SeizyML
🧠 What is SeizyML?
SeizyML is an open-source Python toolkit for semi-automated detection of electrographic seizures. It combines simple, interpretable machine learning models with manual validation resulting in high accuracy and sensitivity, reducing bias, saving time, and improving reproducibility.
⚙️ Key Features
- Fully open-source and built in Python.
- GUI for manual curation of detected seizures.
- Requires minimal training data to get started.
- Lightweight (Just a standard CPU and enough memory).
💡 Why I Built It
Despite many seizure-related studies, there was no accessible tool to help researchers automate seizure detection from raw LFP/EEG data. SeizyML fills that gap by offering a validated, easy-to-use pipeline designed to augment human expertise.
📊 Validation
We tested SeizyML on extensive EEG data from chronically epileptic mice. The Gaussian Naive Bayes model detected all seizures, had the lowest false detection rate, and required only a small training set.
🔧 Tech Stack
- Python 3.
- scikit-learn, NumPy, matplotlib.