About
I am a Postdoctoral Research Associate in Department of Computer Science, Dartmouth College, working with Prof. Andrew Campbell. Previously, I received my Ph.D. in Department of Computer Science and Technology at University of Cambridge, advised by Prof. Cecilia Mascolo.
I develop foundation models for longitudinal and multimodal health sensing data, spanning wearable sensors, physiological signals, and electronic health records. Real-world health data are heterogeneous and incomplete, with sensor configurations, available supervision, and population distributions varying substantially across datasets and deployment settings. My research addresses these challenges through pretraining across heterogeneous datasets, domain adaptation under distribution shift, and large language model (LLM) enhancement for aligning sensing representations with language and reasoning over long-term health histories. I design model architectures and rigorous evaluation frameworks that turn heterogeneous health data into reliable health insights.
News
- 2026.09: I will be serving as a Workshop Chair of WellComp 2026 (Computing for Well-being), co-located with UbiComp/ISWC 2026 in Shanghai.
- 2026.01: Excited to share that I’m starting my Postdoc at Dartmouth College!
- 2025.05: Invited as the Web Chair of ACM MobiSys’26.
- 2024.10: I will be interning at Microsoft Research.
Selected Publications
EMNLP 2026BALMS: Benchmarking Agentic LLMs for Longitudinal Mental Health Sensing, Yu Yvonne Wu, Arvind Pillai, Yuliang Chen, Yuwei Zhang, Sudarshan Regmi, et al.ICLR 2026Beyond Hearing: Learning Task-agnostic ExG Representations from Earphones via Physiology-informed Tokenization, Hyungjun Yoon*, Seungjoo Lee*, Yu Yvonne Wu*, Xiaomeng Chen*, et al.NeurIPS 2024Towards open respiratory acoustic foundation models: Pretraining and benchmarking, Yuwei Zhang, Tong Xia, Jing Han , Yu Yvonne Wu, Georgios Rizos, et al.CIKM 2024StatioCL: Contrastive Learning for Time Series via Non-Stationary and Temporal Contrast, Yu Wu, Ting Dang, Dimitris Spathis, Hong Jia, Cecilia MascoloMLHC 2023Udama: Unsupervised domain adaptation through multi-discriminator adversarial training with noisy labels improves cardio-fitness prediction, Yu Wu, Dimitris Spathis, Hong Jia, Ignacio Perez-Pozuelo, Tomas I Gonzales, Soren Brage, Nicholas Wareham, Cecilia MascoloNPJ Digital MedicineLongitudinal cardio-respiratory fitness prediction through wearables in free-living environments, Dimitris Spathis, Ignacio Perez-Pozuelo, Tomas I Gonzales, Yu Wu, Soren Brage, Nicholas Wareham, Cecilia Mascolo
Invited Talks
- 2025.11: Towards Accessible, Effective, and Trustworthy Human-Centered Digital Healthcare System, University of Melbourne
- 2025.04: Generalizable Representation Learning for Time-Series in Mobile Health:Advancements and Applications, Singapore Management University
Experience
Postdoctoral Scholar
Dartmouth College
Research Intern
Microsoft Research Asia