Hongzhou Luan

Hongzhou Luan

PhD Candidate, UC Berkeley – UCSF

I work on maximizing AI's impact on healthcare for the underprivileged majority. A key barrier to clinical AI deployment is the statistically challenging, yet informative not-at-random missingness in healthcare data: patterns often discarded but containing valuable signals about patient behavior and access. My research develops methods to both appropriately mitigate missingness and extract insights from the missingness itself. More broadly, I work to ensure people benefit from safe AI—from bias analysis in clinical algorithms to AI governance and policy.

Education

PhD Computational Precision Health, UC Berkeley – UCSF
2023 –
Advisor: Prof. Ida Sim · Funding: JupyterHealth Fellowship
MEng Engineering Science, University of Oxford
2018 – 2022
Jardine Scholar (full scholarship)

Publications

Anna Zink, Hongzhou Luan, Irene Y. Chen
Nature Health 2025 (In Press)
Hugo Campos Jr, Daniel Wolfe, Hongzhou Luan, Ida Sim
Journal of Participatory Medicine 2025

Current Projects

Wearable Data Missingness & Acute Event Prediction 2025 –
with Prof. Jessilyn Dunn (Duke) & Prof. Ida Sim

Analyzing 50,000+ participants with Fitbit data linked to EHR in the All of Us dataset to investigate whether shifts in wearable data missingness patterns precede acute cardiovascular events. Developing early warning methods that leverage routinely collected data without requiring continuous wear compliance.

Missingness-Avoidant Neural Networks 2025 –
with Prof. Fredrik Johansson (Chalmers)

Designing attention-based architectures that make predictions using only observed features, eliminating reliance on imputation—which assumes missing-at-random data and lacks clinical interpretability. Enabling models clinicians can trust by grounding predictions in actually measured values.

AI Scribes Evaluation in Clinical Practice 2025 –
with Prof. Ahmed Alaa, Prof. Ida Sim & Prof. Paul Tang (Stanford)

Leading the first systematic evaluation of AI medical scribes at UCSF, analyzing 400 clinical encounters. Identifying AI errors including hallucinations, omissions, and unlicensed statements, and determining their potential to cause patient harm.

AI Policy & Legislative Drafting 2025 –
UC Berkeley CITRIS Policy Lab, CITED

Drafting AI governance legislation for the California Initiative for Technology and Democracy (CITED) to be presented to California bill authors. Conducting policy transfer analysis examining EU AI regulatory frameworks for California adoption.

Dermatology Triage for Skin Cancer Screening 2024 –
with Prof. Katrina Abuabara & Prof. Meghan Shan

Developing a novel triage algorithm using 58,000+ patient-uploaded facial images—the first extraction of such data from UCSF. Addressing a critical access gap: Medicaid patients wait over twice as long for dermatology appointments. Independently filed the successful IRB for this project.

Awards

Jardine Scholar 2018 – 2022

Full scholarship for four years at Oxford with annual stipend of £10,300. Awarded to exceptional students from Asia demonstrating academic excellence and leadership potential.