Two Phase Reverse Neural Network Based Human Vital Sign Prediction System

The objective of this paper is to propose a two phase reverse neural network (NN) model to predict sequential human bio signal data for decreasing the risk of acute and likely fatal diseases and syndromes in patients (e.g., COPD and hypertension) based on a simulated learning algorithm. Our approach uses a two phase sequential neural network with back-propagation learning. The modeling is done using time-varying inputs such as heart rate, SpO₂, and blood pressure as measured and communicated by wireless devices. The output of the prediction system results into distinct classification levels for patient´s condition through sequential points. Such a prediction facilitates an early detection of exacerbating risk factors thereby saving a sudden death and other high risk diseases. The system is plausible to be used for the prevention and management of high risk diseases.