This thesis presents a comprehensive study on the topic of nonlinear respiratory airway resistance and breathing effort estimation for respiratory disease monitoring and care. This section provides a concise overview of the main contributions of this study, insights gained, and discusses their potential impact on the field of respiratory disease monitoring and care using model-based, patient-specific methods.

Presently, spirometry is widely regarded as the foremost standard for diagnosing COPD, with a focus on identifying a reduction in lung function by measuring the FEV₁/vital capacity (VC) ratio and employing methods based on linear airway resistance. However, incorporating nonlinear resistance modelling may offer valuable insights into an individual patient’s condition and the progression of the disease. Analysis of plethysmographic patient data with nonlinear modelling successfully identifies flow limited COPD patients with a nonlinear expiratory resistance in Chapter 5. Nonlinear modelling can provide novel insight into COPD patient-specific lung mechanics and potentially provide patient-specific, model-based metrics suitable for regular monitoring of COPD and lung status; and thus the ability to track disease state/progression.

Chapter 6 introduces a novel approach to estimate work of breathing (WOB) and elastance (E) from plethysmographic data, aimed at developing a diagnostic and monitoring tool for COPD patients. The proposed method employs nonlinear modelling. There are novel further induced metrics using patient specific models which can be used to track disease and, using work of breathing, are directly related to patient specific feelings of disease management, as well as to critical or/and key clinical metrics such as WOB.

While expiration is typically regarded as a passive process, some degree of expiratory flow is generated through patient effort, referred to as expiratory effort. Like expiratory resistance, expiratory effort is nonlinear in nature and requires modelling using nonlinear methods. In Chapter 7, a successful approach is presented for identifying nonlinear expiratory effort in flow-limited COPD patients, utilising nonlinear modelling techniques. These findings, combined with methods for estimating nonlinear resistance, may offer a valuable alternative for identifying COPD patients without the need for spirometry. Expiration is not a passive process in disease states like COPD and FL COPD, which means modelling must be able to account for this issue, which is typical of all lung mechanics model assumptions, and thus typical of models developed for healthy subjects which in turn are not optimal for dysfunction.

Chapter 8 investigates the presence of nonlinear resistance in expiration for patients who have undergone mechanical ventilation (MV). A large subset of patients undergoing MV have underlying COPD. The aim of identifying this nonlinear expiratory resistance in MV patients is to improve optimisation of MV through model-based methods, reducing ventilator induced injury, and improving care. The expected result of translating methods from Chapter 5 to Chapter 8 was not realised without the quantification of inspiratory effort. There are potential areas where the key results translate to in hospital fully MV patients, but they require further modelling for some MV modes. However, the results can be generalised or have the potential to be generalised. This outcome contributes toward the aim of a single useful model in all or most all dysfunction scenarios.

Chapter 9 presents three attempts at quantifying non-invasive, model-based estimation of patient effort in the MV patient data using esophageal pressure measurements. Each of the three methods had their respective trade-offs. While none of the methods demonstrated an ideal solution across all patients in the data, the second analysis provided the best overall approach. The models and methods presented can be translated to MV patients and initial invasive validation efforts showed good, though not perfect correlation where errors are attributable to model errors or assumptions made to ensure models remain identifiable and thus clinically useful. These final results show the trade-off of modelling efforts for accuracy with maintaining identifiability to ensure clinical utility.