Fig. 1: Simulation pipeline.
From: A myoelectric digital twin for fast and realistic modelling in deep learning
a General strategy of using Myoelectric Digital Twin to train an artificial intelligence (AI), that then can be used to process real surface electromyography (sEMG) signals in real-world applications. b Schematic representation of the software’s input, main elements of the simulation pipeline and output. User can define a large set of simulation parameters describing anatomical and physiological properties of the tissues, geometry and montage of electrodes, fibres geometry and their physiological properties, motor unit (MU) distribution and their recruitment model, activation of individual muscles, etc. The main components of the simulation pipeline are described in details in ‘Methods’ section. For a given set of the parameters, the software outputs the resulting simulated EMG data as well as all the metadata of the simulation (e.g. individual motor unit action potentials (MUAPs)).
