Attention-based Learning for Sleep Apnea and Limb Movement Detection using Wi-Fi CSI Signals

Chi-Che Chang,An-Hung Hsiao,Li-Hsiang Shen,Kai-Ten Feng,Chia-Yu Chen

2023-03-27

Abstract:Wi-Fi channel state information (CSI) has become a promising solution for non-invasive breathing and body motion monitoring during sleep. Sleep disorders of apnea and periodic limb movement disorder (PLMD) are often unconscious and fatal. The existing researches detect abnormal sleep disorders in impractically controlled environments. Moreover, it leads to compelling challenges to classify complex macro- and micro-scales of sleep movements as well as entangled similar waveforms of cases of apnea and PLMD. In this paper, we propose the attention-based learning for sleep apnea and limb movement detection (ALESAL) system that can jointly detect sleep apnea and PLMD under different sleep postures across a variety of patients. ALESAL contains antenna-pair and time attention mechanisms for mitigating the impact of modest antenna pairs and emphasizing the duration of interest, respectively. Performance results show that our proposed ALESAL system can achieve a weighted F1-score of 84.33, outperforming the other existing non-attention based methods of support vector machine and deep multilayer perceptron.

Signal Processing,Machine Learning

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to non - invasively detect sleep apnea and periodic limb movement disorder (PLMD) through Wi - Fi channel state information (CSI). Specifically, the author points out that existing research usually conducts the detection of abnormal sleep disorders in unrealistic controlled environments, and it is difficult to distinguish complex macro - and micro - scale sleep movements and intertwined similar waveforms. In addition, existing methods have poor generalization ability among different sleeping postures and patients.

To solve these problems, the author proposes a learning system based on the attention mechanism (ALESAL) for the joint detection of sleep apnea and PLMD. The following are several key points of this system:

1. **Problem Background**:
- Sleep apnea and PLMD are common sleep disorders that may lead to serious health problems.
- Most existing research is carried out in controlled environments and cannot be applied to real - life scenarios.
- A method that can effectively detect these disorders among different sleeping postures and patients is required.

2. **Solution**:
- A learning system based on the attention mechanism (ALESAL) is proposed, which can detect sleep apnea and PLMD simultaneously.
- ALESAL includes an antenna - pair attention mechanism and a temporal attention mechanism to reduce the impact of antenna - pair quality differences and emphasize the time period of interest.
- This system can perform detection in different sleeping postures and among multiple patients, improving robustness and accuracy in practical applications.

3. **Main Contributions**:
- This is the first study to jointly detect sleep apnea and PLMD in any sleeping posture using CSI signals combined with open medical data.
- A temporal - attention neural network is designed to emphasize the time period related to the respiration rate, and an antenna - pair attention mechanism is introduced to handle the signal quality differences provided by different antenna pairs.
- The experimental results show that the weighted F1 - score of the ALESAL system reaches 84.33%, which is superior to other existing methods.

Through these improvements, the ALESAL system can accurately detect sleep apnea and PLMD under a wider range of conditions, thus providing new possibilities for non - invasive health monitoring.