The use of Symmetric Projection Attractor Reconstruction (SPAR) as a novel assessment tool in Asthma

最佳的哮喘护理依赖于对患者临床状况的准确和详细的评估。评估的基石是肺功能测试。这类测试经常涉及在最大呼吸努力期间测量口腔内的气流。虽然被广泛使用，但这种测量相对不敏感，而且往往与总体疾病严重程度关系不大。缺乏敏感性导致难以诊断早期肺部疾病或恶化后的早期干预，这可能最终影响预后。现有的技术也可能是具有挑战性的患者执行，导致不准确或难以解释的测量和潜在的误诊。对称投影吸引子重建(SPAR)是一种新的分析工具，量化生理波形的变化，如呼吸流量，并将高保真波形数据转换为更简单的、可量化的图像(称为吸引子)。反映哮喘时发生的肺力学变化影响的呼吸流波形、形状和变异性的微小变化被吸引器中的较大变化所吸收。这意味着常规测试可能遗漏的病理生理变化(例如，PFT中使用的简单指标)可以被量化，从而更好地利用所有可用的数据。这可能导致对病变和/或恶化的更敏感的识别。当系统跟踪时间序列波形数据时，会形成重叠的吸引子，使我们能够绘制随时间变化的生理变化图。波形的形态和可变性的微小变化将分别导致吸引子的形状和颜色的具体变化。每个吸引子与创建它的输入波形数据具有唯一的关系。因此，不同类型的波形存在不同的吸引子形态和变异性特征，反映了潜在的表型。本研究的目的是为SPAR分析作为哮喘患者临床决策支持工具的概念提供证据。这是一项详细的、观察性的临床生理学研究，目的是评估SPAR在确定哮喘严重程度和对诱导的气道口径改变的反应中的使用。在一大群健康受试者(n=81)和哮喘患者(n=152，n=76轻度，n=76中度/重度)和在呼吸道口径减少或增加的情况下，记录呼吸流量波形。最初的健康受试者数据将通过试点研究得出的呼吸波形SPAR编码进行处理，以生成相应的、依赖于时间的攻击者。这将使关键的Spar特征能够被识别，并在计算机建模中将Spar特征与生理意义联系起来。随着健康受试者和患者数据的收集，将开发特定的呼吸系统编码来量化特定的呼吸系统特征，并开发、优化和测试机器学习算法，以分类不同的疾病表型以及对支气管收缩和支气管扩张的反应，以反映疾病控制的急剧变化。这一项目最重要的成果将是一种诊断和监测哮喘的新方法，这将有可能显著改善相当一部分人口的生活质量。这项研究涉及一个由生物医学科学家、临床医生和数学家组成的多学科研究小组，从而避免了潜在的过时技术或新方法的次优实施。SPAR有可能成为哮喘患者的诊断工具，特别是在儿童等难以评估的群体中，因为这些措施源于静息潮气呼吸，不涉及特定的呼吸动作，以及一种个性化的监测/警报工具，允许在病情恶化时进行早期干预，从而改善结果。

Optimal asthma care relies on accurate and detailed assessment of the patient's clinical condition. A cornerstone of assessment is pulmonary function testing. Such testing frequently involves measuring airflow at the mouth during maximal breathing efforts. Although widely employed, such measurements are relatively insensitive and often relate poorly to overall disease severity. Lack of sensitivity leads to difficulty diagnosing early lung disease or early intervention following deterioration, which may ultimately affect prognosis. Existing techniques can also be challenging to perform by patients resulting in inaccurate or difficult to interpret measures and potential misdiagnosis.Symmetric Projection Attractor Reconstruction (SPAR) is a novel analysis tool to quantify morphology changes from physiological waveforms such as respiratory flow and transforms high-fidelity waveform data into a simpler, quantifiable image (termed an attractor). Small changes in the respiratory flow waveform shape and variability reflecting the impact of changes in lung mechanics that occur in asthma are picked up as bigger changes in the attractor. This means that pathophysiological changes which may be missed with conventional testing (eg simple metrics used in PFT) can be quantified, making better use of all the available data. This could lead to more sensitive identification of pathological change and/or deterioration. As the system tracks through time series waveform data, overlapping attractors are formed, allowing us to map physiological changes over time. Small changes in the waveform's morphology and variability, will result in specific changes in the attractor's shape and colour, respectively. Every attractor bears a unique relationship with the input waveform data that created it. Therefore, different attractor morphology and variability features exist for different types of waveform which reflect the underlying phenotype.The aim of the proposed study is to show proof of concept for SPAR analysis as a clinical decision support tool in patients with asthma. This is a detailed, observational clinical physiological study with the aim of assessing the use of SPAR in determining asthma severity and response to induced change in airway calibre. Respiratory flow waveforms be recorded in a large cohort of healthy subjects (n=81) and asthmatic patients across a range disease severities (n=152, n=76 mild, n=76 moderate/severe) and under conditions of reduced or increased airway calibre. Initial healthy subject data will be processed through pilot study derived respiratory waveform SPAR coding to generate corresponding, time dependent attactors. This will enable key SPAR features to be identified and in silico modelling to associate SPAR features with physiological meaning. As healthy subject and patient data are collected, specific respiratory coding will be developed to quantify specific respiratory features and machine learning algorithms developed, optimised and tested to classify different disease phenotypes and the response to brochoconstriction and bronchodilation, reflecting acute change in disease control.The most significant outcome of this project will be a new approach for diagnosis and monitoring of asthma which has the potential of realising an appreciable improvement in quality of life for a significant proportion of the population. The study involves a multidisciplinary research group of biomedical scientists, clinicians and mathematicians thus avoiding potentially outdated technologies or suboptimal implementation of novel approaches. SPAR has the potential to be a diagnostic tool in patients with asthma, particularly in difficult to assess groups such as children, as measures are derived from resting tidal breathing and do not involve specific respiratory manoeuvres as well as an individualised monitoring/alert tool allowing early intervention at the onset of deterioration, thereby improving outcomes.