Authors: María Teresa García-Ordás, José Alberto Benítez-Andrades, Isaías García-Rodríguez, Carmen Benavides, Héctor Alaiz-Moretón

Published on: February 03, 2024

Impact Score: 8.2

Arxiv code: Arxiv:2402.02183

Summary

• What is new: Using a Variational Convolutional Autoencoder to generate new labeled data for respiratory sound classification, addressing the issue of class imbalance.
• Why this is important: The challenge of detecting pathologies through respiratory sounds from an unbalanced dataset.
• What the research proposes: A combination of Variational Convolutional Autoencoder for data generation and a Convolutional Neural Network for classification.
• Results: Achieved up to 0.993 F-Score in three-label classifications and 0.990 F-Score in six-class classifications.

Technical Details

Technological frameworks used: Variational Convolutional Autoencoder, Convolutional Neural Networks

Models used: nan

Data used: ICBHI (International Conference on Biomedical and Health Informatics) Benchmark dataset

Potential Impact

Healthcare industry, medical diagnostics companies, medical device manufacturers

Want to implement this idea in a business?

We have generated a startup concept here: RespiroTrack.