Towards Interpretable Sleep Stage Classification Using Cross-Modal Transformers

Abstract

Accurate sleep stage classification is significant for sleep health assessment. In recent years, several deep learning and machine learning based sleep staging algorithms have been developed and they 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 which have limited their use in clinical settings. Here, we propose Cross-Modal Transformers, which is a transformer-based method for sleep stage classification. Our models achieve both competitive performance with the state-of-the-art approaches and eliminates the Black-box behavior of deep-learning models by utilizing the interpretability aspect of the attention modules. The proposed cross-modal transformers consist of a novel crossmodal transformer encoder architecture along with a multiscale 1-dimensional convolutional neural network for automatic representation learning. Our sleep stage classifier based on this design was able to achieve sleep stage classification performance on par with or better than the state-of-the-art approaches, along with interpretability, a fourfold reduction in the number of parameters and a reduced training time compared to the current state-of-the-art. Our code is available at https://github. com/Jathurshan0330/Cross-Modal-Transformer.

Automatic Sleep Staging Using BiRNN with Data Augmentation and Label Redirection

“…All automatic sleep-staging algorithms, including the most advanced deep learning algorithms, show that the sleep-staging results of the Wake and REM stages in the 5-stage classification are the most accurate, almost reaching clinical levels. However, the accuracy of the N1 stage is much lower than the average accuracy of other stages [21]. Most research results show that the average accuracy of the N1 stage is only about 40% [22], which largely hinders the pace of automatic sleep staging replacing manual sleep staging.…”

Micro SleepNet: efficient deep learning model for mobile terminal real-time sleep staging

“…However, the high parameter counts of 3.7 M and the long input sequences hinder deployment and real-time inference on mobile devices. Compared to two other relatively lightweight one-to-one CNN Transformer models (Pradeepkumar et al, 2022;Yao and Liu, 2022), Micro SleepNet achieves better staging performance with an order of magnitude fewer parameters. This result indicates that introducing Transformer structures does not significantly improve CNN model performance.…”

Advances in Modeling and Interpretability of Deep Neural Sleep Staging: A Systematic Review

“…Then, multiple FC layers for sleep staging are utilized. Another transformer-based model is composed of a multi-Scale CNN block with intra-modal and cross-modal attention [167]. The main difference (besides the model architecture) between this work and ref.…”

“…The authors of [167] proposed a cross-modal transformer, which enables them to use the attention to learn and interpret: (1) intra-modal relationships, (2) cross-modal relationships, and (3) inter-epoch relationships. The intra-modal relations are similar to [166], with a difference, where they introduce the CLS c token, as discussed in Section 4.3.…”

“…It can be observed that the model highlights the important parts of the signal, which indicates which parts are most relevant to that sleep stage. Please refer to [167] for more examples. In [159], the authors used empirical mode decomposition (EMD) [160], which is a data-driven algorithm that decomposes the signal into intrinsic mode functions (IMFs).…”