This page provides specific FAQs for the 2021 Challenge. Please read the general Challenge FAQs for more general questions about the Challenges.
Yes, you can still participate. An accepted CinC abstract is required for prizes, rankings, and the opportunity to present your work at CinC, but you can still submit algorithms to the official phase without an accepted abstract.
A ‘wildcard’ entry is reserved for a high-scoring team who submitted an abstract that was not accepted or who were unable to submit an abstract by the deadline. Please read here for more details, including the deadline.
Due to the unique circumstances of 2020 and 2021, remote attendance is allowed for both CinC 2020 and 2021. Participants were still eligible for prizes if they attend remotely (as long as they satisfied the other criteria).
Yes, as long as no one from either team competes in a different team.
Yes, the philosophy of the Challenge is to encourage researchers to make their code free to use for research. We hope that companies will approach you to license the code, too! If you do not specify any license, then we will assume that the license is the BSD 3-Clause License.
Each database is labelled using a different ontology, or subset of terms in an ontology (or sometimes no ontology, i.e., just a free-text description). We needed to decide how to map these ontologies to a consistent set of labels. For example, we have the following four labels for ventricular ectopic beats:
|Premature ventricular complexes||164884008||PVC|
|Premature ventricular contractions||427172004||PVC|
|Ventricular ectopic beats||17338001||VEB|
|Ventricular premature beat||17338001||VPB|
We have chosen to retain the distinction between these in terms of SNOMED codes (but merged PVCs because we could really see no reason they had two separate codes), but the scored labels carry the same weight in scoring matrix, so mixing them up doesn’t cost you any points. You may then ask, ‘why not merge them all in the labelling’? That’s a question you have to answer for yourself! You are certainly welcome to do that - but you may not want to. You may note that only VPB indicates the temporal location of the beat relative to the preceding normal beat. This may, or may not, affect your algorithm, depending on how you write your code. You may or may not want it to affect your algorithm - the relative timing of beats certainly gives you information!
In general, we have tried to provide you with as much useful information as possible, without overwhelming you with a complete data dump.
No, the leaderboard contains scores on a subset of the hidden data during the unofficial and official phases of the Challenge. The final scores on the full test data are released after the conference for the “best” model selected by each team.
We are creating a large database of heterogeneous data with varying labels, some of which are wrong or incomplete. Leads can be inverted, noisy, and/or mislabeled. We have deliberately made no attempt to clean this up. The test data contains better labels, but it is not perfect either, and although it roughly correspond to the training data, it includes some deliberate differences.
No, we have the training, validation, and test data along with the evaluation code?
You will be able to choose which model you would like to have scored on the full test set. We will ask for teams to choose their “best” model shortly before the end of the official phase of the Challenge. If you do not choose a model, or if there is any ambiguity about your choice, then we will use the model with the highest score on the current subset of the test data.
Yes, most certainly. We encourage you to do this. You do not need to include your data in the code stack for training the algorithm, but you do need to include the pre-trained model in the code and provide code to retrain (continue training) on the training data we provide. You must also thoroughly document the content of the database you used to pre-train.
Yes, this is a required (and exciting) part of this year’s Challenge.
No, the training code is an important part of this year’s Challenge.
We run your training code on Google Cloud using 10 vCPUs, 65 GB RAM, and an optional NVIDIA T4 Tensor Core GPU with 16 GB VRAM. Your training code has a 48 hour time limit using the GPU and a 72 hours time limit without using a GPU.
We run your trained model on Google Cloud using 6 vCPUs, 39 GB RAM, and an optional NVIDIA T4 Tensor Core GPU with 16 GB VRAM. Your trained models have a 24 hour time limit on each of the validation and test sets.
No, please only submit your code to the submission system.
No, please only submit an entry after you have finished and tested your code.
No, please use the submission form to submit your entry through a repository.
No, not yet. If you change your code after submitting, then we may or may not run the updated version of your code. If you want to update your code but do not want us to run the updates (yet), then please make changes in a subdirectory or in another branch of your repository.
If you used Python for your entry, then test it in Docker.
No, only scored entries (submitted entries that receive a score) count against the total number of allowed entries.
For more general Challenge FAQs, please visit here.
Supported by the National Institute of Biomedical Imaging and Bioengineering (NIBIB) under NIH grant R01EB030362.