I am interested in developing techniques to extract information and reveal the intrinsic behavior of time series, signals, and dynamical systems. My research centers on large-scale models capable of performing such tasks, with a particular emphasis on robustness and generalization. I am especially motivated by the new possibilities these advances open for medicine.
My work spans from fundamental advances in learning theory to the development of practical, clinically deployable systems. Together with an interdisciplinary team, I have developed large-scale machine learning models to process biomedical signals for improved diagnosis and risk stratification. To support this work, we have curated and maintained some of the largest ECG datasets available (CODE and CODE-II), comprising millions of patients, and developed several state-of-the-art models for automatic ECG analysis.
I have also contributed to the broader research infrastructure underlying these efforts, as a core contributor to open-source scientific libraries such as SciPy, and by openly releasing multiple ECG models and datasets that have now been downloaded tens of thousands of times by the research community.
Group Members
Postdoc: Elis Stefansson.
Postdoc: Petrus Abreu.
Ph.D. (main supervisor): Jiawei Li.
Ph.D. (main supervisor): David Väviggreen.
Ph.D. (co-supervisor): Bror Hultberg.
MSc (supervisor): Sabereh Hassanyazdi.
MSc (supervisor): André Ramos Ekengren.