Utilizing Machine Learning Algorithms to Predict Subject Genetic Mutation Class from In Silico Models of Neuronal Networks

Epilepsy affects an estimated 50 million patients across the globe making it the fourth most common neurological disorder. Up to 40% of patients have uncontrolled seizures, yet incur approximately 80% of the cost. Anti-epileptic drugs commonly result in resistance and reversion to uncontrolled drug resistant epilepsy and are commonly associated with significant adverse effects. This has led to a trial-and-error system in which physicians spend months to years attempting to identify the optimal therapeutic approach. The primary reason for this is that the diagnostic system for patients with epilepsy commonly has minimal predictive capacity for treatment efficacy and tolerability because it fails to account for the multifactorial nature of epilepsy and the effects of unique patient biology. Therefore, the investigators hypothesize that by increasing the diagnostic precision of patients with epilepsy by characterizing electrophysiology using a novel approach, better initial predictions for optimal therapy can be made. Several studies have shown that genomic data alone can be predictive of effective therapeutic approaches in some patients. Thus, in order to assess the predictive power of electrophysiological data in silico , a multitude of machine learning strategies were implemented to attempt to predict a subject’s genetically defined class in an in silico model using brief electrophysiological recordings obtained from microelectrode arrays. This study found that various machine learning algorithms excel in using electrophysiological data from isogenic neurons to accurately predict genetic class. This indicates it is possible for machine learning algorithms to use electrophysiological data to predict clinically valuable metrics such as optimal therapeutic approach.