Introduction & Objective: Preventing type 2 diabetes and managing its complications is very important because of the many complications it causes and the high cost of treating it. Metabolic syndrome (MetS) and impaired fasting glucose (IFG) represent interrelated risk factors for developing type 2 diabetes and cardiovascular diseases. Alterations in the cardiovascular system can affect the characteristics of peripheral blood vessels in MetS and IFG. The present study aimed to use only noninvasive arterial pulse signals and combine machine-learning-based and self-developed pulse-distribution analysis (PDA) to discriminate the MetS and IFG-induced effects on vascular properties.
Methods: Divided into three groups categorized according to the five MetS factors (Control, Pre-MetS, and MetS), 113 participants receiving geriatric health checkups with standard medical inspection at Tri-Service General Hospital (TSGH) in Taiwan were enrolled in this study. The participants' radial blood pressure waveform (BPW) signal was measured, to which frequency-domain analysis (FDA) was applied. Two classification methods were used and the features of each pulse signal were collected from 40 spectral indices.
Results: BPW signal differed significantly among the three groups. Several pulse waveform indices were significantly lower in Pre-MetS and MetS than in Control. The AUC of PDA (0.83) was better than that of machine learning (AUC=0.66) in discriminating MetS. IFG and MetS factors were verified as having no obvious interference effect on the PDA classification.
Conclusion: Using pulse indices without other clinical parameters performs well discrimination in detecting MetS and IFG. The present findings facilitate a noninvasive, easy-to-use, and objective evaluation technique for the vascular properties, and could hence aid in reducing the cardiovascular risk associated with MetS and IFG.
L. Wu: None.
