Daniel Oberski
I hold a joint appointment at UMC Utrecht, Julius center and Utrecht University, department of Methodology and Statistics. Together with Charlotte Onland, I am currently the quartermaster data science for the strategic theme Circulatory Health.
My work focuses on applications of data science and machine learning, especially in the biomedical and social sciences, as well as in clinical practice, and quantifying and mitigating the effects of errors in (human) data by developing new statistical and machine learning techniques (see “human data science”). I work, among other topics, on applications to text mining of cardiovascular doctors’ notes, risk factors in dilated cardiomyopathy from EHR, anomaly detection in routine data from ventricular assist devices, estimation of diagnostic accuracy without a gold standard, and app and wearable sensor data analysis for COVID19 patients. Moreover, as lead data scientist of the UMCU digital health data analytics program (ADAM), I advise on a range of data science projects across UMCU. On the methodological side, my work tends to focus on measurement error and latent variable modeling, with links to high-dimensional data analysis, causal modeling, algorithmic fairness, differential privacy, and distributed learning.
Contact
Email: d.l.oberski@umcutrecht.nl
Phone: N/A
Keywords
Data science, machine learning, digital health, methodology, causal modeling, NLP, text mining, statistics, measurement error, latent variable modeling, latent class analysis, model-based clustering
Research topic
Methodology of data science; computational and applied statistics & machine learning.
Type of research
Fundamental data science.
Collaborations / partnerships
- UMC Utrecht: cardiology, Julius Center, Digital Health, reuma, psychiatrie
- IMI COVID RED
- UU M&S, informatica
- Exposome
- Utrecht open science programme.
Pure profile
N/A