CardioAge Challenge

Technical Submission Procedure

• Technical submissions must be sent by email to cardioage_icbhi2026@dei.uc.pt by the registered corresponding team member.
• Email subject format: CardioAge Technical Submission - TeamName - Submission N.
• Submission filename format: team_name_submission_N.zip, where N is 1, 2, or 3.
• The submission deadline is 30 August 2026, 23:59 AoE (Anywhere on Earth, UTC-12).
• The organizers will confirm receipt and technical validity by email.
• Invalid submissions count toward the maximum of 3 technical submissions unless the invalidity is caused by a file-transfer issue confirmed by the organizers.

Inference Interface

• The committee will execute the submitted inference.py entry point in the isolated evaluation environment.
• The exact hidden-test input path and command-line arguments will follow the interface documented in participant_package.zip.
• The solution must generate one chronological-age prediction per hidden-test subject.
• The expected prediction output format will follow the template included in participant_package.zip.
• Submissions are technically invalid if they fail to run, omit required files, produce malformed predictions, require internet access, or attempt to use prohibited data or resources.

Technical Submission Package

team_name_submission_N.zip
├── inference.py
├── requirements.txt
├── README.md
├── model/
│   └── trained model files or weights required for inference
└── src/
    └── optional supporting source code

• inference.py: executable entry point used by the committee to generate predictions on the hidden test set.
• requirements.txt: complete declaration of Python dependencies required for execution.
• README.md: technical documentation, model description, execution instructions and required compliance declarations.
• model/: all trained artifacts needed for inference.
• src/: optional auxiliary code required by inference.py.

The submitted package must be self-contained for inference after installation of the declared Python dependencies. No trained weights, external data or additional model resources may be retrieved during official inference.

Requirements and Individual Environments

Each technical submission must include a requirements.txt file specifying all Python dependencies required to execute the proposed solution. The organizing committee will create an isolated Conda environment for each team/submission and install the declared dependencies before executing official inference.

Participants may use Python libraries and machine learning frameworks required for their implementation, including, for example, classical machine learning libraries or neural network frameworks such as PyTorch or TensorFlow, provided that all dependencies are declared in requirements.txt, all trained model files required for inference are included, the solution runs successfully in the official controlled environment, no models, datasets or external resources are downloaded during official inference, and the solution does not depend on remote APIs or external services. Data augmentation or synthetic data may be used during model development and training, but the submitted package must contain only inference-ready trained artifacts or weights and source code required for prediction.