Early prediction of developing spontaneous activity in cultured neuronal networks

David Cabrera-Garcia,Pablo De La Fuente,Joaquín M Villanueva-Balsera,M Teresa Fernández-Sánchez,Antonello Novelli,Davide Warm

doi:10.1038/s41598-021-99538-9

Abstract

Synchronization and bursting activity are intrinsic electrophysiological properties of in vivo and in vitro neural networks. During early development, cortical cultures exhibit a wide repertoire of synchronous bursting dynamics whose characterization may help to understand the parameters governing the transition from immature to mature networks. Here we used machine learning techniques to characterize and predict the developing spontaneous activity in mouse cortical neurons on microelectrode arrays (MEAs) during the first three weeks in vitro. Network activity at three stages of early development was defined by 18 electrophysiological features of spikes, bursts, synchrony, and connectivity. The variability of neuronal network activity during early development was investigated by applying k-means and self-organizing map (SOM) clustering analysis to features of bursts and synchrony. These electrophysiological features were predicted at the third week in vitro with high accuracy from those at earlier times using three machine learning models: Multivariate Adaptive Regression Splines, Support Vector Machines, and Random Forest. Our results indicate that initial patterns of electrical activity during the first week in vitro may already predetermine the final development of the neuronal network activity. The methodological approach used here may be applied to explore the biological mechanisms underlying the complex dynamics of spontaneous activity in developing neuronal cultures.

Highlights

While gold-standard methods are lacking, some methods provide more robust results
The spontaneous activity of cortical neurons in culture was recorded with microelectrode arrays (MEAs) (Fig. 1a, b) during the first three weeks in vitro
The values of the 18 electrophysiological features analyzed from MEA recordings were grouped in three DIV intervals (6–8, 9–12, and 13–18) which captured the changes occurring from the first to the third week in vitro (Fig. 2a, Table S2)

Summary

Introduction

While gold-standard methods are lacking, some methods provide more robust results. For example, Maximum Interval or logISI are highly consistent burst detectors among different dynamics of spontaneous a ctivity[18]. Integration of spiking, bursting, synchrony, and connectivity features creates a multidimensional profile of network activity which renders the analysis difficult to tackle with traditional statistical methods. To overcome these limitations, machine learning methods arise as an alternative approach for extracting information from large datasets of neural r ecordings[24], as well as for predicting v ariables[25]. We successfully used three machine learning models (MARS, SVM, and Random Forest) to predict the levels of electrophysiological features at the third week in vitro, suggesting that the development of network activity is determined by the early electrical activity of neuronal networks. The methodology presented here may help to identify the biological factors determining the maturation of in vitro neural networks

Methods

Results

Conclusion