Research fellow Melbourne University, Victoria, Australia
Rationale: Seizures are a symptom of underlying brain disease and reliable seizure prediction will greatly improve the quality of life of people with epilepsy. Recording electrical activity in the brain using intracranial electroencephalography (iEEG) enables detection of brain wave patterns. Learning and analysing these patterns is useful for diagnosing brain conditions and estimating the likelihood of a seizure. Accurate seizure prediction can change epilepsy management by warning patients to move to a safe zone or effect interventions. Although many predictive machine-learning algorithms have been developed, standardised and benchmarked algorithms will improve the clinical and commercial viability of seizure-prediction devices. Methods: Using the iEEG data from the first-in-human long-term trial of a seizure prediction device, the ‘Melbourne University AES-Math Works-NIH Seizure Prediction Challenge’ globally crowdsourced predictive algorithms in a standardized framework but improvements in prediction performance are still needed. For ongoing development of these and new algorithms, Epilepsyecosystem.org hosted the contest iEEG data of three patients in the form of ten-minute windows without overlap containing preictal (duration of one hour) and interictal data for development of machine learning algorithms to distinguish between the two classes. The algorithms were ranked on the contest data based on AUC (area under ROC curve) performance that took into account the number of true and false positives. Via our Epilepsyecosystem.org website we have also launched an on-going independent evaluation of the top ranked algorithms and are currently recruiting participants and their algorithms. This valuation focuses on the full iEEG data record of the 15 patients from the long-term trial. We have independently evaluated top algorithms on the contest data and we will soon evaluate these methods on the full trial dataset. Evaluation on the full trial dataset will focus on clinically-relevant performance metrics (i.e., high sensitivity and low proportion of time in warnings) and, for developing battery-saving algorithms usable in wearable/implantable devices, computational efficiency (i.e., maximum RAM usage and average time taken for classification of ten-minute data windows). Results: The best algorithm to date on the contest data has scored an average AUC of 0.89. Other top algorithms are presented in Table 1. Conclusions: Current results suggest viable algorithms are being developed with the contest data and ongoing work through the independent evaluation on the full trial dataset should yield a new set of high-performing and efficient seizure prediction algorithms. This will help make seizure prediction both clinically and commercially viable and help alleviate the stress associated with seizures. Funding: Please list any funding that was received in support of this abstract.: NHMRC GNT1160815. Click here to view image/table
