Atrial fibrillation detection method based on electrocardiogram compressed sensing transposition projection signal

The invention provides an atrial fibrillation detection method based on an electrocardiogram compressed sensing transposition projection signal. According to the technical scheme, firstly, random projection is conducted on electrocardiogram signals through a deterministic observation matrix to complete compression, then transposition projection operation is conducted on compressed signals through transposition of the deterministic observation matrix at the cloud end to obtain approximate signals, and similarity verification is conducted on the approximate signals and original signals; and finally, the approximate signals are input into a transposition projection CNN (TP-CNN) to carry out atrial fibrillation detection. According to the atrial fibrillation screening method combining compressed sensing and a residual neural network, atrial fibrillation detection is directly carried out on the transposed projection signals of compressed electrocardiogram, and compared with the most advanced method, the result of