Fig. 3From: Opportunities and challenges of supervised machine learning for the classification of motor evoked potentials according to musclesMEP properties to illustrate the high variability of intrinsic MEP signal features. A MEP traces recorded during an entire surgery plotted on top of each other for 2 different patients and muscles (left: all APB MEPs of one patient, right: all TA MEPs of a second patient). Different colors indicate different MEP recordings. B Distribution of onset latencies and amplitude distributions of the APB muscle for all patients. C Dimensionality reduction of test data via principal component analysis (left) and feature extraction (right). For the PCA plot, the data is displayed along the 2 components of highest data variability, whereas on the right, 2 intuitive features were selected for the x- and y-axis. Each dot represents an MEP signalBack to article page