Development of an Intelligent Model for Cardiac Arrest Prediction using Radio Frequency Identification (RFID) And Machine Learning Algorithm

Abstract

Physiological instability or abnormal vital signs such as heart beat rate, respiratory rate or blood pressure suggests
altered physiology. It is well recognized that abnormal physiology is associated with adverse clinical outcomes.
The higher physiology deviates from normal, the higher the risk of mortality such as cardiac arrest. Most Patients
showed evidence of physiological abnormality prior to the event of the arrest. This research work is intended to
provide an improved model for the investigation and prediction of physiological characteristics of cardiac arrest
via prediction using Radio Frequency Identification (RFID) and Machine Learning techniques. The model was
developed using Bidirectional Long Short Term Memory with Conditional Random Field and a One Dimensional
Convolutional Neural Network. The model takes as input a vector of the heart disease dataset and produces as
output a classification of the prediction based on the analysed heart disease dataset. The accuracy of this research
work is rated 98.3%. It was also discovered that the use of an autoencoder to actively learn the dataset played an
important role in the outcome of the results.