Current Research Interests
- AI-Med Datasets. Constructing large-scale, multi-source biomedical databases to provide the foundational data infrastructure for training, benchmarking, and validating biosignal AI models across diverse clinical populations.
- Biosignal Foundation Models. Developing generalist foundation models by leveraging large-scale, multi-modality databases to enable robust representation learning and zero-shot generalization across diverse clinical tasks.
- Biosignal Specialized MLLMs and Agents. Constructing domain-specific multimodal large language models and clinical agents to bridge the semantic gap between time-series biosignals and clinical language, and to enable tool-augmented clinical reasoning and decision support.
- Digital Biomarker Discovery. Establishing a systematic paradigm to transform high-dimensional biosignals into structured, quantifiable metrics that capture latent physiological aging, organ function, and disease risk.
- AI-enhanced Health Devices. Inventing and enhancing AI-ECG products, device-repurposing techniques, and generative pipelines to expand the diagnostic capacity of low-cost, ubiquitously deployable sensors.
- Real-World Clinical Validation. Conducting rigorous real-world studies to evaluate the clinical efficacy and implementation pathways of AI-ECG interventions.
We are recruiting team members who have a strong passion of AI for digital health. If you are interested, please send an email with your CV attached:
- PostDoc with clinical research methodology and writing skills, coding skills not required
[2026/05] 🔥 Try our latest ECG interpretation MLLM in ECG-R1 at Code. Paper in ICML 2026!