Inhalers are devices which deliver medication to the airways in the treatment of chronic respiratory diseases. When used correctly inhalers relieve and improve patients' symptoms. However, adherence to inhaler medication has been demonstrated to be poor, leading to reduced clinical outcomes, wasted medication, and higher healthcare costs. There is a clinical need for a system that can accurately monitor inhaler adherence as currently no method exists to evaluate how patients use their inhalers between clinic visits. This paper presents a method of automatically evaluating inhaler adherence through acoustic analysis of inhaler sounds. An acoustic monitoring device was employed to record the sounds patients produce while using a Diskus dry powder inhaler, in addition to the time and date patients use the inhaler. An algorithm was designed and developed to automatically detect inhaler events from the audio signals and provide feedback regarding patient adherence. The algorithm was evaluated on 407 audio files obtained from 12 community dwelling asthmatic patients. Results of the automatic classification were compared against two expert human raters. For patient data for whom the human raters Cohen's kappa agreement score was [Formula: see text], results indicated that the algorithm's accuracy was 83% in determining the correct inhaler technique score compared with the raters. This paper has several clinical implications as it demonstrates the feasibility of using acoustics to objectively monitor patient inhaler adherence and provide real-time personalized medical care for a chronic respiratory illness.
Trinity College Dublin ->