Development of audio-based signal processing methods to objectively assess patient adherence in respiratory medicine

Abstract

Over four million people die every year from chronic respiratory diseases such as asthma and chronic obstructive pulmonary disease (COPD). There is no cure currently available for these diseases; however, they may be controlled using inhaled medication which directly targets the airways to relieve symptoms. Medication is usually administered to asthma and COPD patients using inhaler devices. Patients are required to inhale through the inhaler device in order to deliver medication to the airways. Despite the proven clinical efficacy of inhaler devices, up to 80% of patients make critical user technique errors which significantly reduce the amount of medication delivered to the patient. Poor adherence to correct user technique is associated with increased hospitalisations and healthcare costs. Inhaler user technique is currently assessed using checklists from healthcare professionals during treatment. However, they are subjective and have been reported to generate inaccurate assessments of patient inhaler user technique. Therefore, there is an urgent clinical need for objective methods to assess patients? inhaler user technique during clinical consultation and remotely over the course of treatment. In this thesis, audio-based signal processing methods were developed to objectively assess patient inhaler user technique. Audio recordings of patients using inhaler devices were obtained using microphones attached and distant to different inhalers to simulate attachable and wearable monitoring devices. A range of temporal and spectral features were extracted from the inhaler audio recordings to automatically assess critical events of inhaler use. A new novel method of objectively assessing patient inhaler user technique is introduced for the first time. Furthermore, this thesis also investigates the potential of employing audio-based methods to monitor patients? clinical response to treatment, providing promising insights into new potential clinical measures in respiratory medicine. The central aim of this research was to investigate the clinical applicability of audio-based signal processing methods to assess critical steps of patient inhaler use. The main findings of the studies detailed in this thesis suggest that temporal and spectral audio-based features of inhaler inhalation sounds can be used to objectively estimate pertinent clinical measures of inhalation technique such as the peak inspiratory flow rate (PIFR), inspiratory capacity and ramp time of inhalation. Furthermore, there exists a relationship between audio-based measures of the PIFR of inhaler inhalations and lung function. This may allow healthcare professionals to intervene before the onset of an exacerbation. Moreover, these audio-based signal processing methods can accurately detect the release of medication during inhaler use. It was shown that the audio-based methods presented in this thesis provide more accurate assessment of patient inhaler user technique than standard subjective clinical checklists. This highlights the need to implement these new audio-based objective measures of user technique assessment into clinical practice. In conclusion, the original contribution to knowledge in this thesis lies in the development of audio-based signal processing methods to provide objective assessment of patient inhaler user technique. This thesis builds on previous research by advancing signal processing methods to assess user technique in a range of different inhalers and providing new clinical measures in the treatment of chronic respiratory diseases. The main findings of this thesis establish the acoustic properties of inhaler sounds and provide significant value for future research in the field of audio-based inhaler monitoring systems. The audio-based methods described in the studies in this thesis can provide healthcare professionals with more accurate objective information regarding patient inhaler user technique. This information can be used to give patients feedback regarding their inhaler user technique which can improve the clinical efficacy of inhaler medication, reduce healthcare costs and enhance precision medicine in the treatment of chronic respiratory diseases.