Clinical Decision Support for the Assessment of Syncope and Falls: Novel Applications of Cerebral Perfusion Measurement and Big Data Analytics
File Type:
PDFItem Type:
ThesisDate:
2023Author:
Access:
embargoedAccessEmbargo End Date:
2025-04-23Citation:
Perez Denia, Laura, Clinical Decision Support for the Assessment of Syncope and Falls: Novel Applications of Cerebral Perfusion Measurement and Big Data Analytics, Trinity College Dublin, School of Medicine, CentreFor Medical Gerontology, 2023Abstract:
Syncope, defined as a transient loss of consciousness due to cerebral hypoperfusion, is common in both young and old, with syncope an attributable cause of unexplained falls (UF) in older adults. Both syncope and falls have significant negative physical and psychosocial consequences for an individual, with high associated healthcare costs ($2.4 billion/year) at a societal level. Identification of causative and modifiable risk factors for syncope and falls is therefore paramount.
International guidelines recommend the Active Stand (AS) test with concurrent monitoring of continuous blood pressure (BP) and heart rate as part of a comprehensive clinical assessment. Despite the presumed causal role of cerebral hypoperfusion in syncope and UF, no direct measure of cerebral perfusion is currently used in clinical assessment. Near-infrared spectroscopy (NIRS) has emerged in recent years as a candidate tool for this purpose, enabling easy, continuous monitoring of cerebral oxygenation during dynamic testing.
In this thesis, four studies were developed to investigate the clinical utility of applying NIRS in this context, with the overarching goal of developing explainable AI/ML based clinical decision support tools to assist in the management of these complex patients. Firstly, NIRS was integrated into standard patient assessments, enabling the collection of cerebral oxygenation signals during the AS from a large cohort of consecutive patients attending a national Falls and Syncope Unit. This dataset was then integrated with population data from The Irish Longitudinal Study on Ageing (TILDA) resulting in a unique harmonised database of clinical information, and peripheral and cerebral circulation measures. A software framework was developed to automate signal processing, feature extraction, statistical analysis and fitting of artificial intelligence/machine learning (AI/ML) models to this high dimensional data for predictive modelling.
Novel findings suggest that both age and sex strongly modulate cerebral oxygenation responses to standing. In young patients with syncope, peripheral cardiovascular hypersensitivity with unaltered cerebral oxygenation responses to an AS were observed, suggesting a peripheral origin of syncope, with no alterations in cerebral function evident. In older adults, multimorbidity/frailty was strongly associated with impairments in cerebral oxygenation independent of BP, suggesting that biological ageing is associated with repeated periods of hypoperfusion, increasing vulnerability of these individuals and potentially accelerating their rate of biological ageing. Lastly, novel explainable AI/ML models were developed for the first time, combining multimorbidity, BP and cerebral oxygenation to identify patients with UF and syncope with high accuracy (~75%), enabling the development of future clinical decision support tools to assist in the management of these patients. These explainable AI/ML models also provide novel insights into possible physiological mechanisms underlying syncope and UF, supporting the hypothesis that UF are central in origin, associated with more significant brain ageing and pathology, while syncope is likely present in those with younger intact brains with a peripheral origin. Overall, an independent future role for NIRS may be found in the assessment of brain ageing in older adults at risk of syncope and UF.
In conclusion, this work has furthered the scientific knowledge on orthostatic cerebral oxygenation in those with syncope and UF, while also highlighting the role of multimorbidity in these responses. Importantly, this thesis demonstrates the clinical value of using NIRS in the assessment of patients with syncope and falls. Additionally, it contributes to the development of a future AI/ML based clinical decision support tool to assist clinicians in both community and specialised clinical settings, aiding in the early identification of patients at risk of falls and syncope. This tool has the potential to assist in the design of appropriate personalised interventions to prevent the occurrence of such episodes, and ultimately improving a patient?s quality of life.
Sponsor
Grant Number
Fundaci? Universit?ria Agust? Pedro i Pons - Universitat de Barcelona - Barcelona - Spain
Irish Research Council - Government of Ireland Postgraduate Scholarship Programme 2018 (Grant No
GOIPG/2018/134) - Dublin - Ireland
Author's Homepage:
https://tcdlocalportal.tcd.ie/pls/EnterApex/f?p=800:71:0::::P71_USERNAME:PREZDENLDescription:
APPROVED
Author: Perez Denia, Laura
Sponsor:
Fundaci? Universit?ria Agust? Pedro i Pons - Universitat de Barcelona - Barcelona - SpainIrish Research Council - Government of Ireland Postgraduate Scholarship Programme 2018 (Grant No
GOIPG/2018/134) - Dublin - Ireland
Advisor:
Finucane, CiaranPublisher:
Trinity College Dublin. School of Medicine. CentreFor Medical GerontologyType of material:
ThesisCollections
Availability:
Full text availableMetadata
Show full item recordLicences: