Mobile Heart Murmur Classifier Using Spectral Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Primary care physicians, such as pediatricians, face challenges in distinguishing between Still's murmur and pathological heart murmurs, leading to unnecessary referrals and increased healthcare costs due to their inability to reliably diagnose Still's murmur, which is a benign condition.
Innovation Solution
A mobile device-based system that includes a stethoscope attachment and circuitry to process heart signals, allowing for the classification of Still's murmur by segmenting the heart signal, computing spectral width and peak frequency, and analyzing the shape of the signal's envelope, thereby differentiating it from pathological murmurs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If primary care physicians use only a stethoscope to detect heart murmurs, then the detection method is simple and cost-effective, but the ability to reliably distinguish between Still's murmur and pathological murmur is insufficient
Solution Approach 1:
A mobile device is introduced as an intermediary between the stethoscope and the physician. The mobile device receives heart signals from the stethoscope, processes them through spectral analysis algorithms, and provides classification results. This intermediary approach enhances diagnostic accuracy without requiring the physician to directly perform complex signal processing.
Solution Approach 2:
The patent replaces the purely mechanical/subjective listening process with an automated digital signal processing system. The mobile device performs spectral analysis, extracts features such as spectral width and peak frequency, and automatically classifies murmurs, substituting the mechanical stethoscope listening process with electronic analysis while retaining the simple stethoscope interface.
2Reliability
If primary care physicians refer all detected murmurs to specialists, then patient safety is ensured by avoiding misdiagnosis, but healthcare costs increase and unnecessary referrals occur
Solution Approach 1:
The mobile device performs preliminary classification of heart murmurs at the primary care level before specialist referral. By analyzing spectral features and providing a preliminary diagnosis of Still's murmur versus pathological murmur, the system enables physicians to confidently manage benign cases without referral, reserving specialist consultation for cases that require further evaluation.
Solution Approach 2:
The system applies partial automation by focusing specifically on detecting Still's murmur characteristics rather than attempting to classify all types of heart murmurs. This targeted approach provides sufficient diagnostic confidence for benign cases while maintaining safety margins for ambiguous cases that may still require specialist evaluation.
3Measurement precision
If automated signal processing is implemented to improve murmur detection accuracy, then diagnostic precision improves, but the device complexity and processing requirements increase
Solution Approach 1:
The patent extracts only the most discriminative spectral features from the heart signals, specifically spectral width and peak frequency within the 80-800 Hz range. By focusing on these key features rather than performing exhaustive spectral analysis, the system achieves accurate Still's murmur detection with reduced computational requirements and lower energy consumption.
Data Source
Figure 1
Figure 2A~2B
Figure 3
AI summary
Discussed herein are an apparatus and a corresponding method, for classifying a murmur detected in a heart signal as a Still's murmur. Circuitry included in the apparatus segments a heart signal to detect a first lobe (S1) and a second lobe (S2) of the heart signal. Further, the circuitry computes a spectral width of an extracted feature of the heart signal that lies in a portion of a time interval between the S1 lobe and the S2 lobe based on an estimated power spectral density. Additionally, a peak frequency of the extracted feature of the heart signal and a shape of an envelope of the extracted feature of the heart signal are computed. The extracted feature of the heart signal is classified as the Still's murmur based on the spectral width, the peak frequency, the shape of the envelop, and additional murmur discriminant features.