Estimation of inhalation flow profile using audio-based methods to assess inhaler medication adherence

Area of Science:

• Respiratory Medicine
• Biomedical Engineering
• Acoustic Signal Processing

Background:

• Patients with asthma and chronic obstructive pulmonary disease (COPD) require proper inhalation technique with dry powder inhalers (DPIs) for effective medication delivery.
• Objective monitoring of patient inhalation technique is crucial for remote patient management and adherence.
• Previous audio-based methods for estimating inhalation flow profiles required multiple calibration recordings, limiting their clinical applicability.

Purpose of the Study:

• To develop and validate an audio-based method for accurately estimating DPI inhalation flow profiles using a single calibration recording.
• To assess the accuracy of power law and linear regression models in correlating acoustic envelopes with inhalation flow signals.
• To evaluate the potential clinical utility of this method for inhaler technique training and remote monitoring.

Main Methods:

• Twenty healthy participants performed inhalations through a placebo Ellipta™ DPI at various flow rates.
• Inhalation flow was measured using a pneumotachograph spirometer, and audio signals were captured simultaneously using an Inhaler Compliance Assessment device.
• Acoustic envelopes were extracted from audio signals, and power law and linear regression models were used to estimate flow profiles based on single calibration recordings.

Main Results:

• Power law regression models demonstrated significantly higher average flow estimation accuracy (90.89±0.9%) compared to linear models (76.63±2.38%).
• The method accurately estimated key inhalation parameters, including peak inspiratory flow rate and inspiratory capacity, even in the presence of noise.
• The single-calibration audio-based approach proved effective in characterizing inhalation flow profiles across participants.

Conclusions:

• An audio-based method utilizing a single calibration recording can accurately estimate DPI inhalation flow profiles.
• Power law models offer superior accuracy for flow profile estimation compared to linear models.
• This technology holds promise for improving inhaler technique training and enabling remote patient monitoring in respiratory care.