Functional Clustering of Continuous Cardiovascular and Brain Oxygenation Signals During an Active Stand Test in The Irish Longitudinal Study on Ageing (TILDA)

In this study, we performed K-shape functional clustering of two non-invasive continuous cardiovascular (systolic Blood Pressure: sBP) and brain oxygenation (left frontal lobe Tissue Saturation Index: TSI) times series signals, simultaneously collected 20 seconds before to 40 seconds after an active stand (AS) test in 2158 participants of The Irish Longitudinal Study on Ageing. Both signals were normalised and adjoined prior to clustering, and the gap statistic method was used to determine the optimum number of clusters. Eight clusters were obtained and characterised. Cluster 5 (n=287) and Cluster 1 (n=173) were independently predictive of not reporting dizziness post-AS or falls after 8 years, and were characterised by complete (Cluster 5) or near-complete (Cluster 1) sBP recovery and a moderate/transient (Cluster 5) or large/sustained (Cluster 1) TSI overshoot. Cluster 3 (n=304) and Cluster 6 (n=326) were independently predictive of post-AS dizziness and had large sBP recovery deficit/lack of TSI overshoot (Cluster 3) or moderate sBP recovery deficit/TSI overshoot (Cluster 6). Cluster 8 (n=289) was the most multimorbid and showed both sBP and TSI recovery deficits. This data-driven approach produced clinically meaningful results and may help clinicians recognise combined neurovascular signatures of orthostatic regulation in older people, both in health and disease.