Hongjian Zhou

Hongjian Zhou

DPhil candidate in Engineering Science
University of Oxford

My research focuses on AI scientist systems for medicine and biomedical research.

I develop language-model agents that work with biomedical literature, tools, and evidence, and study how to evaluate their capabilities and failure modes reproducibly. I am especially interested in making these systems capable, reliable, and scientifically useful.

I am a Clarendon Scholar working with Professor David A. Clifton in Oxford's Computational Health Informatics group.

Google ScholarOxfordGitHubXLinkedIn
Hongjian Zhou

Research interests

I study how AI agents can support biomedical discovery to help people live longer, healthier lives while remaining evidence-grounded and safe.

  • Biomedical AI scientists

    AI agents that use literature, data, and scientific tools to support scientific discovery in biomedicine.

  • Evaluation of AI scientists

    Open environments and benchmarks for measuring end-to-end research capability with real-world signals, identifying failure modes, and making agents more capable, reliable, and safe.

  • Reliable AI scientists

    Methods and evaluations for evidence-grounded reasoning, especially when information is incomplete, conflicting, or misleading in the real world.

Recent news

Selected publications

All publications on Scholar

2025

Application of large language models in medicine

Fenglin Liu*, Hongjian Zhou*, Boyang Gu, Xinyu Zou, Jinfa Huang, et al.

Nature Reviews Bioengineering 3, 445–464, 2025

Reviews medical language models, their applications and evaluation, and the barriers to dependable clinical use.

[paper]

* Equal contribution

Background

I was born in Cixi, China, and moved to Oakville, Canada, at an early age. Outside research, I enjoy making music; some of my work is available on NetEase Cloud Music.