Inhalation Flow Profiles from Audio Recordings of Placebo Inhaler Use

Featuring audio and flow signal recordings from 20 healthy participants performing 15 inhalations each through a placebo dry powder inhaler. It was created to develop a method for estimating inhalation flow profiles using a single calibration audio recording. The study found power law regression models achieved an average flow estimation accuracy of 90.89%.

Use Cases

• Train regression models to estimate inhalation flow signals from the acoustic envelope of audio recordings.
• Analyze the relationship between inhalation audio amplitude and flow parameters like peak inspiratory flow rate.
• Compare the accuracy of linear versus power law models for predicting flow profiles from a single calibration recording.
• Develop methods for remote monitoring of inhaler technique by estimating inspiratory capacity from audio signals.

Strengths

• Includes 300 total inhalation recordings (20 participants x 15 inhalations each).
• Provides simultaneous recordings of inhalation audio signals and pneumotachograph flow signals.
• Documents a specific method achieving 90.89% average flow estimation accuracy with power law models.

Limitations

• Data is from a controlled study with only 20 healthy participants, limiting generalizability to patient populations.
• Inhalations were performed using a placebo Ellipta DPI, which may not reflect real-world medication use.
• The dataset's size and specific format are unknown from the provided input.

Provenance

Source

Dryad digital repository.

Collection Method

Recordings from 20 healthy participants performing inhalations through a placebo inhaler, with simultaneous audio and pneumotachograph spirometer capture.

Freshness

Last updated in June 2020.