Blood Pressure Modeling using Statistical and Computational Intelligence Approaches

Prevalence of hypertension in recent years have raised questions about the shape of association between blood pressure (BP) and various factors such as alcohol consumption, age, physical activity level (PAL) to assess the overall cardiovascular risk. However, the risk complexity and the process Dynamics are difficult to analyze. The study presents modeling of Systolic Blood Pressure (SBP), Diastolic Blood Pressure (DBP) and Pulse pressure (PP) variations with biological parameters like age, pulse rate, alcohol addiction, and PAL by statistical multiple regression and Feed-forward Artificial Neural Network (ANN) approaches. Different algorithms likes Quasi-Newton, Gradient descent and Genetic Algorithm were used in training the ANN to improve the training performance and better optimization of the predicted values. The statistical analyses clearly indicate that the variation of SBP, DBP and PP with the parameters is statistically significant. ANN approaches give a much more flexible and non-linear model for prognosis and prediction of the blood pressure parameters than classical statistical algorithms. Moreover, evolutionary algorithms like Genetic Algorithms (GA) provide many advantages over usual training algorithms like Gradient descent or Quasi-Newton methods. Although all measures of blood pressure were strongly related to the variables, the analyses indicated that SBP was the best single predictor of cardiovascular events for the present study. More detailed study with varied parameters that influence BP is required for a exhaustive modeling.