Feature-Gated Machine Learning for Low-Power Arrhythmia Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing medical devices face challenges in accurately detecting cardiac arrhythmias while efficiently managing power consumption, as feature delineation algorithms require expert design and machine learning methods are computationally prohibitive for battery-powered devices.
Innovation Solution
A medical device system that combines feature delineation with machine learning to verify arrhythmia detection, using low-power feature delineation to identify relevant data for machine learning analysis, thereby increasing accuracy and conserving battery life.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning methods are used for arrhythmia detection, then detection accuracy is improved, but power consumption becomes prohibitive for battery-powered devices
Solution Approach 1:
The system segments the detection process into two distinct stages: a low-power feature delineation stage that processes all cardiac signals continuously, and a machine learning verification stage that is selectively applied only to episodes flagged by the feature delineation algorithm. This segmentation allows the computationally intensive machine learning model to be used sparingly, preserving battery life while maintaining high detection accuracy for suspected arrhythmia episodes.
Solution Approach 2:
The feature delineation algorithm performs preliminary analysis of cardiac signals to identify potential arrhythmia episodes before applying the machine learning model. By pre-processing signals and flagging only suspicious episodes for machine learning verification, the system prepares data in advance in a way that minimizes subsequent computational requirements and power consumption.
2Use of energy by moving object
If feature delineation algorithms are used, then power consumption is reduced, but detection accuracy and flexibility are limited
Solution Approach 1:
The system merges two complementary approaches: the low-power feature delineation algorithm that provides continuous monitoring with reduced energy consumption, and the high-accuracy machine learning model that verifies suspected episodes. The combination leverages the strengths of both methods—the efficiency of feature delineation and the accuracy of machine learning—achieving both low power consumption and high detection accuracy simultaneously.
Solution Approach 2:
The feature delineation algorithm serves as an intermediary between the continuous cardiac signal monitoring and the computationally intensive machine learning verification. It processes all incoming signals at low power consumption and selectively passes only promising episodes to the machine learning model, acting as a gatekeeper that optimizes the overall system efficiency while maintaining accuracy.
3Measurement precision
If machine learning models are applied continuously, then detection accuracy is maximized, but device complexity and computational burden increase
Solution Approach 1:
The detection system is segmented into a simple continuous monitoring layer using feature delineation and a complex verification layer using machine learning. This segmentation reduces the average computational complexity by ensuring the machine learning model is invoked only when necessary, rather than continuously processing all cardiac signals.
Solution Approach 2:
Instead of applying the full machine learning model to all cardiac signals (excessive action), the system applies it partially—only to episodes that are flagged by the feature delineation algorithm as potential arrhythmias. This partial application maintains high detection accuracy for critical episodes while dramatically reducing overall computational burden.
Data Source
AI summary
Techniques are disclosed for using feature delineation to reduce the impact of machine learning cardiac arrhythmia detection on power consumption of medical devices. In one example, a medical device performs feature-based delineation of cardiac electrogram data sensed from a patient to obtain cardiac features indicative of an episode of arrhythmia in the patient. The medical device determines whether the cardiac features satisfy threshold criteria for application of a machine learning model for verifying the feature-based delineation of the cardiac electrogram data. In response to determining that the cardiac features satisfy the threshold criteria, the medical device applies the machine learning model to the sensed cardiac electrogram data to verify that the episode of arrhythmia has occurred or determine a classification of the episode of arrhythmia.