We present our GLOBAL UPDATE of the first open benchmark of the epigenetic aging clocks to test their ability to predict biological age! Now it is equipped with the
Training Split suitable for training 1-st generation aging clocks on 7k+ blood samples of healthy people.What was added?
- 46 additional datasets (see table above) of healthy blood samples suitable for training 1-st generation aging clocks;
- everything is easy to download in a couple of minutes from our huggingface repo;
- our preprint was improved with a new figure A3 explaining the new Training split internals;
- some bugs were fixed and, as always, everything is accessible from our ComputAge repo;
- a notebook exemplifying the training procedure.
Useful links:
📜 paper-v2
🖥 code
🤗 dataset
💬 community
🤩 notebook