Signal Processing Apparatus for Heartbeat Interval Determination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for determining heartbeat intervals and heart rate from bio-vibration signals, such as those derived from piezoelectric sensors, face challenges due to disturbances in waveforms caused by sensor position and subject movement, making it difficult to obtain accurate results.
Innovation Solution
A signal processing apparatus and system that processes bio-vibration signals using differentiation, high-pass filtering, and absolutization to extract heart sound signals, and employs machine learning to predict ECG signals equivalent to those obtained from traditional ECG measurements, allowing for the determination of heartbeat intervals and heart rate.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional methods are used to determine heartbeat intervals from bio-vibration signals, then the measurement is non-intrusive and comfortable for the subject, but the waveform disturbances make it difficult to obtain accurate results
Solution Approach 1:
The patent introduces an intermediary machine learning model that maps bio-vibration signals to ECG signals. This intermediary transformation allows the system to leverage the comfort of non-intrusive bio-vibration measurement while achieving the accuracy of ECG-based heartbeat interval determination, effectively resolving the contradiction between ease of operation and measurement precision
Solution Approach 2:
The patent replaces the direct mechanical signal processing approach (filtering and analyzing raw bio-vibration signals) with a data-driven machine learning system. This substitution enables accurate heartbeat interval extraction from disturbed waveforms by learning the complex nonlinear relationships between bio-vibration and ECG signals, thereby maintaining measurement comfort while improving accuracy
2Measurement precision
If ECG electrodes are attached on extremities or chest for continuous measurement, then accurate ECG signals are obtained, but it puts a substantial burden on the subject
Solution Approach 1:
The patent creates a virtual copy of ECG signals by training a machine learning model to reproduce ECG waveforms from bio-vibration signals. This copying approach allows the system to obtain accurate ECG-like signals without requiring physical ECG electrodes, thereby eliminating the subject burden while maintaining measurement precision
Solution Approach 2:
The patent makes the bio-vibration sensor multi-functional by enabling it to capture not only pulse wave information but also ballistocardiac movements that contain ECG-correlated data. This universality allows a single non-intrusive sensor to replace both ECG electrodes and pulse wave sensors, achieving accurate heartbeat interval measurement without subject burden
3Adaptability or versatility
If the piezoelectric sensor is placed on or under bedding or attached on body surface, then bio-vibration signals are acquired non-restrictively, but the waveforms are occasionally disturbed depending on sensor position
Solution Approach 1:
The patent performs preliminary action by training the machine learning model on diverse data collected from multiple sensor positions and conditions during the learning phase. This preliminary exposure to various disturbances enables the model to compensate for position-dependent waveform variations during actual measurement, maintaining signal reliability while preserving placement flexibility
Solution Approach 2:
The patent introduces dynamics by using a machine learning model that can adapt to varying signal characteristics. The model dynamically adjusts its predictions based on the input bio-vibration signal patterns, enabling reliable heartbeat interval determination even when sensor position changes cause waveform disturbances, thus maintaining reliability while preserving adaptability
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 effectively generates signals equivalent to ECG signals from bio-vibration data, enabling accurate determination of heartbeat intervals and heart rate, even in cases where conventional methods fail due to waveform disturbances.
Implementation Method 1
Existing piezoelectric sensors use a sheet-type piezoelectric element made of polyvinylidene difluoride (PVDF), a fluoride organic ferroelectric material. PVDF, a piezoelectric polymer material, produces electric polarization in response to increased displacement of ions arranged in crystal lattices by pressure or deformation.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
An apparatus yields signals that are equivalent to ECG signals and allow determination of a heartbeat interval or heart rate from bio-vibration signals including vibrations derived from heartbeats. An ECG meter 30 acquires ECG signals of a sample, and a piezoelectric sensor 40 acquires bio-vibration signals of the sample simultaneously. The bio-vibration signals include beating vibration signals derived from heartbeats. A learning unit 13 of a prediction modeling apparatus 10 establishes a prediction model 24 by machine learning in which ECG signals are used as teaching data, and model input signals obtained by performing a specified processing on the bio-vibration signals are input. The learning unit 13 delivers the prediction model 24 to a prediction unit 23 of a signal processing apparatus 20. The prediction model 24 predicts and outputs pECG signals upon input of model input signals obtained by performing a specified processing on bio-vibration signals acquired from a subject under prediction with a piezoelectric sensor 40.