Hi all,
I had a few questions regarding sub-challenge 1.
1. When will the test data be released? (if they are not out yet)
2. From yesterday's webinar I had the impression that the demographics training table will not include columns such as (profesional-diagnosis, age, sex, etc.) and the feature extraction should only focus on the accelerometer, device motion and pedometer data. Am I missing something?
3. In the walking activity training table the medTimepoint column could be used to infer if the entry refers to a PD patient or control. I assume such feature derived from this column will not be acceptable?
Thanks!
Ioannis
Created by Ioannis Papavasileiou johnppv The medTimepoint will not be available in the test data and should not be used to build features - medTimepoint contains the information about case/control status. We will also release a supplemental training data as was discussed in the webinar 9/6. This training dataset contains more data from the same individuals in the released training data. You can use the demographics for the training data, however you will not have these data for the test data.
Sorry, I clearly haven't had my morning coffee, I meant to say you will not be provided medTimepoint for the _test_ data. Thanks for the response.
So we can use the demographics training (syn10146552) to build our own models and tune our features.
Also, for the walking activity table (syn10146553) I see in the schema that there is a medTimepoint column. Is there going to be another training set released? Apologies, we will release the test data tomorrow or Monday at the latest. @johnppv-
1. We will provide the test data on the September 15th with the launch of the scoring queues.
2. That is correct. You will not be provided demographics info on the test set.
3. You will not be provided medTimepoint in the training data, because it can be used to infer PD status.
Solly