Welcome to the SeFEF Community!

[anascacais]

Hello everyone,

Thank you for your interest in SeFEF – the Seizure Forecasting Evaluation Framework.

SeFEF was developed to address a key challenge in seizure forecasting research: the lack of standardized development and validation protocols, which makes it difficult to compare algorithms and hinders translation into clinical practice.

The framework automates several steps in the algorithm development pipeline, including:

• Data handling: seizure labeling and cross-validation splitting
• Post-processing: transforming forecasts into meaningful outputs
• Evaluation: computation of deterministic and probabilistic performance metrics
• Reporting: automated forecasting plots, model cards, and reliability diagrams

🎯 Why this discussion space?

This is a place for the community to:

• Share experiences using SeFEF (datasets, algorithm types, use cases)
• Ask questions and troubleshoot
• Suggest new features or improvements
• Propose extensions (e.g., new metrics, new data handling modules)
• Discuss the broader goal of building a unified methodology for seizure forecasting

🚀 How to contribute

• Tried SeFEF? Please let us know what worked and what didn’t.
• Have ideas? Open a new Discussion to suggest features or use cases.
• Found a bug? Report it in Issues.

🙌 Next steps

We are currently running first-user experiments and would love feedback from the community. At this stage, your input will directly shape how SeFEF evolves.

👉 You can also share feedback anonymously via our Feedback Form.

Together, we can work towards a robust and standardized path for seizure forecasting research.

Looking forward to your thoughts and contributions,
— The SeFEF team