Research

My research focuses on causal inference, target trial emulation, and longitudinal clinical modeling using electronic health records and other real-world healthcare data. I build methodological frameworks that aim to make longitudinal clinical data more useful for real treatment decisions, while being explicit about support, bias, censoring, and the structure of routine care.

Featured research directions

Target Trial Emulation with EHR

Reconstructing clinically meaningful treatment comparisons from routine clinical records.

This work develops operational target trial frameworks for electronic health records, with emphasis on:

• treatment action reconstruction from routine EHR data
• decision-time aligned treatment evaluation
• support-aware restriction before comparing strategies
• clinically interpretable longitudinal treatment contrasts

Longitudinal Treatment Strategies in Routine Care

Defining treatment histories that better reflect how chronic care actually evolves over time.

This line of work studies longitudinal treatment strategies such as:

• add-on versus monotherapy strategies
• early versus delayed treatment intensification
• treatment persistence, gaps, restarts, and switching
• sustained exposure and lag-aware treatment effects

These projects aim to define treatment histories and comparisons that remain clinically meaningful when medication use changes over time.

Interpretable Disease Progression and Decision-Oriented Modeling

Building longitudinal models that support interpretation, uncertainty awareness, and decision relevance.

This work includes:

• interpretable disease trajectory modeling
• uncertainty-aware longitudinal prediction
• decision-facing representations of evolving patient state
• connections between predictive structure and causal questions

Application areas include Alzheimer’s disease and neurodegeneration, cancer outcomes and treatment safety, cardiovascular adverse events, chronic disease progression, and drug repurposing using longitudinal clinical data.