Early Diagnoses of Chronic Heart Failure using Neural Network Classifier of Tensiometric Blood Test Results

• Published in: 2022 International Conference on Data Science and Intelligent Computing (ICDSIC) pp. 181 - 185
• Main Authors: Alqezweeni, Mohie M., Gorbachenko, Vladimir I., Zenin, Oleg K., Gribkov, Dmitry N., Potapov, Vladimir V., Miltykh, Ilia
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-11-2022
• Subjects: blood surface tension; Cardiovascular diseases; classification quality metrics; feature engineering; generate synthetic data; Heart; heart failure; ischemic heart disease; neural network training; Neural networks; Reliability; Sociology; tensiometry; Training; Training data
• DOI: 10.1109/ICDSIC56987.2022.10076007

Cardiovascular pathology, including chronic heart failure, is the leading cause of disability and mortality in the population. The high incidence and particular severity of cardiovascular diseases require the development of new effective diagnostic tools, treatment, and methods for monitoring the effectiveness of treatment. Early diagnosis reduces the number of patients requiring hospital treatment, reduces the number of days off work, and reduces the likelihood of adverse outcomes. An analysis of the known methods for early diagnosis of cardiovascular insufficiency showed that the known methods are complex and do not provide high diagnostic reliability. For early diagnosis of cardiovascular insufficiency, it is proposed to use tensiometry methods that are more reliable and less complicated than known methods. The disadvantage of this diagnostic method is the difficulty of interpreting the results of tensiometric analyzes for diagnosticians. To improve the accuracy of interpreting the results of the study, it is proposed to use a neural network classifier. The data for network training was obtained using a tensiometer. The problem of an insufficient amount of training data is solved by generating synthetic data using a variational autoencoder. A fully connected neural network has been implemented, which made it possible to diagnose chronic heart failure with an accuracy of 98%.