Addressing the current practice where most apnea detection systems rely on time-frequency features extracted from respiratory signal samples for classification, this paper proposes a 1D-Res&SENet classification model. This approach takes complete respiratory signal waveforms as input, extracts features through a one-dimensional convolutional neural network, and incorporates a residual network structure to mitigate gradient vanishing and grid degradation. Furthermore, recognising the varying importance of features across channels, it introduces an SE attention mechanism to identify and strengthen associative information between feature channels, thereby enhancing the accuracy of apnoea detection. Experimental results show that after adding the residual network and SENet module, the accuracy, recall and specificity of the model are increased by 2.0%, 4.9% and 1.7%, respectively.