Exploring Model Architectures for Real-Time Lung Sound Event Detection (ESANN 2025)

Code for the publication "Exploring Model Architectures for Real-Time Lung Sound Event Detection" by Michiel Jacobs, Lode Vuegen, Tom Verresen, Marie Schouterden, David Ruttens & Peter Karsmakers, which was presented at European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN) (April 2025).

This work was built starting from the work "Patch-Mix Contrastive Learning with Audio Spectrogram Transformer on Respiratory Sound Classification" by Sangmin Bae, June-Woo Kim, Won-Yang Cho, et al (Interspeech 2023).

Data: ICBHI 2017

Get the ICBHI 2017 lung sound dataset through https://bhichallenge.med.auth.gr/. TBD (how and where should this data be saved)

Running The Code

TBD (Conda venv requirements).

Training Models

TBD (bash script, make file).

Contact

Michiel Jacobs (KU Leuven)

Citing

If you use this repository or ideas presented in the corresponding paper, please consider citing (BibTeX):

@inproceedings{esann2025_jacobsm_etal,
  author = {Jacobs, Michiel and Vuegen, Lode and Verresen, Tom and Schouterden, Marie and Ruttens, David and Karsmakers, Peter},
  title = {Exploring Model Architectures for Real-Time Lung Sound Event Detection},
  year = {2025},
}