Digital Stethoscope Respiratory Event Prediction Using Convolutional Neural Networks

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current digital stethoscopes face limitations in real-time performance for detecting respiratory abnormalities, inability to forecast future respiratory events, and inadequate data cataloging while ensuring patient privacy, with existing systems requiring a doctor's presence and lacking severity prediction capabilities.

Innovation Solution

A digital stethoscope system that uses convolutional neural networks to analyze auditory signals, tracks coughs, predicts respiratory event severity, and forecasts future events by generating convolution vectors and applying trained weights, with a base station for enhanced processing and wireless charging, and cloud-based data storage for secure data management.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If a doctor uses a traditional stethoscope to diagnose respiratory illnesses, then the diagnosis can be made, but the need for a doctor makes daily monitoring impractical and the system requires human presence

Engineering Contradiction:

Improveautomation of respiratory abnormality detectionVSAvoidcomplexity of the monitoring system

Core Design Contradiction:

Extent of automationVSDevice complexity

Solution Approach 1:

The patent replaces the mechanical/stethoscope-based system with a digital electronic system that uses microphones, processors, and algorithms to detect and classify respiratory sounds, eliminating the need for a doctor's physical presence while maintaining diagnostic capability

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables self-monitoring by automatically detecting, classifying, and alerting patients or caregivers about respiratory abnormalities without requiring professional medical intervention for each monitoring event, allowing continuous independent use

Inventive Principle:

Principle #25Self-service

2Measurement precision

If a classification system is used to distinguish normal from abnormal noises, then detection accuracy improves, but real-time performance is insufficient for local execution on patient devices

Engineering Contradiction:

Improveaccuracy of respiratory abnormality detectionVSAvoidreal-time processing speed

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The classification system is divided into multiple stages: preliminary filtering of obvious normal sounds, extraction of key acoustic features, and then application of the classification algorithm only to extracted features rather than raw audio streams, enabling real-time processing

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system performs preliminary processing by extracting and pre-processing acoustic features before feeding them to the classification algorithm, preparing the data in advance to reduce computational load during real-time classification

Inventive Principle:

Principle #10Preliminary action

3Measurement precision

If a classification system predicts whether a noise is normal or abnormal, then detection capability is provided, but the system cannot predict the severity of future respiratory events or their characteristics

Engineering Contradiction:

Improvedetection capability of respiratory abnormalitiesVSAvoidpredictive capability for future events

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The system continuously monitors respiratory patterns and maintains baseline data, performing preliminary analysis of trends and deviations before actual events occur, enabling prediction of future respiratory events based on accumulating pattern recognition

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system uses feedback from continuous monitoring of respiratory patterns, comparing current readings against historical data and established baselines to detect evolving conditions and predict future events before they manifest as acute abnormalities

Inventive Principle:

Principle #23Feedback

4Quantity of substance

If data from in-home stethoscopes is cataloged to improve monitoring, then comprehensive data collection is achieved, but patient privacy interests are compromised

Engineering Contradiction:

Improvevolume of respiratory data collectedVSAvoidrisk to patient privacy

Core Design Contradiction:

Quantity of substanceVSObject-affected harmful factors

Solution Approach 1:

The system extracts and analyzes only the necessary acoustic features and classification results, storing minimal de-identified data while removing raw audio files and detailed personal information from the storage system, reducing the privacy risk surface area

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The system introduces an intermediary processing layer that anonymizes and aggregates data before storage and analysis, acting as a buffer between raw patient data and the storage system to prevent direct exposure of identifiable information

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS10709414B1Predicting a respiratory event based on trend information, and applications thereof

Publication Date: 2020.07.14 SONAVI LABS INC

AI summary

Embodiments disclosed herein improve digital stethoscopes and their application and operation. A first method detects of a respiratory abnormality using a convolution. A second method counts coughs for a patient. A third method predicts a respiratory event based on a detected trend. A fourth method forecasts characteristics of a future respiratory event. In a fifth embodiment, a base station is provided for a digital stethoscope.