Héctor Climente-González

Interpretable machine learning leverages proteomics to improve cardiovascular disease risk prediction and biomarker identification - Communications Medicine Climente-González, Oh et al. introduce an interpretable machine learning model that integrates plasma proteomics with clinical risk factors and employs an explainable boosting machine algorithm for risk prediction. Authors show that their model outperforms existing models in predicting risk for cardiovascular disease.

🚨 NEW RESEARCH! ⚕️ 🩺 🖥️

Climente-Gonzalez, Oh et al. develop an interpretable #ML model that integrates plasma proteomics with clinical risk factors to improve risk prediction for CVD 🫀 .

https://bit.ly/3ZUqyY8

@hclimente.eu @nkoell.bsky.social

I've spent the last few weeks digging into InstaDeep's Nucleotide Transformer DNA language model using @hf.co 🤗

If you're keen on seeing how LLM libraries can be applied to DNA, my latest post breaks it all down. Give it a read: hclimente.eu/blog/hf-tran...

Why I'm more worried about AI safety now than 6 months ago Exponentials are all you need

New post: things have been moving very fast in AI. But has safety caught up to capabilities? open.substack.com/pub/naix/p/w...

I have been using uv to manage my Python projects lately. Faster setups, better reproducibility, and it even trimmed down my stack.

Here's a short tutorial if you're curious: hclimente.eu/blog/python-...

I’m a Lead Scientist at Novo Nordisk in London, applying ML to genetics, epigenetics and real-world data. I’m particularly into sequence learning, explainable AI & graph-based methods.

Never too late for cool science! I just put together a post on SHAP values — a popular model-agnostic approach to explain model predictions. Would love to hear your thoughts or critiques!
#MachineLearning #ExplainableAI #XAI #SHAP

Even More Thoughts on ML Method Comparisons Introduction A few things motivated this post.   Some recent discussions about the virtues of LightGBM vs XGBoost Posts on TabPFN by Jon...

Good read on visualizing ML model performances.

In our updated TraitGym preprint (w/ @gonzalobenegas.bsky.social & Gökcen Eraslan), we evaluate Evo 2 on regulatory variants associated with human traits. We see marked performance gains with scale on Mendelian traits, although still a bit behind alignment-based methods.
doi.org/10.1101/2025...
1/n

RNA xkcd.com/3056

Part one of a collaboration with @3blue1brown.bsky.social on presenting the mathematics of the cosmic distance ladder in an accessible fashion.

A DNA language model based on multispecies alignment predicts the effects of genome-wide variants https://www.nature.com/articles/s41587-024-02511-w (read free: https://rdcu.be/d5oQZ) 🧬🖥️🧪 https://github.com/songlab-cal/gpn

Thanks for sharing, I’ll keep an eye out.

I would love to know more about the piece of work on caQTLs. Is it published?