Seizure-prediction

Seizure prediction contains innovative methods using adaptive filtering for detecting epileptic seizures based on energy signals. The paper was part of the final project submitted for the course, ECE 251B on Digital Signal Processing II, at UC San Diego during Spring 2017.

The curated and trained data is available at x_dir_new.mat, which is used by the predicting functions

Types of results

• Use kf_predict.m for prediction using kalman filtering
• Use myparticle.m for prediction using particle filtering
• Use rls_predict.m for prediction using RLS filtering

The description of the code and the results are available in the attached paper - "Seizure Prediction using Serial-Parallel Block Concatenated Adaptive Filters"

The data is downloaded from 2 sources :

1. Sleep activity eye-tracking dataset (avaiable in data folder)
2. The Bern-Barcelona EEG database, and it is huge and is not included in the repo.

The Test folder contains the sleep activity dataset which is used as the test data.

The file Data_N_Ind2261.txt is provided as an example of the training data used.