Accuracies were up to 84 percent for women and 77 percent for men using HTN threshold of SBP \u2265135 mmHg and DBP \u226585 mmHg<\/p>\n<\/h3>\n
By Elana Gotkine HealthDay Reporter<\/b><\/p>\n
<\/b><\/p>\n
FRIDAY, Sept. 13, 2024 (HealthDay News) — Predictive models for hypertension screening have been developed using speech recordings, with accuracies up to 84 percent for women and 77 percent for men, according to a study published online Sept. 10 in IEEE Access<\/em>.<\/p>\n Behrad Taghibeyglou, from the University of Toronto, and colleagues proposed a novel framework for detecting hypertension through acoustic analysis of speech. Speech was recorded across multiple sessions, and its temporal and spectral characteristics were analyzed with an aim of identifying indicators of hypertension. Two thresholds were explored for labeling individuals with hypertension: systolic blood pressure (SBP) \u00e2\u0089\u00a5135 mmHg or diastolic blood pressure (DBP) \u00e2\u0089\u00a585 mmHg and SBP \u00e2\u0089\u00a5140 mmHg or DBP \u00e2\u0089\u00a590 mmHg. The study included 245 participants (91 female).<\/p>\n The researchers developed predictive models for each gender; their performance was assessed using leave-one-subject-out validation. For the first threshold, the balanced accuracy achieved was 84 percent for women and 77 percent for men. The corresponding balanced accuracies for the second threshold were 63 percent for women and 86 percent for men.<\/p>\n “These insights pave the way for further development of generalizable, non-invasive, accessible, and cuffless hypertension screening methods utilizing speech analysis,” the authors wrote.<\/p>\n