🤖 AI Summary
To address the limited interpretability and clinical applicability of deep learning models in sleep staging, this paper proposes an interpretable cross-modal Transformer architecture. It is the first to introduce a cross-modal Transformer to sleep staging, integrating multi-scale 1D convolutions with self-attention mechanisms for end-to-end automatic feature learning and stage classification from multi-channel physiological signals (e.g., EEG, EOG, EMG). The model provides traceable decision rationales via attention weights and supports visual attribution analysis. Experiments demonstrate performance on par with state-of-the-art methods while reducing parameter count by 40% and training time by 35%. The proposed approach thus achieves a favorable trade-off among high accuracy, computational efficiency, and clinical interpretability. Code and an interactive demo are publicly available.
📝 Abstract
Accurate sleep stage classification is significant for sleep health assessment. In recent years, several machine-learning based sleep staging algorithms have been developed, and in particular, deep-learning based algorithms have achieved performance on par with human annotation. Despite improved performance, a limitation of most deep-learning based algorithms is their black-box behavior, which have limited their use in clinical settings. Here, we propose a cross-modal transformer, which is a transformer-based method for sleep stage classification. The proposed cross-modal transformer consists of a cross-modal transformer encoder architecture along with a multi-scale one-dimensional convolutional neural network for automatic representation learning. The performance of our method is on-par with the state-of-the-art methods and eliminates the black-box behavior of deep-learning models by utilizing the interpretability aspect of the attention modules. Furthermore, our method provides considerable reductions in the number of parameters and training time compared to the state-of-the-art methods. Our code is available at https://github.com/Jathurshan0330/Cross-Modal-Transformer. A demo of our work can be found at https://bit.ly/Cross_modal_transformer_demo.
