Analysis system of electrocardiography rhythms and anomaly detection

Abstract

Today, there is a need in medicine for the intellectual analysis of medical data. Modern medical monitoring systems are obliged to carry out centralized monitoring of patients. Electrocardiogram (ECG) monitoring of patients is a difficult and unsolved task at the moment. Researches in this field are conducted by scientists from different universities and countries. William Buxt of the University of California, San Diego, used a neural network - a multilayer perceptron - to detect myocardial infarction in patients who are receiving rest in acute chest pain. The fact is that the neural network has been well aware of the noise of its weight in previous times and periods. And the fact that neural networks are very effective in predicting missing sequences, and reasonably showed the results and consequences of competing predictable programs conducted by the University of Santa Fe - neural networks were able to take first place and dominate among the best methods.
The above technologies do not exhaust the full range of application of such ECs. The purpose of the study was to improve the quality of diagnosis for patients with cardiovascular disease by developing an online system for the analysis of electrocardiogram rhythms to detect abnormalities in the cardiovascular system.
A typical ECG signal is a time series that is non-stationary and prone to numerous types of interference. The task of intelligent analysis of ECG rhythms in monitoring data systems helps in the processing and early detection of anomalies in humans for the diagnosis of cardiovascular problems. A database called MIT-BIH Arrhythmia was used to evaluate the ST and T tooth segment, which is often used in the detection of cases of myocardial infarction and for thorough studies of the dynamics of the heart. An integrated clustering system was developed for the analysis.
The following tasks were accomplished in the course of the research: analysis of existing methods and approaches to clustering of time series; the optimal algorithm for clustering data in the form of time series is selecteddesigned own system for detecting anomalies in ECG rhythms; analysis of data collapsing methods and approaches; developed own software for analysis of ECG rhythms and specialized database for it; experiments were conducted based on test and real data.