Ruinan Jin
Trustworthy medical AI · foundation models · clinical reasoning
PhD Researcher
Trustworthy Medical AI
I am Ruinan Jin, a PhD researcher working on trustworthy machine learning systems for medicine. I earned my B.S. (Honours with Distinction) in Computer Science from the University of British Columbia. My industry experience includes machine-learning and software engineering work at Google, Amazon Web Services, and Sierra Wireless.
My research centers on foundation and vision-language models, medical AI agents, and clinical reasoning systems that work reliably beyond simulation. I study how these systems behave under longitudinal decision-making, distribution shift, and heterogeneous patient populations—then develop practical evaluations and defenses with measurable guarantees.
I am actively open to research collaborations on trustworthy medical AI, foundation models, federated learning, and clinical AI evaluation. Reach me by email, LinkedIn, or GitHub.
Research directions
- Security & safety — backdoor attacks and defenses in medical and multimodal models; adversarial robustness; safety-oriented evaluation for clinical AI.
- Privacy & data governance — memorization and data exposure in medical VLMs, machine unlearning, differential privacy, and synthetic data for sensitive domains.
- Reliability & generalization — foundation models under distribution shift, heterogeneous clients, and longitudinal clinical use.
- Fairness & responsible evaluation — bias measurement and group fairness tied to downstream clinical risk and real-world impact.
Open-source projects
- FairMedFM — fairness benchmarking for medical imaging foundation models.
- MedVLMBench — an evaluation benchmark for medical vision-language models.
- RVCBench — robustness evaluation for modern text-to-speech and voice-cloning models.
Community service
I review for NeurIPS (Top Reviewer, 2025), MICCAI, IEEE Transactions on Medical Imaging (Distinguished Reviewer, Bronze), IEEE Transactions on Dependable and Secure Computing, IEEE TPAMI, IEEE TNNLS, IEEE JBHI, IEEE TAI, ACM Computing Surveys, Medical Image Analysis, Neural Networks, and Pattern Recognition.
selected publications
- MICCAI
DUCX: Decomposing Unfairness in Tool-Using Chest X-ray AgentsIn International Conference on Medical Image Computing and Computer-Assisted Intervention, 2026Oral; early acceptance rate 9% - TNNLS
Forgettable Federated Linear Learning with Certified Data UnlearningIEEE Transactions on Neural Networks and Learning Systems, 2026 - Preprint
RVCBench: Benchmarking the Robustness of Voice Cloning Across Modern Audio Generation Models2026 - NeurIPS
FairMedFM: Fairness Benchmarking for Medical Imaging Foundation ModelsIn Advances in Neural Information Processing Systems, 2024Acceptance rate 25.8% - MedIA
Backdoor Attack and Defense in Federated Generative Adversarial Network-based Medical Image SynthesisMedical Image Analysis, 2023Impact Factor 10.9