szzs1693/edge-ai-cough-count

Edge Artificial Intelligence (edge-AI) Cough Counting

This is a mirrored dataset of edge-AI Cough Counting Dataset, converted to Parquet format and hosted on Hugging Face for more accessibility.

Additional preprocessing has been made, such as data aggregation and audio visualizations, for in-depth data details without downloading the whole dataset.

Background

Counting the number of times a patient coughs per day is an essential biomarker in determining treatment efficacy for novel antitussive therapies and personalizing patient care. There is a need for wearable devices that employ multimodal sensors to perform accurate, privacy-preserving, automatic cough counting algorithms directly on the device in an Edge-AI fashion. To advance this research field, our team from the Embedded Systems Laboratory (ESL) of EPFL contributed the first publicly accessible cough counting dataset of multimodal biosignals. The database contains nearly 4 hours of biosignal data, with both acoustic and kinematic modalities, covering 4,300 annotated cough events. Furthermore, a variety of non-cough sounds and motion scenarios mimicking daily life activities are also present, which the research community can use to accelerate ML algorithm development.

Data access

The edge-AI cough counting dataset can be found at the following Zenodo link: https://zenodo.org/record/7562332#.Y87MenbMKUm

Citations

If you use the open-source dataset in your work, please cite our publication:

@inproceedings{orlandic_multimodal_2023,
    address = {Sydney, Australia},
    title = {A {Multimodal} {Dataset} for {Automatic} {Edge}-{AI} {Cough} {Detection}},
    copyright = {https://doi.org/10.15223/policy-029},
    url = {https://ieeexplore.ieee.org/document/10340413/},
    doi = {10.1109/EMBC40787.2023.10340413},
    language = {en},
    urldate = {2024-04-10},
    booktitle = {2023 45th {Annual} {International} {Conference} of the {IEEE} {Engineering} in {Medicine} \& {Biology} {Society} ({EMBC})},
    publisher = {IEEE},
    author = {Orlandic, Lara and Thevenot, Jérôme and Teijeiro, Tomas and Atienza, David},
    month = jul,
    year = {2023},
    pages = {1--7},
}

Contact

For questions or suggestions, please contact lara.orlandic@epfl.ch.