Dear BoA Team,
I am new to this challenge and might have missed something, but I can't find very specific information about the Phase 2 goals.
What exactly do we have to predict for the samples? Days till death? In the submission tutorial the information is only about the age prediction (so Phase 1 i think).
Thanks in advance!
Created by Korbinian Traeuble KorbiT Ah I see. Thank you very much for the information!
I did not take part in the first phase so I did not receive any email. Found the email:
For Phase 2 of the challenge:
? Participants will submit a file with a ?biological age? between 20 and 100 for each sample representing the predicted mortality risk based on the molecular measurement for that sample.
? Judging will then be done via Cox proportional-hazards model as implemented in the biolearn library (https://github.com/bio-learn/biolearn/blob/0361a44945273a5edb610118a064d4cd54097b68/biolearn/mortality.py#L35).
? Submissions will be scored based on their hazard ratio relative to chronological age (higher being better) though only submissions with a p-value < 0.05 will qualify for scoring. NOT an official answer, just until you get that:
We have to predict "biological age". A Float number from 20 to 100 like before, but this time it is (very roughly) judged by how well it predicts mortality, not by how well it approximates actual age like in phase 1.
There was an email about the exact judging criteria, not sure where it is on the synapse page.