A Novel Approach to Mortality Prediction of ICU Cardiovascular Patient Based on Fuzzy Logic Method

Objective: Health-care methods have an undeniable progress in protecting the health of patients. Risk assessment is very valuable for important clinical events in evaluating new treatments, controlling resource consumption, and improving the quality control of Intensive Care Unit (ICU). Methods: To predict the patient's future status, the phenomenon and related physiological effects should be accurately identified. The purpose of this paper is to predict the mortality of cardiovascular patients in the intensive care unit. So far, several methods have been proposed for prediction in this regard, each of which has been able to predict mortality somewhat, but many of these methods require the registration of a large number of patients, which in most cases is not possible to record all data, while this paper focuses only on Heart Rate Variability (HRV). In this paper, using the information obtained from the electrocardiogram (ECG) signal and by means of biological signal processing techniques, the changes in these signals will be initially investigated during the hospitalization period and then predicting the mortality rate of cardiovascular patients through the extracted features of the return mapping generated from the heart rate variability signal will be discussed. For the purpose of this paper, 80 recorded data from patients with coronary artery disease who were admitted to the intensive care unit, were collected. Given the fact that the return mappings detect the hidden phenomenon within the biological signal, a two-dimensional return mapping of the HRV signal was reconstructed in the first step of the quantification process. Then, the extraction of the traditional features of this mapping was dealt with, because these features were not able to provide important information about how the patient status was at the time of death. In the following, new features of return mapping were introduced and it was shown that the use of these features can provide a better distinction at the time the patient goes to death. Results: The results demonstrate that using the new method presented in this paper is comparable with other methods or leads to better results. The specificity and sensitivity of the new features were 92.54% and 91.73% respectively for coronary artery disease with a prediction horizon of 0.5 h before death. The higher the prediction horizon, the nurses or patient's relatives have more opportunity to take therapeutic measures, the time interval found in this paper is 0.5 and 1 h before the death. In order to determine the patient's future status and ECG signal analysis, the minimum requirement was 9.2 h for hospitalization, which had a significant reduction compared with other methods. Conclusion: As a result, the implementation of these methods leads to more favorable treatment of patients and the allocation of appropriate equipment to them. Another of these aims is to reduce costs, avoid unnecessary treatments and reduce the risk for patients. (C) 2018 Elsevier Ltd. All rights reserved.