BioML, Uncertainty Estimation/XAI, Scientific Reproducibility @ FU Berlin
Joined October 2019
- Tweets221
- Following115
- Followers50
- Likes6.7K
Pascal Iversen retweeted
📊 Latest research in Bioinformatics Advances: "Leveraging uncertainty estimates for drug response prediction in cancer cell lines"
Find it here: doi.org/10.1093/bioadv/vbag2…
Authors include: @Iv_Pascal
Code is the blind spot of peer review, and many serious errors that may invalidate a paper's conclusions can stay hidden in scientific repositories. We study whether coding agents can help make code review feasible for science.
Check out our paper, where we apply agentic scientific code review to 100 NeurIPS 2025 papers (link below), and meet us at NeurIPS 2026!
We find that codebases are often missing, incomplete, or contain methodological problems.
We think that AI-auditing can make computational science more rigorous and reproducible!
Thanks to my co-first @ferbsx, and co-authors Katharina Baum and Bernhard Renard.
Quick paper overview video:
youtube.com/watch?v=QiP65fft…
The paper:
openreview.net/forum?id=xjCJ…
Try the beta version of our tool for free at auditowl.ai. We'd be super interested in your feedback!
ML-based cancer cell line drug response models are well-motivated, and significant research effort has gone into developing complex modeling approaches (over 100 papers in 2025).
The problem: under rigorous evaluation, we found none that actually works.
Most are published based on inflated metrics, break down when probed in realistic application scenarios, and are outperformed by simple baselines.
We built DrEval, a pipeline for unbiased evaluation of these models, to encourage more meaningful progress in the field.
Now published in Nature Communications:
nature.com/articles/s41467-0…
DrEval could also function as an unbiased reward signal for AI agent–based model development in biological ML. We are happy to chat about this if anyone is interested in details :)
A recent npj Digital Medicine paper reads as a complete LLM-slop. Inconsistent p values. Released code doesn't produce what the paper shows. Wrong metadata in the citations. How does this get through review? nature.com/articles/s41746-0…
Pascal Iversen retweeted
Simon Witzke presenting his poster @ECCBinfo! #ECCB2024
@witsyke and @Iv_Pascal recent work in explaining aleatoric uncertainty in regression can help increase transparency in digital health. – at Turku, Finland
ALT Simon Witzke happily standing in front of his poster.
Pascal Iversen retweeted
Lukas will present SimbaML 🦁 at the #ICLR2023 Tiny Papers Showcase Day (a #DEI initiative) tomorrow (05/05) at 11:15am CAT! Come by to learn more about connecting mechanistic models and ML with augmented data! w/ @JulianZabbarov @witsyke @Iv_Pascal
How can a century of scientific knowledge about system dynamics be harvested in the age of ML? Augment ML pipelines with ODE-generated synthetic data with SimbaML 🦁, our new Python tool accepted as #ICLR2023 Tiny Paper: arxiv.org/abs/2304.04000 @JulianZabbarov @witsyke @Iv_Pascal
Pascal Iversen retweeted
How can a century of scientific knowledge about system dynamics be harvested in the age of ML? Augment ML pipelines with ODE-generated synthetic data with SimbaML 🦁, our new Python tool accepted as #ICLR2023 Tiny Paper: arxiv.org/abs/2304.04000 @JulianZabbarov @witsyke @Iv_Pascal
Pascal Iversen retweeted
🥳Zwei HPIler für herausragende Abschlussarbeiten von der @unipotsdam ausgezeichnet: HPI-Masterabsolvent Hans Gawendowicz erhält den #Absolventenpreis 2022, HPI-Doktorand Pascal Iversen den Jacob-Jacobi-Preis. Wir gratulieren herzlich! ➡️Mehr Infos unter: bit.ly/3OJKXYn
ALT HPI-Absolvent Hand Gawendowicz mit Prof. Dieter Wagner (links) und Prof. Tobias Friedrich (rechts)
Pascal Iversen retweeted
