Hi there, As I understand all three sub-challenges are time series (longitudinal) questions. So, for every single SUBJECTED we will have multiple predictions in different times, but there is no TIME column in Submission Templates?! The other point is using cutoff dependent performances measures, i.e. Precision, recall etc, to evaluate performance of models. Why you do not use cutoff independent measures, e.g. area under precision-recall curve or if models could show good performance AUC? Using cutoff dependent measures may provide chance of cheating and fake better performance with focusing on improving 2 or even three cutoff dependent measures. Samad

Created by S J SAJA
Thanks, Solveig. - S
Given the original design of the challenge, participants were to submit one timepoint per phase, thus there was no need for multiple columns. Unfortunately, we do not have time to redesign the scoring system now. Performance measures for Subchallenge 1 and 2 are the average of the ranks of the area under the ROC (AUROC) and area under the PR (AUPR). I hope this clarifies things.

Submission template and performance measures page is loading…