Trustworthy AI and Uncertainty
Principled methods for uncertainty quantification, robustness, interpretability, transparency, and accountable AI systems.
University of California, Santa Barbara
Center for Statistical Foundations of AI
Advancing the statistical science behind reliable, interpretable, uncertainty-aware, personalized, and trustworthy artificial intelligence.
NIH/NCI R01 on wearable multimodal health AI
NSF awards supporting heterogeneous data integration and AI policy learning
CSFAI Annual Conference: Statistics Foundations of AI
JASA, Annals of Statistics, JMLR, Biometrika, JRSSB, and AOAS publications
Mission
Building statistical foundations for trustworthy AI.
The Center for Statistical Foundations of AI serves as an interdisciplinary hub for developing validated, interpretable, transparent, accountable, and personalized AI. CSFAI brings together statistics, probability, mathematics, computer science, engineering, health sciences, and domain sciences to solve fundamental AI challenges driven by real-world applications.
Research
Core research themes
Trustworthy AI
Uncertainty quantification, robustness, interpretability, transparency, accountability, and responsible AI.
Foundation Models
LLM and NLP methodology, representation learning, retrieval learning, personalization, and agentic AI.
Precision Health
Mobile health, wearable devices, digital phenotyping, dynamic treatment regimes, and personalized interventions.
Heterogeneous Data
Data integration, distribution shift, block-wise missing data, optimal transport, and multimodal learning.
Causal AI
Causal discovery, mediation pathways, reinforcement learning, off-policy evaluation, and individualized decision-making.
Scientific AI
AI for neuroscience, climate, imaging, spatial-temporal data, and biological discovery.
Featured Research
Flagship research directions
CSFAI connects statistical theory, machine learning methodology, and scientific applications through a set of flagship research programs.
Heterogeneous Data Integration
Representation retrieval learning, optimal transport, distribution shift, posterior drift, and block-wise missing data.
Reinforcement Learning for Precision Health
Individualized decision-making, off-policy evaluation, dynamic treatment regimes, and resource-aware interventions.
Foundation Models and Personalized LLMs
Statistical foundations for personalized language models, retrieval, fine-tuning, and user-aligned AI systems.
Wearable and Mobile Health AI
Learning individualized signals from Oura, Fitbit, and multimodal longitudinal health data to support real-time insight.
Neuroscience, Aging, and Scientific AI
Statistical learning for memory, Alzheimer’s disease, brain dynamics, climate, imaging, and scientific discovery.
Projects
Main projects
Representation Retrieval Learning
General frameworks for heterogeneous data integration across sources, populations, and modalities.
Personalized and Trustworthy LLMs
Statistical methods for personalized language models, retrieval, and user-aligned AI systems.
Wearable and Mobile Health AI
Learning individualized health patterns from wearable sensors and intensive longitudinal data.
Reinforcement Learning for Precision Health
Policy learning, off-policy evaluation, and dynamic treatment regimes for heterogeneous populations.
Differential Privacy and Digital Twins
Privacy-preserving synthetic data and digital twin methods for sensitive biomedical and social data.
AI for Neuroscience and Aging
Data integration and statistical learning tools for memory, Alzheimer’s disease, and brain dynamics.
Funding
Current Federal Funding
SCH: Individualized Learning and Prediction for Heterogeneous Multimodal Data from Wearable Devices
Statistical Training to Enhance the Excellence of Research in Biomedical Sciences
Collaborative Research: Causal Discovery and Individualized Policy Optimization for Human Text Data
Data Integration for Heterogeneous Data: A General Framework for Distribution Shift, Posterior Drift and Block Missing Data
Epigenomic Predictors of PTSD and Traumatic Stress in an African American Cohort
People
CSFAI community
Annie Qu
Professor, Department of Statistics & Applied Probability, UC Santa Barbara
Founding Director, Center for Statistical Foundations of AI
Annie Qu is Professor of Statistics and Applied Probability at the University of California, Santa Barbara, and Founding Director of the Center for Statistical Foundations of AI (CSFAI). Her research develops statistical foundations for artificial intelligence, with interests spanning statistical learning, heterogeneous data integration, reinforcement learning, causal inference, trustworthy AI, uncertainty quantification, precision health, and multimodal data analysis. She is an ASA Fellow, IMS Fellow, AAAS Fellow, and an elected member of the International Statistical Institute (ISI). She received the IMS Medallion Award (2024), the IMS Harry Carver Medal (2025), and the ICSA Distinguished Achievement Award (2026) in recognition of her contributions to statistics, machine learning, and artificial intelligence. She currently serves as Co-Editor of the Journal of the American Statistical Association (Theory and Methods). Through CSFAI, she leads interdisciplinary research that advances the statistical foundations of trustworthy AI while fostering collaborations across statistics, computer science, engineering, health sciences, and industry.
