Abstract

Arpita Bhattacharjee and Arup Ratan Ray*

ABSTRACT

An accurate estimation of systolic and diastolic pressure from oscillometric pressure signal has a great influence in health monitoring and clinical applications. This paper proposes a model based predictive algorithm comprising a predictive Volterra model, predictive statistical model and Chi Square test for the goodness of fit. Volterra models are developed from the ADC value of the oscillometric pressure signal data. Recursive least square filter is used to compute the Volterra kernels from input-output data. Local models based on Volterra model are created for individual systolic and diastolic group. Predictive statistical model is developed based on the variables like age, weight, diet, illness profile, geographic location etc. The accuracy of the model is improved using Chi Square goodness of fit test. Training and test data set can be used to create and test the model respectively.