README.md

metadata

license: cc-by-nc-sa-4.0
task_categories:
  - audio-classification
tags:
  - medical
configs:
  - config_name: default
    data_files:
      - split: csa
        path: aware_csa.csv

Acoustic Waveform Airway and Respiratory Examination (AWARE/PTEase) Dataset

Guidelines

AWARE/PTEase is a smartphone-based sensing system that examines human airway's internal physiological conditions, developed by the Intelligent Systems Laboratory at University of Pittsburgh. AWARE/PTEase probes the airway with acoustic pulses through mouth, and collect the airway's reflections for analysis. Please refer to our paper and github repo for more details.

This dataset includes the raw and pre-processed acoustic data of airway measurements collected from 382 human subjects recruited at the Children’s Hospital of Pittsburgh. Demographics of the recruited human subjects are as follows：

Category	Characteristics	Number
Demographics	Age (years)	22.55 ± 15.63
	Adults (%)	166 (43.46)
	Female (%)	205 (53.66)
	Caucasian (%)	287 (75.13)
	African-American (%)	93 (24.35)
Body Conditions	Height (cm)	159.76 ± 15.67
	Weight (kg)	65.79 ± 27.28
Disease Conditions	Healthy (%)	115 (30.10)
	Asthma (%)	190 (49.74)
	Cystic Fibrosis (%)	64 (16.75)
	COPD (%)	8 (2.09)
	Others (%)	5 (1.31)

The contents of this dataset include three parts:

• Raw WAV files

  • "aware_raw.zip"

    • Subfolders are named by the subjects' ID (from 1 to 382). In each subfolder, there are three "cali_X.wav" files and multiple "test_X.wav" files

      • "cali_1.wav" : Calibration step 1 - long tube calibration - consists of 12 pairs of received pulses. Each pair contains one wide-band (0-16Hz) pulse and one narrow-band (0-200Hz) pulse. The narrow-band pulses are deperacated and only the wide-band pulses are used. Transmitting period for each pair is 0.4 sec.

      • "cali_2.wav" : Calibration step 2 - standard tube close-end calibration - is the same as above except the calibrations are conducted in a different setting.

      • "cali_3.wav" : Calibration step 3 - standard tube open-end calibration - is the same as above except the calibrations are conducted in a different setting.

      • "test_X.wav" : Test number X - the subject's X-th attempt of conducting the test - follows the protocol of "Nasal breathing -> Inhale -> Exhale -> Inhale -> Exhale -> Inhale -> Exhale". Each segment contains 25 pulses collected at an interval of 0.2 sec. NOTE: Some of the samples have 50 pulses for the "Nasal breathing" phase.

• Segmented and aligned acoustic pulses

  • "aware_segmented.pkl"

    • Python pandas.DataFrame saved as a PICKLE file. Same contents as "aware_segmented.csv", but without flattening.

    • To load the data, use the following commands:

      import pandas as pd  
      df = pd.read_pickle("aware_segmented.pkl") 
      

    • Column 0-45 : Basic subject information, including demographics, clinical measurements and labels.

      • Column 0 : Subject ID
      • Column 1-9 : Demographics
      • Column 10 : Disease label
      • Column 11-21 : Impulse Oscillometry Measurements
      • Column 22-30 : Spirometry measurements
      • Column 31-44 : Spirometry measurements in post-bronchodilator test
      • Column 45 : Test number

    • Column 46-55 : Acoustic signals (16-bit PCM, sample rate = 48000) arranged by different phases. Each calibration phase contains 12 measurements and each testing phase contains 25 measurements. Each measurement lasts for 0.1 sec (4800 sample points).