Statistics & Applied Probability
Computer Science
Molecular, Cellular & Developmental Biology
Environmental Science & Management
Ecology, Evolution & Marine Biology
Psychological & Brain Sciences
Quantum Computing
Xuan BiUniversity of Minnesota
Xiaowu DaiUCLA
Yujia DengZoom
Louis EhwerhemuephaCHOC Research Institute
Haoda FuAmgen Inc.
Yuqing GuoUCI
Lei LiuWashington University in St. Louis
Regina LiuRutgers University
Rui MiaoUT Dallas
Qing NieUCI
Babak ShahbabaUCI
Weining ShenUCI
Xiaotong ShenUniversity of Minnesota
Padhraic SmythUCI
David TalbyJohn Snow Labs
Xiwei TangUT Dallas
Monica UddinUniversity of South Florida
Derek WildmanUniversity of South Florida
Qi XuUniversity of Minnesota
Fei XuePurdue University
Lan XueOregon State University
Henry YeGoogle
Yubai YuanPenn State
Jiuchen ZhangUniversity of Michigan
Jin ZhouUCLA
Ruoqing ZhuUIUC
Postdoctoral Fellows
- Hanjia Gao
- Cora Ziyi Liang
Current UCSB Ph.D. Students
- Cadence Pinkerton
- Junchao Han
- Yuting Ma
- Chunyin Lei
- Kathy Xu
- Monte Davityan
- Hannah Nelson
Current UCI Ph.D. Students (Co-mentored)
- Spencer Hilligoss
- Xiaoke Cao
- Kuancheng Ye
- Yuanpeng Li
Partnerships
Industry and clinical partnerships
CHOC Research Institute
John Snow Labs
Amgen
CSFAI works with clinical and industry partners to translate statistical AI foundations into high-impact systems for health, science, and responsible technology.
Publications
Recent publication highlights
Selected recent work connected to CSFAI themes in statistical learning, AI, health, and heterogeneous data.
Representation retrieval learning for heterogeneous data integration
A general framework for heterogeneous data integration.
Reinforcement learning of individual optimal policy for heterogeneous data
Reinforcement learning for individualized decision-making.
Heterogeneous effects of physical activity on physiological stress during pregnancy
Wearable AI for individualized stress monitoring.
Differential private data release for mixed-type data
Privacy-preserving statistical learning for sensitive heterogeneous data.
Individualized dynamic latent factor model for multi-resolution data
Dynamic latent modeling for intensive longitudinal health data.
Optimal Individualized Treatment Rule for Combination Treatments Under Budget Constraints
Personalized combination treatment under practical constraints.
Stage-aware Learning for Dynamic Treatments
Adaptive learning for multi-stage treatment decisions.
News
News and Highlights
2026
CSFAI Inaugural Conference
The inaugural Statistics Foundations of AI Conference brings together leading researchers in statistics, machine learning, and artificial intelligence at UC Santa Barbara.
Visit Conference Website2026 ICSA Distinguished Achievement Award
Annie Qu received the International Chinese Statistical Association Distinguished Achievement Award for outstanding contributions to statistical methodology and the foundations of data science, including pioneering work on heterogeneous data integration, representation learning, and individualized decision-making; for advancing the interface between statistics, machine learning, and artificial intelligence through impactful applications in mobile health and biomedical sciences; and for distinguished leadership in shaping the global statistical community.
Award Citation2025
Smart Rings and Statistical Timing Study Unlock Personalized Exercise Advice for Pregnant Women
A UCSB research team led by Annie Qu developed statistical learning methods that combine wearable-device data with individualized modeling to determine when physical activity is most effective for reducing physiological stress during pregnancy, illustrating how AI and statistical methodology can deliver personalized health recommendations.
Read the UCSB News Story2024
IMS Medallion Lecture: Data Integration for Heterogeneous Data
Annie Qu delivered the 2024 IMS Medallion Lecture, Data Integration for Heterogeneous Data. The lecture presents statistical foundations for heterogeneous data integration, representation learning, uncertainty quantification, and individualized decision-making, which are central to the research vision behind CSFAI.
Watch the LectureEvents
CSFAI Annual Conference
The annual Statistics Foundations of AI conference brings together leaders in statistics, machine learning, AI, health, and scientific discovery.
Contact
Center for Statistical Foundations of AI
Department of Statistics & Applied Probability, University of California, Santa Barbara
5512 South Hall, UCSB, Santa Barbara, CA 93106
https://qu.pstat.ucsb.edu