Standardized database development for EEG epileptiform transient detection: EEGnet scoring system and machine learning analysis

► We describe our new web-based software for collecting expert opinion on paroxysmal activity in routine scalp EEG. ► We report that inter-rater correlation among our groups of 11 board-certified EEG scorers was only moderate. ► Our machine learning analysis suggests that our EEG database needs to be larger than its current size to adequately represent the variability of waveform morphologies in EEG. ► Our artificial neural network machine learning classifiers performed better than our Bayesian classifiers and the wavelet features were the most useful.