About Me
Hello, I’m Ruinan Jin, and I am currently the PhD student focus on medical and trustworthy machine learning. I hold a B.S. (Hon.) in Computer Science from the University of British Columbia (UBC). I have also gained rish industry experiences at Google, Amazon Web Services and Sierra Wireless, working on machine learning research and IoT solutions. Below is a list of my research interests, and I am open to various forms of collaboration. If you share any of these interests, please feel free to reach out!
Research Directions and Past Experiences
My research focuses on building trustworthy machine learning systems for medicine, with an emphasis on foundation and vision-language models (VLMs), medical AI agents, and clinical reasoning and dialogue systems deployed in real-world healthcare settings. I am particularly interested in understanding and improving how medical AI systems behave beyond simulation, including their reliability, safety, fairness, and robustness under real-world clinical use (e.g., longitudinal decision-making, distribution shift, and heterogeneous patient populations).
A central theme of my work is to diagnose failure modes in medical AI systems, such as spurious reasoning, memorization, privacy leakage, bias, and vulnerability to adversarial or distributional perturbations, and to develop defenses and evaluation methodologies with measurable guarantees. My research spans both model development and rigorous evaluation, with a strong focus on high-impact, sensitive domains such as healthcare.
Security & Safety (Failure Modes and Defenses)
Backdoor attacks and defenses in medical and multimodal models; robustness of foundation models and medical AI agents to spurious correlations, adversarial perturbations, and deployment-time shifts; safety-oriented evaluation for clinical reasoning and dialogue systems.Privacy & Data Governance (Trust and Unlearning)
Memorization and data exposure in modern medical VLMs and agents; implications for privacy leakage, fairness, and backdoors; machine unlearning; differential privacy in centralized and federated settings; and privacy-preserving synthetic data generation for sensitive medical domains.Reliability & Generalization (From Simulation to Real World)
Generalization behavior of medical foundation models, VLMs, and federated learning systems under distribution shift, heterogeneous clients, and longitudinal use; robustness-oriented training and evaluation protocols designed to reflect real-world clinical deployment.Fairness & Responsible Evaluation (Clinical Risk Awareness)
Bias measurement and group-fairness constraints in medical foundation models and vision-language systems; fairness-aware benchmarks and evaluation frameworks that connect algorithmic behavior to downstream clinical risk and real-world impact.
Open-source Projects
FairMedFM: Collection of fairness evaluation on Foundation Models for medical image analysis.
MedVLMBench: Collection of medical vision-language models for medical image analysis.
RVCBench: Collection of robustness evaluation on modern TTS and voice clone models.
Community Service
Conferences
- NeurIPS (2025) Top Reviewer
- MICCAI (2025)
Journals
- IEEE Transactions on Medical Imaging Distinguished Reviewer (Bronze)
- IEEE Transactions on Pattern Analysis and Machine Intelligence
- IEEE Transactions on Neural Networks and Learning Systems
- IEEE Journal of Biomedical and Health Informatics
- IEEE Transactions on Artificial Intelligence
- ACM Computing Surveys
- Medical Image Analysis
- Neural Networks
- Pattern Recognition