Classifying haemodialysis sessions, on the basis of the evolution of specific clinical variables over time, allows the physician to identify patients that are being treated inefficiently, and that may need additional monitoring or corrective interventions. In this paper, we propose a deep learning approach to clinical time series classification, in the haemodialysis domain. Specifically, grounding on our previous experience in adopting convolutional neural networks on haemodialysis time series, we have defined an inception-based architecture, able to exploit kernels of different sizes in parallel. The proposed architecture has outperformed the results obtained by resorting both to a more standard convolutional neural network, and to the state of the art approach ROCKET, since we reached higher accuracy values, coupled with a good Matthews Correlation Coefficient.