I build, validate, and monitor clinical AI.
At UCSF Health, I validate and monitor AI models such as Epic’s 30-day readmission model, develop LLM-based summarization of inpatient notes, and build real-time bed-availability prediction models and patient-flow simulations for hospital operations.
10+
Years in healthcare analytics and research
2–4hrs
Cut from emergency department wait times in a prior role
6
Peer-reviewed journal articles
130+
Research citations (OpenAlex)
01About
I’m a data scientist at UCSF Health, where I validate AI models for production use and monitor their performance after deployment. A background in industrial and systems engineering shapes how I approach hospital operations, from real-time bed-availability prediction to patient-flow simulation.
Before UCSF, I was a healthcare data scientist at the University of Maryland Medical System, where I applied simulation, forecasting, and quasi-experimental methods to emergency department wait times, COVID-19 hospitalizations, and patient satisfaction.
My doctoral research at Virginia Tech, which included a project funded by the Ohio Department of Medicaid, used system-dynamics and machine-learning models to study infant mortality and preterm birth. It produced first-authored articles on state-level variation in infant mortality and early signs of preterm birth, and I co-authored a study benchmarking states’ healthcare systems on birth outcomes.
02Selected work
Clinical AI at UCSF Health, forecasting and simulation for hospital operations, and population-health research.
At UCSF Health
Validating models for production use and tracking their performance after deployment, across projects including:
Using large language models to summarize inpatient notes, hospital course, and discharge instructions, and validating their output.
Real-time predictions of bed availability for UCSF Health’s Patient Capacity Management Center.
Patient-flow simulation that informs decisions such as MRI capacity expansion; previously, discrete-event models that cut ED wait times by 2–4 hours.
At the University of Maryland Medical System: quasi-experimental studies on improving patient satisfaction, and COVID-19 hospitalization and ICU forecasts that outperformed external benchmarks.
Research
Peer-reviewed research on infant mortality, preterm birth, and PTSD, including modeling for a project funded by the Ohio Department of Medicaid.
03Experience
AI and machine-learning tools for patient care and hospital operations.
Process improvement and predictive analytics.
Doctoral research on infant mortality and preterm birth.
04Research and publications
Machine learning, system dynamics, and operations research applied to preterm birth, infant mortality, PTSD, and healthcare-system performance.
Featured · 2022
Information, 13(7), 310 · doi:10.3390/info13070310
Expert Systems with Applications, 182, 115278 · doi:10.1016/j.eswa.2021.115278
American Journal of Perinatology, 36(12), 1271–1277 · doi:10.1055/s-0038-1675835
Systems Research and Behavioral Science, 36(4), 476–493 · doi:10.1002/sres.2563
System Dynamics Review, 33(1), 71–84 · doi:10.1002/sdr.1575
PLOS ONE, 11(10), e0161405 · doi:10.1371/journal.pone.0161405
Covered by Newsweek and Business Insider
System Modeling of a Health Issue: The Case of Preterm Birth in Ohio
Predicting Preterm Birth Using Machine Learning Techniques
Interstate Analysis of Variation in Infant Mortality, 2000–2014
John White Graduate Student Award
Scholarship, System Dynamics Summer School
Second place, FlexSim Simulation Competition
05Skills
The techniques and software behind my work, from model validation and monitoring to simulation and optimization.
06Contact
I welcome conversations about clinical AI, health-system analytics, research collaborations, and new opportunities. LinkedIn is the best way to reach me.