70COMBINING THE ACTIVE STAND TEST AND PATTERN RECOGNITION ENABLES VASOVAGAL SYNCOPE PREDICTION. (October 2014)

• Record Type: Journal Article
• Authors: Carmody, M.; Nolan, H.; Fan, C. W.; O'Dwyer, C.; Kenny, R. A.; Finucane, C.
• Is Part Of: Age and ageing. Volume 43(2014)Supplement 2
• Page Start: ii21
• Page End: ii21
• Publication Date: 2014-10
• Subjects: Aging -- Periodicals; Geriatrics -- Periodicals; 618.97
• DOI: 10.1093/ageing/afu133.1
• Languages: English
• ISSNs: 0002-0729

Abstract

Abstract : Introduction: Vasovagal syncope (VVS) is the most common form of syncope, accounting for 50-60% of unexplained syncope. The gold standard of diagnosis is the Head-Up Tilt (HUT) test, a resource intensive procedure. This study aims to assess the accuracy of applying a pattern recognition methodology to predicting HUT outcome based on AS responses. Methods: Continuous blood pressure records obtained during an AS were acquired from patients attending a Falls and Blackout Unit. Patients were categorized into 3 groups based on their clinical history and HUT response: controls (CON), tilt-positive (HUT+) and tilt-negative (HUT-). Data from subjects diagnosed with VVS i.e. HUT+ and HUT- were combined to form a vasovagal positive (VVS+) group. Hemodynamic features (n = 33) were extracted from AS responses and entered into a linear discriminant classifier. Classifier training and accuracy was achieved using an N-fold cross validation procedure. Results: N = 101 patients were recruited (25 ± 9 years; 66% male) of whom 37 were CON, 30 were HUT- and 34 were HUT+. Maximum prediction accuracy of HUT response was 60.9% (range: 58.2-60.9%), with a sensitivity of 58.8% and specificity of 63.3%. A multivariate classifier enabled us to distinguish between VVS+ and CON with a maximum accuracy of 80.2% (range: 76.4-80.2%), sensitivity of 84.3% and specificity of 72.9%. Conclusion: This study highlights the existence of an alternative hemodynamic response to an AS test exhibited by young patients prone to VVS. Based on these responses, it was possible to identify the presence of VVS in younger people, using multi-physiological-parameter classification approaches from active stand, with an accuracy of 80% - a potential improvement on the HUT classification accuracy reported in literature (26% to 87%). With prospective verification, this approach may form the basis of a novel tool for syncope diagnosis, population studies and the tracking of treatment efficacy.