Atrial fibrillation detection method and device based on U-Net and LSTM network

The invention discloses an atrial fibrillation detection method and device based on U-Net and LSTM networks, and the method specifically comprises the steps: inputting obtained to-be-detected electrocardiogram data into a trained U-Net-LSTM network, and outputting a predicted electrocardiogram signal segmentation result with the same length as the input data; dividing the predicted electrocardiosignal segmentation result according to the heart beat position, counting the number of data points segmented into atrial fibrillation in each heart beat, and judging whether each heart beat is an atrial fibrillation signal or not according to the number of data points segmented into atrial fibrillation in each heart beat. According to the method, the problem of low accuracy of mainstream CNN and other models can be overcome, the backbone network is constructed by using the U-Net-LSTM framework, the recognition precision can be improved to a greater extent, and the actual recognition requirement is met. The LSTM networ