Heart Murmur Classification with Sequential Machine Learning for MMVD
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for diagnosing myxomatous mitral valve disease (MMVD) in dogs are challenging due to the need for expert training and experience, and there is a high rate of misdiagnosis or late diagnosis, particularly in stage B2, which can lead to untreated or undertreated heart failure.
Innovation Solution
An apparatus and method using machine-learning-based classifiers to analyze audio signals, specifically heart murmur patterns, to automatically classify heart sounds into regular and irregular categories, and further sub-classify irregular sounds into different severity levels, utilizing pre-processing and multiple machine-learning algorithms to enhance accuracy and reduce computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual diagnosis by veterinarians is used, then diagnostic accuracy can be achieved with expert training, but it requires significant training and experience and leads to high misdiagnosis rates in stage B2
Solution Approach 1:
The patent replaces the mechanical system of manual auditory assessment by veterinarians with an automated acoustic signal processing system. The system uses microphones to capture heart sounds, digital signal processing to analyze audio patterns, and machine learning algorithms to classify murmurs, eliminating the need for extensive veterinary training while maintaining or improving diagnostic accuracy.
Solution Approach 2:
The system enables self-diagnosis capability by automatically analyzing heart sounds and providing staging classifications without requiring professional veterinary intervention. The automated system performs the diagnostic function that previously required expert veterinarians, making the service accessible to pet owners while maintaining diagnostic quality.
2Measurement precision
If manual heart murmur grading is performed, then staging can be determined, but it is time-consuming and relies on expert experience
Solution Approach 1:
The patent replaces the time-consuming manual grading process with automated acoustic analysis. The system uses digital signal processing to extract features from heart sound recordings and machine learning models to rapidly classify murmurs into stages, reducing diagnosis time from minutes of expert listening to automated processing while maintaining staging accuracy.
Solution Approach 2:
The system performs preliminary automated analysis of heart sound patterns to pre-classify murmurs before expert review or treatment decisions are made. The automated staging classification can be performed immediately upon recording, providing rapid preliminary diagnosis that guides immediate care while allowing for subsequent verification.
3Measurement precision
If multiple machine-learning classifiers are used in sequence, then classification accuracy improves, but computational complexity increases
Solution Approach 1:
The patent divides the classification task into sequential stages using multiple specialized machine learning classifiers. A first classifier performs preliminary screening to identify potential murmurs, and a second classifier provides refined staging classification. This segmentation allows each classifier to focus on specific aspects of the data, improving overall accuracy while managing computational complexity through staged processing.
Solution Approach 2:
The system dynamically adjusts the classification process based on the input data characteristics. The machine learning models adapt their analysis depth and computational resources based on the complexity of the audio signal and the required precision level, allowing the system to optimize between accuracy and computational efficiency in real-time.
Data Source
AI summary
An apparatus for classifying at least one audio signal has an input interface configured to receive input information of the audio signal, a trained first machine-learning-based classifier configured to map the input information to one of a first and a second class of audio signals; a trained second machine-learning-based classifier configured to, if the audio signal belongs to the first class of audio signals, map the input information of the audio signal belonging to the first class of audio signals to one of a plurality of third classes of audio signals, and an output interface configured to output information on which classes the audio signal belongs to.