Our research focuses on the development, evaluation, and trustworthy use of AI in medicine. We work across language, imaging, and multimodal health data, with particular interests in biomedical NLP, medical LLMs, multimodal and foundation models, ophthalmology and medical imaging AI, and rigorous evaluation for real-world applications in biomedicine and healthcare.
We develop and evaluate large language models for biomedical and clinical text. Our work includes the development of domain-specific LLMs and broad applications in biomedical natural language processing.
Selected recent work:
A Federated and Parameter-Efficient Framework for Large Language Model Training in Medicine, Preprint
Benchmarking Large Language Models for Biomedical Natural Language Processing Applications and Recommendations, Nature Communications, 2025; selected for the AMIA 2025 Year in Review
Outpatient Reception via Collaboration Between Nurses and a Large Language Model: A Randomized Controlled Trial, Nature Medicine, 2024
We build foundation and multimodal models that integrate language, medical images, and other health data. Our goal is to support scalable representation learning and clinically relevant prediction across medical tasks.
Selected recent work:
Are Multimodal LLMs Ready for Clinical Dermatology? A Real-World Evaluation in Dermatology, Preprint
Building the world’s first truly global medical foundation model, Nature Medicine, 2025
Medical Foundation Large Language Models for Comprehensive Text Analysis and Beyond, npj Digital Medicine, 2025
We develop AI methods for ophthalmology and broader medical imaging applications, including disease screening, diagnosis, prognosis, and multimodal reasoning. We are especially interested in open and clinically useful models for eye care.
Selected recent work:
LEME: Open Large Language Models for Ophthalmology with Advanced Reasoning and Clinical Validation, Preprint
VOLMO: Versatile and Open Large Models for Ophthalmology, Preprint
Ophthalmological Question Answering and Reasoning Using OpenAI o1 vs Other Large Language Models. JAMA Ophthalmology, 2025
LMOD+: A Comprehensive Multimodal Dataset and Benchmark for Developing and Evaluating Multimodal Large Language Models in Ophthalmology, ACM Transactions on Computing for Healthcare, 2025; the preliminary version was published in NAACL Findings
We study how AI systems in medicine should be evaluated, validated, and trusted. Our work focuses on factuality, memorization, generalization, real-world evaluation, and responsible deployment in biomedicine and healthcare.
Selected recent work:
Memorization in Large Language Models in Medicine: Prevalence, Characteristics, and Implications. Accepted in principle, Nature Communications
Reasoning-Driven Large Language Models in Medicine, The Lancet Digital Health, 2026
AI workflow, external validation, and development in eye disease diagnosis, JAMA network open, 2025, NLM Honor Award