Sleep Stage Detection Using PPG Spectral Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional methods for determining an individual's sleep stage are inaccurate and require costly or cumbersome equipment, failing to provide reliable measurements of sleep quality and potential physiological conditions.

Innovation Solution

A computer-implemented system that uses probabilistic estimation based on photoplethysmographic (PPG) signals and accelerometer data to classify sleep stages, incorporating signal features such as beat-to-beat intervals and spectrograms, without the need for additional equipment, and can identify abnormal sleep behaviors and conditions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional measurement systems are used to determine sleep stages, then equipment cost and complexity are reduced, but measurement precision and reliability deteriorate

Engineering Contradiction:
Improvesleep stage detection accuracyVSAvoidequipment complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces complex mechanical/electrical measurement systems (EEG electrodes, EMG sensors) with optical sensing (photoplethysmography) and simple motion detection (accelerometry). This substitution maintains measurement precision for sleep stage detection while dramatically reducing device complexity and making the system wearable and portable.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces intermediate signal processing steps including spectral analysis of heart rate variability and motion spectral analysis as mediators between the raw sensor data and sleep stage classification. These intermediary analyses extract meaningful features that bridge the gap between simple sensor measurements and accurate sleep stage determination.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Reliability

If conventional measurement systems are used, then device simplicity is maintained, but reliability of sleep stage determination deteriorates

Engineering Contradiction:
Improvesleep stage determination reliabilityVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent performs preliminary spectral analysis of heart rate variability and motion data before final sleep stage classification. By pre-processing the signals to extract frequency domain features and patterns, the system builds a more reliable foundation for classification, reducing the need for complex post-processing and improving overall determination reliability.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system incorporates feedback mechanisms where the classified sleep stages are used to refine and update the classification model over time. This allows the system to learn from accumulated data and improve its reliability continuously, compensating for the simplicity of the underlying sensors.

Inventive Principle:
Principle #23Feedback

3Measurement precision

If spectral analysis of heart rate and motion is used, then sleep stage detection accuracy is improved, but computational complexity increases

Engineering Contradiction:
Improvesleep stage classification accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts only the most relevant spectral features from heart rate and motion signals for sleep stage classification, rather than analyzing the complete signal spectrum. By selectively extracting specific frequency components and temporal patterns that are most discriminative for different sleep stages, the system maintains high accuracy while reducing computational burden.

Inventive Principle:
Principle #2Taking out (Extraction)

4Device complexity

If photoplethysmographic signals and accelerometer data are used, then specialized equipment requirements are reduced, but measurement precision may deteriorate

Engineering Contradiction:
Improveequipment requirementsVSAvoidsleep stage measurement accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent transitions from analyzing signals in the time domain to analyzing them in the frequency domain through spectral analysis. This dimensional transformation allows the system to extract subtle patterns and rhythms from the simple PPG and accelerometer signals that are not apparent in the time domain, thereby maintaining measurement precision while using simple sensors.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

The system provides more accurate measurements of sleep stages and detects abnormal sleep conditions without requiring specialized equipment, improving the accuracy of sleep stage classification and identifying conditions like Insomnia and REM Behavior Disorder.

Implementation Method 1

A PPG signal is received from a PPG sensor

Methodology Applied
Scientific EffectPhotoplethysmography: Photoelectric Effect

Data Source

PatentEP3410924B1Machine learnt model to detect REM sleep periods using a spectral analysis of heart rate and motion
Publication Date: 2023.09.13 VERILY LIFE SCIENCES LLC
  • EP3410924B1 patent drawingFigure 1
  • EP3410924B1 patent drawingFigure 2
  • EP3410924B1 patent drawingFigure 3

AI summary

Systems and methods are provided for probabilistically estimating an individual's sleep stage based on spectral analyses of pulse rate and motion data. The embodiments include receiving signals from sensors worn by the individual, including a photoplethysmographic (PPG) signal and an accelerometer signal. The embodiments may divide the PPG signal into segments and determining a beat interval associated with each segment. The embodiments may resample the set of beat intervals to generate an interval signal. The embodiments may generate signal features based on the interval signal and the accelerometer signal, including a spectrogram of the interval signal. The embodiments may determine a sleep stage for the individual by comparing the signal features to a sleep stage classifier included in a learning library, wherein the sleep stage classifier comprises one or more functions defining a likelihood that the individual is in the sleep stage based on the signal features.