♾Epistemic uncertainty challenges aging clock reliability in predicting rejuvenation effects

We finally published our paper about the role of epistemic uncertainty in aging clocks predictions (biological age predictions actually)!

Insights:
- Rejuvenation by cell reprogramming can not be reliably predicted with aging clocks;
- Ground zero of epigenetic age in embryogenesis can not be reliably predicted with aging clocks;
- Any clinically relevant aging clocks must be equipped with uncertainty estimation of its predictions;
- Covariate shift between training and testing datasets is a huge problem that you have to remember.

Thoughts:
It's been a while since the preprint came out. Since then, I've changed my attitude towards the aging clock a lot, which can be summarized in the following simple thought. The aging clock is not about machine learning and data science - the aging clock is about the methodology and philosophy of science. If you don't like the combination of words methodology and philosophy, you'll most likely produce another useless aging clock research, sorry.

📜paper
🖥code
💬community