      • Column 46 : Phase "Calibration #1". Each element is a [12,4800] NumPy ndarray
      • Column 47 : Phase "Calibration #2". Each element is a [12,4800] NumPy ndarray
      • Column 48 : Phase "Calibration #3". Each element is a [12,4800] NumPy ndarray
      • Column 49 : Phase "Nasal breathing". Each element is a [25,4800] NumPy ndarray
      • Column 50 : Phase "Inhale #1". Each element is a [25,4800] NumPy ndarray
      • Column 51 : Phase "Inhale #2". Each element is a [25,4800] NumPy ndarray
      • Column 52 : Phase "Inhale #3". Each element is a [25,4800] NumPy ndarray
      • Column 53 : Phase "Exhale #1". Each element is a [25,4800] NumPy ndarray
      • Column 54 : Phase "Exhale #2". Each element is a [25,4800] NumPy ndarray
      • Column 55 : Phase "Exhale #3". Each element is a [25,4800] NumPy ndarray

  • "aware_segmented.csv" and "headers.txt"

    • NOTE: Directly loading this CSV file is not recommended. It could be time-consuming and buggy. Try "aware_segmented.pkl" first.

    • NOTE: "aware_segmented.csv" has no headers. See column definition below.

    • Column 0-45 : Basic subject information, including demographics, clinical measurements and labels.

      • Column 0 : Subject ID
      • Column 1-9 : Demographics
      • Column 10 : Disease label
      • Column 11-21 : Impulse Oscillometry Measurements
      • Column 22-30 : Spirometry measurements
      • Column 31-44 : Spirometry measurements in post-bronchodilator test
      • Column 45 : Test number

    • Column 46-1012845 : Acoustic signals (16-bit PCM, sample rate = 48000) arranged by different phases. Each calibration phase contains 12 measurements and each testing phase contains 25 measurements. Each measurement lasts for 0.1 sec (4800 sample points).

      • Column 46-57645 : Phase "Calibration #1". Flattened from [12,4800] by row-major order.
      • Column 57646-115245 : Phase "Calibration #2". Flattened from [12,4800] by row-major order.
      • Column 115246-172845 : Phase "Calibration #3". Flattened from [12,4800] by row-major order.
      • Column 172846-292845 : Phase "Nasal breathing". Flattened from [25,4800] by row-major order.
      • Column 292846-412845 : Phase "Inhale #1". Flattened from [25,4800] by row-major order.
      • Column 412846-532845 : Phase "Inhale #2". Flattened from [25,4800] by row-major order.
      • Column 532846-652845 : Phase "Inhale #3". Flattened from [25,4800] by row-major order.
      • Column 652846-772845 : Phase "Exhale #1". Flattened from [25,4800] by row-major order.
      • Column 772846-892845 : Phase "Exhale #2". Flattened from [25,4800] by row-major order.
      • Column 892846-1012845 : Phase "Exhale #3". Flattened from [25,4800] by row-major order.

• Processed airway cross-sectional area (CSA) curves inferred from the acoustic signals

  • "aware_csa.csv"

    • NOTE: Not all raw samples are used for airway CSA reconstructions. Low-quality data is discarded, thus the sample size of this part is smaller than above

    • Column 0-45 : Basic subject information. Same as "aware_segmented.csv".

    • Column 46-129: Reconstructed airway cross-sectional area at different depths from mouth.

For more information about this dataset please contact: [email protected]

Citation

Use of this dataset in publications must be acknowledged by referencing the following publication:

@inproceedings{yin2023ptease,
    title={PTEase: Objective Airway Examination for Pulmonary Telemedicine using Commodity Smartphones},
    author={Yin, Xiangyu and Huang, Kai and Forno, Erick and Chen, Wei and Huang, Heng and Gao, Wei},
    booktitle={Proceedings of the 21st Annual International Conference on Mobile Systems, Applications and Services},
    pages={110--123},
    year={2023},
    url = {https://doi.org/10.1145/3581791.3596854},
    doi = {10.1145/3581791.3596854}
}

Other Related Publications:

@inproceedings{yin2022out,
    title={Out-Clinic Pulmonary Disease Evaluation via Acoustic Sensing and Multi-Task Learning on Commodity Smartphones},
    author={Yin, Xiangyu and Huang, Kai and Forno, Erick and Chen, Wei and Huang, Heng and Gao, Wei},
    booktitle={Proceedings of the 20th ACM Conference on Embedded Networked Sensor Systems},
    pages={1182--1188},
    year={2022},
    url = {https://doi.org/10.1145/3560905.3568437},
    doi = {10.1145/3560905.3568437}
}

License

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.