Figure 3.

Structure of the Unet model for semantic segmentation of functional assessment. It is composed of four encoder blocks and four decoder blocks interconnected with a bridge in the central part where all the characteristics of the input signals are encoded. Each encoder/decoder block is composed of a series of 1D convolutional layers and a normalization (blue arrows). The outputs of these blocks (Sn and Pn) are interconnected with the next encoder block (red arrows) and with the analog decoder (gray arrows). The output of the model is the probability of each timestamp (64 input timestamps) of the activity of the functional test.