Abstract:

The goal of this study is to develop a clinical prediction model for predicting evolution of heart failure parameters acquired experimentally in research applied impedance cardiography. A proper Markov model for thoracic fluid content investigations has been formulated. The presented method is effective, in spite of limited measurements and irregular intervals between measurements for individual patients. It turned out that some important assumptions have to be made to solve this task. All of them are presented herein. Those assumptions allow for prediction in relatively small groups of patients, as it reduces the amount of data needed to approximate model parameters, while maintaining a large number of states. In this article, we are also submitting an algorithm to solve the proposed optimization problem. Parameters and accuracy of the model were determined basing on measurements made using data fusion impedance cardiogram and electrocardiogram. The presented method is successful in predictions for a patient’s medical evolution condition.