Diagnostic Model of Umbilical Artery Blood Flow Using NC-PSO-SVM

The pathological diagnosis and analysis of umbilical artery blood flow signals are very important for prenatal fetal monitoring and fetal health status detection. In this study, we proposed Nonlinear Characteristics (NC)-Particle Swarm Optimization (PSO)-Support Vector Machines (SVM)intelligent diagnosis model of umbilical blood flow to solve the problem of low accuracy of pathological diagnosis of early umbilical blood flow caused by human experience. Firstly, the umbilical artery blood flow SVM diagnostic model was constructed with the Peak valley ratio of umbilical blood flow velocity (S/D), Pulsation Index (PI) and Resistance Index (RI) as the characteristic parameters, and the umbilical artery blood flow PSO-SVM diagnostic model was constructed with the non-linear characteristic matrix as the characteristic parameters. On this basis, according to the non-linear characteristics of the umbilical blood flow time signal, the NC-PSO-SVM intelligent diagnostic model is constructed with the PSO algorithm and SVM. The average accuracy of the 20 times of the diagnostic model is 77.5%, The results showed that the accuracy of NC-PSO-SVM intelligent diagnosis model was 9% higher than that of SVM diagnosis model and 5% higher than that of PSO-SVM diagnosis model due to the anti-noise and nonlinear expressiveness of chaotic sequences.