A short recorded pulse dataset for vascular age prediction in China

Early assessment of cardiovascular disease risk plays an important role in preventing cardiovascular disease, vascular age (VA) is an important indicator for early screening of cardiovascular disease risk. This study presents a pulse signal-based dataset for VA prediction. The dataset comprises 226 subjects with 1364 pulse cycles, spanning both sexes (49.6% male, 50.4% female) and an age range of 20 to 69 years. Pulse signals were denoised by Savitzky-Golay filters, and 4th-order derivatives were calculated to extract the features of pulse signal. We applied the classic statistical model Klemera Doubal method (KDM) and five artificial intelligence models to predict VA. The experimental results showed that these models can predict VA with high accuracy and stability. It indicates that using pulse signals to predict VA is a simple, non-invasive, and effective method for assessing vascular health.