Nils Strodthoff

Prof for AI4Health @UniOldenburg; former
Head of Applied Machine Learning Group
@FraunhoferHHI @nstrodt at

[Nils Strodthoff @nstrodt]

📣I'm excited to share that I will be joining @UniOldenburg as a professor in the School of Medicine and Health Sciences building up a lab for Artificial Intelligence in Health.
https://t.co/Z4WVxz6MH0

[Nils Strodthoff @nstrodt]

Paper: https://t.co/VszEAKD7R5
Code repo and ECG-CPC weights: https://t.co/0ZhMC9bdCS

[Nils Strodthoff @nstrodt]

Our work on benchmarking foundation models for electrocardiography has been accepted at ICLR2026! We benchmarked 7 ECG FMs (proposed a highly efficient FM based on CPC ourselves) on 26 tasks across 12 datasets, looked into label efficiency and representational similarity

[Nils Strodthoff @nstrodt]

Representation Diversity: Models with similar accuracy learn distinct internal patterns, revealing multiple paths to effective ECG understanding.

[Accepted papers at TMLR @TmlrPub]

FeatInv: Spatially resolved mapping from feature space to input space using conditional diffusion...

Nils Neukirch, Johanna Vielhaben, Nils Strodthoff.

Action editor: Zhihui Zhu.

https://t.co/Ub9BMQejHo

#cnns #classifiers #vision

[Nils Strodthoff @nstrodt]

@imin_chiu @JACCJournals @David_Ouyang Amazing! Will datasets also be made available?

[Nils Strodthoff @nstrodt]

I have two vacant PhD positions in my group:
🏃 Machine Learning for Activity Trajectories https://t.co/lK3BTSGdfY (Deadline Oct 20)
🧪 Machine learning for Medical Diagnostics https://t.co/Z0YYUMm2C4 (Deadline Oct 31)
Looking forward to your applications!

[Nils Strodthoff @nstrodt]

Two papers are already available as preprints: ICD codes https://t.co/cIiGDcyD3q and lab values https://t.co/wGoYtdhl68 based on joint work with my PhD student Juan Miguel. Looking forward to interesting discussions at the conference.

[Nils Strodthoff @nstrodt]

Next week I will attend Computing in Cardiology presenting one talk (Open Science to Foster Progress in Automatic ECG analysis) and three posters on ICD diagnosis prediction from ECG features, lab value prediction from ECG features & uncertainty quantification for PPG data #CinC

[Nils Strodthoff @nstrodt]

Interested in general diagnostic (not necessarily cardiac) prediction tasks based on #ECG data? Check out our new Physionet dataset MIMIC-IV-ECG-Ext-ICD https://t.co/CGogzP4HDF which links MIMIC-IV ECGs to MIMIC-IV discharge diagnoses @tompollard @alistairewj

[Nils Strodthoff @nstrodt]

🎉Our unified prediction model for cardiac and non-cardiac conditions from a single ED #ECG has been published by #EHJDigital https://t.co/g6qNoSCxyr @BruiningNico @rafavidalperez @GerdHindricks @rbcasado Never understimate the diagnostic power of the ECG!

[Nils Strodthoff @nstrodt]

Code is available under https://t.co/LSpRogyBSr We will put out a Physionet dataset soon to make the dataset even more convenient to use. Thanks to my amazing PhD student Juan Miguel for the hard work and to Hjalmar Bouma from UMCG for the fun collaboration! 6/6

[Nils Strodthoff @nstrodt]

🧵I am excited to share our latest preprint on multimodal decision support in the emergency department. We propose a multimodal dataset (based on MIMIC-IV) and first baseline models for diagnoses and deterioration prediction. https://t.co/3S3gBzeKMj 1/6

[Nils Strodthoff @nstrodt]

Similarly, the model shows a strong performance across many different deterioration conditions (6 clinical deterioration measures, ICU admission and mortality prediction at different time horizons). 5/6

[Nils Strodthoff @nstrodt]

Our work on pixel flipping methodologies for #XAI methods got accepted at TMLR. Give our SRG measure a try the next time you benchmark your latest XAI method. Thanks @sbluecher @jvielhaben for the fun collaboration @XAI_Research @FraunhoferHHI @bifoldberlin

[Accepted papers at TMLR @TmlrPub]

Decoupling Pixel Flipping and Occlusion Strategy for Consistent XAI Benchmarks

Stefan Bluecher, Johanna Vielhaben, Nils Strodthoff.

Action editor: Wei Liu.

https://t.co/3SYlIwEJoN

#occlusion #features #feature

[Nils Strodthoff @nstrodt]

🎉Our unified prediction model for cardiac and non-cardiac conditions from a single ED #ECG has been published by #EHJDigital https://t.co/g6qNoSCxyr @BruiningNico @rafavidalperez @GerdHindricks @rbcasado Never understimate the diagnostic power of the ECG!