandarin, Hindi, Spanish, and French variants.
Authors
Noopur Zambare, Kiana Aghakasiri, Carissa Lin, Carrie Ye, J Ross Mitchell,Mohamed Abdalla
Beyond a Million Tokens: Benchmarking and Enhancing Long-Term Memory in LLMs
An investigation into the architectural limits of LLMs when processing extremely long contexts, proposing methods to improve retrieval accuracy and memory retention beyond traditional token limits.
Authors
Mohammad Tavakoli, Alireza Salemi, Carrie Ye, Mohamed Abdalla, Hamed Zamani, J Ross Mitchell