Genetic programming of conventional features to detect seizure precursors

Abstract

This paper presents an application of genetic programming (GP) to optimally select and fuse conventional features (C-features) for the detection of epileptic waveforms within intracranial electroencephalogram (IEEG) recordings that precede seizures, known as seizure-precursors. Evidence suggests that seizure-precursors may localize regions important to seizure generation on the IEEG and epilepsy treatment. However, current methods to detect epileptic precursors lack a sound approach to automatically select and combine C-features that best distinguish epileptic events from background, relying on visual review predominantly. This work suggests GP as an optimal alternative to create a single feature after evaluating the performance of a binary detector that uses: 1) genetically programmed features; 2) features selected via GP; 3) forward sequentially selected features; and 4) visually selected features. Results demonstrate that a detector with a genetically programmed feature outperforms the other three approaches, achieving over 78.5% positive predictive value, 83.5% sensitivity, and 93% specificity at the 95% level of confidence.

Novel approach for fetal heart rate classification introducing grammatical evolution

“…In future work, we will examine a fitness function that is not depended on the selection of the classification method and we will try to maximize the separability of the classes in feature space [55,56]. This approach can be combined with simpler classifiers reducing the potential overfitting problems that can occur when neural networks are involved.…”

A Filter Approach to Multiple Feature Construction for Symbolic Learning Classifiers Using Genetic Programming

“…It can be used in conjunction with a decision tree inducer to construct more discriminative features for decision nodes [55]. In the signal classification domain, GP has been used to provide synthetic artificial features for the k-nearest neighbor classifier [13], [56], [57] as well. Since every fitness evaluation involves training a classifier and then testing its performance, the search process is computationally very intensive.…”

A Comparative Study of Intracranial EEG Files Using Nonlinear Classification Methods

“…3,7,11,12,19,20,32 These investigations employ the use of multichannel trends, neural networks applications, the use of orthogonal transforms, such as the Walsh Transform, genetic programming, and all dwell in either time or frequency domains. The results are based on EEG recorded from the scalp that has lower signal-to-noise ratio compared to the EEG recorded from the cortex.…”