Personalized ECG Monitoring Using Abnormal Beat Synthesis Filters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cardiac health monitoring systems require long periods of data collection to accurately classify abnormal heartbeats and lack accuracy when used on new patients, as they rely on patient-specific data and are not as effective as personalized systems.
Innovation Solution
A system that uses a library of abnormal beat synthesis filters to generate potential abnormal heartbeats from normal heartbeats, combined with a 1D Convolutional Neural Network trained on a personalized dataset, to classify heartbeats in real-time and alert for abnormalities, addressing the need for efficient and accurate detection across different patients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If prior personalized classification systems record both normal and abnormal beats to learn characteristics, then detection accuracy is improved, but monitoring duration is excessively long
Solution Approach 1:
The system performs preliminary action by synthesizing abnormal beat patterns from normal beats before actual monitoring begins. The ABS filters are pre-trained on a benchmark dataset to learn abnormal beat characteristics, enabling the system to generate synthetic abnormal beats for new patients without requiring long monitoring periods to collect real abnormal data.
Solution Approach 2:
The system creates copies of normal beats by applying ABS filters to generate synthetic abnormal beats. These synthesized beats are copies that mimic real abnormal patterns, allowing the system to train personalized classifiers without needing to wait for actual abnormal beats to occur during monitoring.
2Loss of time
If generic classification systems use observations from other patients, then monitoring duration is reduced, but detection accuracy deteriorates
Solution Approach 1:
The system applies local quality by personalizing the classification model for each individual patient. The ABS filters are trained on a benchmark dataset and then applied to each patient's normal beats to generate patient-specific synthetic abnormal beats, creating a personalized training set that captures individual cardiac characteristics while still enabling rapid deployment.
3Measurement precision
If personalized systems are developed for each patient, then detection accuracy is improved, but device complexity increases
Solution Approach 1:
The system achieves universality through the ABS filter library that can be applied to any patient. The same filter library, trained on a benchmark dataset, serves multiple patients by generating personalized synthetic abnormal beats for each individual, eliminating the need for separate complex training procedures for each patient while maintaining high detection accuracy.
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
Enables early detection of abnormal heartbeats with high accuracy and speed, reducing the need for extensive data collection and improving performance on new patients, while maintaining real-time monitoring capabilities.
Implementation Method 1
a sensor configured to contact skin of a user and sense electrical changes on the skin that arise from a heart's electrophysiologic pattern to generate an ECG signal
Data Source
AI summary
A method of detecting abnormal heartbeats includes providing a library of abnormal beat synthesis (ABS) filters, wherein each ABS filter corresponds to a specific cause of a cardiac problem. The method further includes obtaining an ECG of a normal heartbeat of a person and applying an ABS filter from the library of ABS filters to the ECG of the normal heartbeat of the person to generate a potential abnormal ECG. The method further includes monitoring a heartbeat of the person and classifying each heartbeat as either normal or abnormal.