Key result
The multivariate 1-dimensional Convolutional Neural Network and Multilayer Perceptron achieved the highest accuracies (99.26% and 96.82%, respectively) for detecting hERG channel blockade.
Why the study?
Accurate cardiac safety assessment requires determining whether a drug alters cardiomyocyte electrical activity by disrupting hERG channels using extracellular field potentials.
MLP and multivariate 1D-CNN architectures provide promising accuracy for automated detection of hERG channel blockade in hiPSC-CMs, aiding early preclinical cardiac safety screening.
This work will present a comparison of several Artificial Neural Network methods for a classification problem related to cardiac safety assessment. Given the extracellular field potential recorded by means of micro-electrode arrays, the aim is to determine whether a given chemical drug is altering the electrical activity of cardiomyocytes by disrupting the normal behavior of the hERG channels. To do so, this work has considered four different Neural Network methods and compared them in terms of accuracy and computational costs. The conclusion is that, among the tested architectures, the Multilayer Perceptron (MLP) and multivariate 1-dimensional Convolutional Neural Network (1D-CNN) give the most promising results.
No takes yet. Share an insight, caveat, or question.
Liu et al. (2022) studied hERG channel blockade (cardiotoxicity) (n=57,831). Artificial Neural Network methods (MLP, 1D-CNN, 2D-CNN) was evaluated on Classification accuracy for detecting hERG potassium channel blockers. The multivariate 1-dimensional Convolutional Neural Network and Multilayer Perceptron achieved the highest accuracies (99.26% and 96.82%, respectively) for detecting hERG channel blockade.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: