Wearable Cardiac Activity Detection with Motion Noise Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting abnormal cardiac activity, such as atrial fibrillation, are cumbersome, invasive, costly, and inefficient, particularly for long-term monitoring, and wearable devices using photoplethysmogram (PPG) data are prone to noise from user movement, masking arrhythmia occurrences.
Innovation Solution
A deep learning architecture utilizing a combination of convolutional neural networks, bidirectional recurrent neural networks, and attention networks to analyze cardiac and motion data, including ECG and accelerometer data, to accurately classify cardiac activity into normal and abnormal categories, incorporating subject contextual information for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If ECG monitoring is used periodically over a few weeks to detect arrhythmias, then abnormal cardiac activity can be detected, but the method becomes cumbersome and invasive requiring multiple electrodes and patches
Solution Approach 1:
The patent replaces the mechanical electrode-ECG system with an optical PPG-based system using light-emitting diodes and photodetectors. This substitution eliminates the need for multiple electrodes and patches while maintaining arrhythmia detection capability, directly resolving the contradiction between detection reliability and ease of operation
Solution Approach 2:
The patent uses photoplethysmogram (PPG) signals as an optical copy/alternative to electrical ECG signals for detecting cardiac arrhythmias. By capturing blood volume changes through light absorption rather than electrical potentials, the system achieves comparable diagnostic value with significantly improved user comfort and ease of operation
2Ease of operation
If PPG data is used in wearable devices to detect cardiac activity, then long-term monitoring becomes feasible and less invasive, but movement noise masks occurrences of arrhythmia
Solution Approach 1:
The patent segments the PPG signal processing into multiple independent components: raw signal acquisition, noise identification module, artifact removal module, and arrhythmia detection module. This segmentation allows targeted processing of movement artifacts while preserving genuine cardiac signals, resolving the contradiction between wearability and measurement precision
Solution Approach 2:
The patent introduces intermediate processing modules between signal acquisition and arrhythmia detection. These modules include noise identification and artifact removal stages that act as mediators to filter movement-induced noise while preserving the underlying cardiac rhythm information, thereby maintaining measurement precision in wearable applications
3Reliability
If current ECG methods are used for long-term monitoring, then arrhythmia detection is possible, but the cost becomes prohibitively high
Solution Approach 1:
The patent employs inexpensive optical components (LEDs and photodetectors) that can be easily manufactured and integrated into consumer-grade wearable devices. This approach replaces expensive medical-grade ECG hardware while maintaining sufficient detection capability for long-term monitoring, directly addressing the cost effectiveness contradiction
Solution Approach 2:
The patent creates a multi-functional system where a single wearable device can perform both cardiac arrhythmia detection and general health monitoring. By combining PPG sensing with motion sensors and implementing comprehensive signal processing, the system replaces multiple specialized devices, reducing overall system cost while maintaining detection reliability
Data Source
AI summary
The systems and methods can accurately and efficiently determine abnormal cardiac activity from motion data and/or cardiac data using techniques that can be used for long-term monitoring of a patient. In some embodiments, the method for using machine learning to determine abnormal cardiac activity may include receiving one or more periods of time of cardiac data and motion data for a subject. The method may include applying a trained deep learning architecture to each tensor of the one or more periods of time to classify each window and/or each period into one or more classes using at least the one or more signal quality indices for the cardiac data and the motion data and cardiovascular features. The deep learning architecture may include a convolutional neural network, a bidirectional recurrent neural network, and an attention network. The one or more classes may include abnormal cardiac activity and normal cardiac activity.