ecg monitoring system

ECG Monitoring System Advantage

The heart is one of the most complex organs in the human body. Heart disease is the most common ailment in the industrialized world. Despite the fact that the heart signals are considered one of the most studied in the world, this analysis process is still overly complex. Modern functional diagnostics have a large number of different instrumental research methods, but one of the most common, informative, and accessible of them is electrocardiography. This method of studying the bioelectrical activity of the heart, since its discovery by Einthoven, has been leading in the diagnosis of rhythm and conduction disorders, coronary heart disease, cardiomyopathy, and other diseases of the cardiovascular system.

One of the significant and obvious advantages of this research method is its safety. This allows you to repeat studies without restrictions, which is important not only for diagnosing the disease, but also for monitoring its course, monitoring treatment effectiveness, and predicting complications. In our time, algorithms for automatic processing and analysis of ECG have become widespread.

What is an ECG monitoring algorithm?

ECG processing or monitoring system is a software program that receives a cardiological signal at the input and, through a complex mathematical analysis of the data array, produces a result in the usual form for a doctor in the form of a set of indicators that characterize the work of the heart. The basis of such software are digital signal processing (DSP) algorithms, applied statistical analysis algorithms, multi-dimensional classification algorithms. ECG algorithms must meet certain requirements in order to be safe for patients and useful for physicians. Such algorithms must provide the required performance (accuracy) and are subject to mandatory certification. These algorithms must be sufficiently robust with respect to the quality of the input signal, since the ECG signal is characterized by a significant noise level in a wide frequency range. This problem is usually solved with complex digital filtering systems. As a result, an ECG algorithm typically produces: annotation of heart beats (including the positions of PQRST waves and their amplitudes, types of heart beats), heart rate, interval measurements, and rhythm events interpretation. One of the brightest representatives of the family of ECG algorithms is the The Kinetic™ ECG algorithm by Monebo.

The Kinetic™ ECG monitoring system

The Kinetic™ ECG algorithm created by Monebo is a software program for processing electrocardiographic signal of any length and number of leads from 1 to 16 which guarantees high accuracy of the result. Our unique processing algorithm consists of three parts: Preprocessor, Core, and Report generator.

The preprocessor provides pre-processing of the raw ECG signal including normalization and filtering. The core receives a filtered ECG signal as input. The main outputs are lists of detected beat and rhythm events. These lists are used to further form the output annotation of the algorithm. The Core of the algorithm itself also has a modular structure – Noise Analyzer, Beat Analyzer, QRS Type Analyzer, Average QRS Analyzer and Rhythms Analyzer (there will be a hyperlink to relevant pages on the site).

To achieve high performance indicators, modern machine learning methods based on automatic optimization are implemented. The operating mode of the algorithm core depends on a set of internal several hundred parameters. These parameters allow analyzers to adapt to the characteristics of the input signal and ultimately effect performance. In the general case, teaching an algorithm, or optimization, is the selection of the optimal set of Core parameters that provides maximum performance. Thus, the ECG analysis software achieves the best performance since the ECG algorithm is subjected to learning to for the selection of the optimal values ​​of the Core parameters.

Manual tuning and automatic learning

The process usually includes two steps: manual tuning and automatic learning. Manual tuning is the initial step before starting an automatic workout. This step involves coarse tuning of the parameters to reach a point in multidimensional space formed by the parameter vector, which can be used as the best initial approximation before starting automatic learning. Automatic learning is a supervised optimization procedure to find the global maximum of a response function in a multivariate factor space.

In this case, the multidimensional factor space is formed by the parameter vector. The performance of the algorithm is used as a response function. Algorithm performance is an integral metric that considers performance against various criteria, a set of which can be customized. This is an iterative process, which, in turn, is divided into separate stages. At each substage, different values ​​of the optimization hyperparameters and different bounded subsets of the factor space can be implemented.

Expert diagnosis of patients

The Kinetic™ ECG monitoring system carefully analyzes the ECG signal to provide highly accurate PQRST detection, heart rate, interval measurement, and rhythm interpretation for one to sixteen ECG leads, The Kinetic algorithm also measures intervals: PQ(R), QT, including corrected QT (QTc), the duration of the QRS complex. This wealth of information about the functioning of the heart helps medical professionals evaluate the performance of the heart and make expert diagnoses for patients.

To get this ECG monitoring system to work or to clarify the details of its work, please contact us using the request form.