I'm a bit confused about the new requirement: "Therefore, the code submitted to Sub-challenge 2 should be able to run for subjects with only the latest exam (in the absence of a metadata table, such as is the case in Sub-challenge 1)". Can you be more specific? Shall I give only the image crosswalk TSV file to my SC2 inference script to pretend there is no meta info and longitudinal images? I don't want my submission to SC2 to become invalid because of a simple change like this. I need to test my code thoroughly. Please offer more details. Thanks!

Created by Li Shen thefaculty
Hi David, The ExamIndex column will be set to 1, so you shouldn't change anything in your algorithm with respect to how you run it in SC2. The test for the surrogate dataset will be simply to assign to each subject of the test set a value of each field in the metadata randomly sampled from the rest of the population in the test set. The results (AUC or pAUC) in the actual data has to be statistically better than in the surrogate, using the tie breaking rules described in the Challenge Description. Let me know if you need further clarifications. Gustavo
Hi, Could you please clarify on how to test for the new requirement? When you are running the code on the surrogate dataset, does the column ExamIndex exist? If so, is it set to all ones, or does it contain the number of the last examinations for each patient? Without this information it is not clear how to implement a robust fallback method. Is it furthermore possible to test the new requirement, i.e. is the test for the surrogate dataset implemented in the fast lanes? Best, David
Dear Li, I have forwarded your question to other challenge organizers. Thanks in advance for your patience. Best, Thomas

How can we test the new SC2 requirement? page is loading…