Juhyun Lee

Hi, I’m Julie, a researcher passionate about AI for Healthcare. I’m currently pursuing M.S. in Computer Science at the University of North Carolina at Chapel Hill, and I hold a B.S. in Applied Statistics from Chung-Ang University in South Korea.

I interned at LG AI Research, where I developed multimodal frameworks for Alzheimer’s disease phenotype prediction using incomplete patient records from ADNI. The model integrates imaging, genetic, and clinical data to improve diagnostic accuracy and was accepted to the NeurIPS 2025 ImageOmics Workshop. I also designed AI agents that automate biomedical analysis pipelines including gene expression profiling on colorectal cancer data using LangChain.

Currently, my projects focus on:

1. Healthcare dialogue systems — applied few-shot prompt engineering and efficient fine-tuning strategies for persona-adaptive medical chatbots with vLLM-optimized inference.
2. Whole-slide image analysis — extracting features from ResNet and UNI encoders, and applying CLAM for binary classification of smoking status using TCGA-BLCA pathology data.

My goal is to build large vision and language models that can reason across multimodal biomedical data, advancing clinical understanding and making healthcare more accessible and equitable.