Sleep staging method, computer device, and readable storage medium

By calculating the cardiopulmonary coupling component in sleep physiological signals and generating a unified representation, the performance degradation problem when PSG signals are transferred to the BCG scenario is solved, and the adaptability and stability of the sleep staging model are improved.

CN122398237APending Publication Date: 2026-07-17
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Applications(China)
Filing Date
2026-06-10
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

Existing deep learning models pre-trained based on PSG signals exhibit significant performance degradation when transferred to BCG scenarios, failing to effectively adapt to the feature differences of BCG signals.

Method used

By extracting the respiratory and instantaneous heart rate components from sleep physiological signals, calculating the cardiopulmonary coupling component, mapping it to a unified time axis, normalizing and weighting it, and generating a unified representation for prediction using sleep staging models.

Benefits of technology

It improves the performance of the sleep staging model in the BCG scenario, reduces structural bias in cross-modal transfer, enhances the model's adaptability to low-quality signals and complex sleep dynamics, and improves the stability of sleep staging results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122398237A_ABST
    Figure CN122398237A_ABST
Patent Text Reader

Abstract

The application discloses a sleep staging method, a computer device and a readable storage medium, and belongs to the sleep medical field. The method comprises the following steps: obtaining a sleep physiological signal, obtaining a respiratory component and an instantaneous heart rate component based on the sleep physiological signal; mapping the respiratory component and the instantaneous heart rate component to a unified time axis and resampling to the same sampling frequency; dividing the time axis into multiple Epochs with fixed time lengths; for each Epoch, respectively calculating the synergistic change coefficient between the resampled respiratory component and the resampled instantaneous heart rate component, and normalizing and weightedly fusing the synergistic change coefficient to obtain a heart-lung coupling component corresponding to each Epoch; obtaining a unified representation based on the respiratory component, the instantaneous heart rate component and the heart-lung coupling component, calling a sleep staging model based on the unified representation, and obtaining sleep stage prediction results corresponding to each Epoch respectively. The application improves the model performance of a pre-training model obtained based on a PSG signal and migrated to a BCG scene.
Need to check novelty before this filing date? Find Prior Art