Challenge Details

Data and Protocol

The AI4PAIN dataset contains a total of 65 participants. The dataset will be partitioned into three sets for the challenge: training (41 participants), validation (12 participants), and testing (12 participants). Currently, the AI4PAIN dataset comprises two sets designated for training and validation purposes. For testing, new, unseen data will be utilised. The objective is to classify sample data into one of three categories: No Pain (NP), Low Pain (LP), and High Pain (HP). Labeled training and validation sets will be accessible early on (by 1 April, 2024), while the new, unlabeled test set will be provided in mid-May, 2024. 

There are no separate fNIRS-only or video-only challenges. Participants have the freedom to utilise either modality or both. They are also permitted to employ their own features and classification techniques. The labels of the testing set will be undisclosed, requiring participants to adhere to the definitions of training, validation, and testing sets. While papers may include results obtained from the training and validation sets, only outcomes from the testing set will contribute to the overall Grand Challenge results. 

Participants are encouraged to provide the labels for the test set to the organisers, facilitating the computation and sharing of classification results during the testing phase (18 May -  1 June). Throughout this phase, participants will have up to two chances to submit test labels and receive classification results. The culmination of this effort will be the presentation of final results at the AAP 2024 Workshop held during the ACII 2024 conference. Additionally, participants are expected to submit a paper outlining their methodologies and presenting their results in detail. 

To register your team, go to the Challenge Registration page.