About me
I am an AI researcher working at the intersection of machine learning, causal inference, and healthcare. My work asks how we can move from predicting patient outcomes to understanding which interventions could improve them.
I hold a PhD in Data and Decisions Sciences from the Technion, advised by Prof. Uri Shalit and Prof. Danny Eytan. My research spans heterogeneous treatment-effect estimation, survival analysis, and hybrid mechanistic models that combine scientific knowledge with learned dynamics.
Selected research
- Hybrid models for intervention outcomes — Combining mechanistic models with neural networks to predict responses to time-dependent interventions, including under distribution shift. Preprint, 2026.
- Personalized treatment effects with censored outcomes — MISTR estimates how treatment benefits vary across patients when follow-up is incomplete. ICML, 2025.
- Survival analysis and PyDTS — Efficient methods and open-source tools for discrete-time outcomes with competing events, with an application to ICU length of stay. Biometrics and JOSS, 2025.
- COVID-19 response — Modeling hospital demand and mortality, alongside national symptom surveillance and vaccine-impact research. Nature Communications and Nature Medicine, 2021.
- Fracture detection in CT — Deep learning for identifying vertebral compression fractures in routine imaging. MICCAI, 2020.
Background
At the Weizmann Institute of Science’s Segal Lab, I developed models and analyses to support the COVID-19 response. Earlier, at Zebra Medical Vision, I worked on deep-learning systems for medical imaging. I also served as head lecturer for Machine Learning 1 at the Technion.
I am interested in translating research into practical tools for personalized medicine, working with electronic health records, longitudinal time series, survival data, and medical images.