Rethinking respiratory diagnostics and monitoring: From hardware to model-based therapeutics
| dc.contributor.author | Guy, Ella Frances Sophia | |
| dc.date.accessioned | 2025-02-04T22:42:03Z | |
| dc.date.available | 2025-02-04T22:42:03Z | |
| dc.date.issued | 2024 | |
| dc.description.abstract | Respiratory disease poses a large and increasing global burden, directly effecting approximately 450 million people. Respiratory disease can be described by combinations of obstructive, respiratory, and neuromuscular dysfunction modes. It follows that mechanically this disease can be described by mechanical airway resistance, lung elastance, or neuromuscular abnormalities. However, direct measurement of these variables is difficult and often impractical in spontaneously breathing (unsedated) patients, and has significant added economic and social costs. Mechanical respiratory parameters have been successfully identified in invasively ventilated, sedated, patients in intensive care units. However, in spontaneous, unsedated, patients, patient effort dominates inspiratory drive, and is difficult to elucidate from passive elastic and restrictive mechanics. The ability to establish an identifiable model of respiratory function in terms of airway resistance, lung elastance, and neuromuscular contributions could be expected to enable faster, more accurate patient-specific diagnoses, testing, and treatment adjustment, but only if it does not require added sensors, measurements, or cost. Overall, this thesis presents developments towards an identifiable and clinically informative model of respiratory function. With the aim of facilitating more frequent respiratory testing to establish patient-specific baseline functions and track progressions with time, without requiring clinical visits and multiple intrusive tests for each data point. Hardware was developed to collect appropriate data for model identification, using simple methods which could be easily followed and thus applied outside clinical settings without specialist operation. In addition, a number of small-scale low-risk trials, with a combined total 160 subjects, were conducted to collect spontaneous breathing data for model and testing methodology development. The first section of this thesis centred on the investigation of appropriate methods of identifying active respiratory mechanics. First, active muscular components of spontaneous breathing were described by scaled second order b-splines within a linear single compartment model framework. These methods were adapted from a technique applied to Neurally Adjusted Ventilatory Assist invasive mechanical ventilation data. Significant parameter trade-off was observed when applying this method in spontaneously breathing subjects. Thus, a fixed literature based resistance value was used and elastance was identified from expiratory data and extrapolated to inspiration. However, the accuracy of identified patient drive was consequently highly dependent on the accuracy of estimated elastances and resistances. Subsequently, the identified patient muscular effort profiles were used to assess a subject’s muscular contributions to work of breathing. Patient-ventilator interactions were then investigated using patient driving pressure relative to positive airway pressure therapy. These modelling efforts provide a foundation for positive airway pressure therapy titration to optimise subject-specific work of breathing profiles, and thus to optimise care. Another practical difficulty in respiratory assessment is differentiation of the symptoms of disease from disordered breathing patterns. Thus, investigation of abdominal and thoracic contributions to respiration were made using externally located dynamic circumference tape measures created specifically for this research, and which are very low cost and simple to use. A key outcome was the graphical differentiation of laboured from resting breathing modes by comparison of abdomino-thoracic pattern between inspiration and expiration. In future, clinical testing of neuromuscular disease could augment these results and prove further differentiation of muscular recruitment modes and their underlying causes. The active respiratory mechanics investigation portion of this thesis resulted in a promising model framework and increased understanding of patient-ventilator mechanics in NIMV, as well as an understanding of variations and trends in breathing mode. However, the identification of active mechanics was highly-dependent on the accuracy of estimated resistance and elastance values from the pulmonary mechanics models. Thus, the second portion of the thesis targeted methods of improving the accuracy of elastance and resistance identification to reduce parameter trade-off and improve clinical applicability and efficacy. Elastances and resistances were fit to rapid expiratory occlusion instances, adapted from interrupter technique methods, using a low-cost shuttering hardware system developed specifically for this research. Occlusion-based elastance and resistance values were identified in mechanical lung bench testing data and in subject spontaneous cases. In both cases, identified elastances and resistances were within expected ranges and were more reliable than compared methods, and performed better with the application of NIMV. The combined active and passive respiratory mechanics identification provides a comprehensive model-based respiratory assessment framework. The combined model was applied to a small spontaneous breathing dataset to illustrate the combined methodology. The combined comprehensive model-based respiratory assessment framework has many associated clinical implications which have the potential to improve patient-specific care across diagnostic, monitoring, and treatment of disease. Overall, the model-based methods and corresponding high-function, low-cost hardware developed in this thesis have shown the ability differentiate key parameters of respiratory function breath-wise in spontaneous breathing data. The identified parameters are closely related to physiological function (and dysfunction), and due to the simplistic physiological base model framework, they are clinically relevant and identifiable. These outcomes demonstrate the clinical utility of these methods and potential to guide and improve patient-specific respiratory care, by providing patients and clinicians with clear metrics, which can be obtained at greater frequency with lower burden to patients and healthcare systems. | |
| dc.identifier.uri | https://hdl.handle.net/10092/108050 | |
| dc.identifier.uri | https://doi.org/10.26021/15640 | |
| dc.language | English | |
| dc.language.iso | en | |
| dc.rights | All Right Reserved | |
| dc.rights.uri | https://canterbury.libguides.com/rights/theses | |
| dc.title | Rethinking respiratory diagnostics and monitoring: From hardware to model-based therapeutics | |
| dc.type | Theses / Dissertations | |
| thesis.degree.discipline | Mechanical Engineering | |
| thesis.degree.grantor | University of Canterbury | |
| thesis.degree.level | Doctoral | |
| thesis.degree.name | Doctor of Philosophy | |
| uc.bibnumber | in1405429 | |
| uc.college | Faculty of Engineering |