A method based on cardiopulmonary coupling analysis for sleep quality assessment with FPGA implementation.

Area of Science:

• Biomedical Engineering
• Physiological Signal Processing
• Artificial Intelligence in Healthcare

Background:

• Assessing sleep quality is crucial for clinical diagnosis, with increasing links to various diseases and overall wellness.
• Current methods relying on self-reports and psychological variables are limited by poor self-observation of sleep behaviors.
• There is a need for objective and reliable sleep quality assessment methods.

Purpose of the Study:

• To develop an objective method for sleep quality estimation using physiological signals.
• To validate the efficacy of a convolutional neural network (CNN) model for sleep quality assessment.
• To create a portable home monitoring device for sleep analysis.

Main Methods:

• Examined single-lead electrocardiogram (ECG) signals to estimate cardiopulmonary coupling.
• Derived respiration and normal-to-normal sinus interbeat interval series from ECG.
• Utilized a CNN model trained on a one-dimensional array of the coupling signal, referencing age-related cyclic alternating pattern (CAP) rate percentages.

Main Results:

• The developed CNN model achieved 91% accuracy in sleep quality estimation.
• The model demonstrated an area under the receiver operating characteristic curve (AUC) of 97%.
• Performance metrics are comparable to or exceed current state-of-the-art methods.

Conclusions:

• The proposed method, based on ECG-derived cardiopulmonary coupling and CNN analysis, offers a highly accurate approach to sleep quality assessment.
• The successful implementation on a field-programmable gate array (FPGA) board enables the creation of a resilient, user-friendly home monitoring device.
• This technology holds significant potential for clinical analysis and remote patient monitoring.