Information technologies and computer sciences

Application of Machine Learning Methods for Evaluating Systemic Vascular Resistance Based on Non-Invasive Hemodynamic Parameters

Authors

O. M. Bryk
Ternopil Ivan Pului National Technical University ROR
O. A. Pastukh
Ternopil Ivan Pului National Technical University ROR
DOI https://doi.org/10.31649/1997-9266-2025-183-6-113-118

Keywords

systemic vascular resistance hemodynamics non-invasive parameters machine learning classification

Abstract

The article presents the results of a study aimed at applying machine learning methods for the non-invasive assessment of systemic vascular resistance (SVR), which is an important integral indicator of peripheral circulation and is widely used in clinical practice for the diagnosis of cardiovascular pathologies. Hemodynamic signals, in particular the pulse wave and related physiological parameters, are characterized by a combination of pronounced rhythmicity and stochastic variations caused by regulatory mechanisms of the cardiovascular system and the influence of external factors. The relevance of this study is due to the fact that traditional methods for determining SVR, such as linear deterministic models, systems of differential equations, spectral and wavelet analysis, are based on invasive procedures related to cardiac output and mean arterial pressure. These methods require specialized equipment, highly qualified personnel, and are associated with risks for patients, which limits their application in routine clinical practice and screening studies and increases the need for safe, accessible, and automated assessment methods.

In this work, a mathematical formalization of hemodynamic signals based on cyclic random processes is proposed, which serves as a theoretical basis for the application of machine learning methods and allows simultaneous description of periodic components of the cardiac cycle and random signal fluctuations. This approach is based on the use of non-invasive hemodynamic parameters, including heart rate, arterial blood pressure, morphological characteristics of the pulse wave, and heart rate variability indices. To improve model quality, data preprocessing was performed using normalization, multicollinearity analysis (VIF), dimensionality reduction by principal component analysis (PCA), and heteroscedasticity testing. Modern machine learning algorithms were applied to build the models, including logistic regression, the k-nearest neighbors method, support vector machines, and random forest. It is shown that the use of cyclic random processes provides a basis for the formation of more informative and physiologically meaningful features suitable for classification and predictive modeling.

A generalized block diagram of hemodynamic signal modeling was developed, covering the stages of preprocessing, mathematical modeling, informative feature extraction, and classification or prediction of hemodynamic states. The obtained results confirm the feasibility of using cyclic random processes as a fundamental tool for constructing hybrid analysis systems that combine mathematical modeling and artificial intelligence methods and can be applied in modern biomedical diagnostic and clinical decision support systems.

152 74

How to Cite

[1]
“Application of Machine Learning Methods for Evaluating Systemic Vascular Resistance Based on Non-Invasive Hemodynamic Parameters”, Вісник ВПІ, no. 6, pp. 113–118, Dec. 2025, doi: 10.31649/1997-9266-2025-183-6-113-118.

Author Biographies

O. M. Bryk, Ternopil Ivan Pului National Technical University

Post-Graduate Student, of the Chair of Software

O. A. Pastukh, Ternopil Ivan Pului National Technical University

Dr Sc. (Eng.), Professor, Professor of the Chair of Software Engineering

References

National Health Service of Ukraine, Analytical reports on the state of population health. Kyiv, Ukraine, 2024.
W. W. Nichols, and M. F. O’Rourke, McDonald’s Blood Flow in Arteries: Theoretical, Experimental and Clinical Principles, 6th ed., Boca Raton, FL, USA: CRC Press, 2020, ISBN 978-1138741581.
S. Khandani, et al., “Machine Learning Applications in Cardiovascular Disease Prediction and Monitoring,” Journal of the American College of Cardiology, vol. 81, no. 10, pp. 1101-1115, 2023, https://doi.org/10.1016/j.jacc.2022.12.054 .
R. Singh, and R. Saini, “Application of machine learning in healthcare: Review and future,” Computer Methods and Programs in Biomedicine, vol. 236, Art. no. 107624, 2023, https://doi.org/10.1016/j.cmpb.2023.107624 .
K. W. Johnson, et al., “Artificial Intelligence in Cardiology,” Nature Reviews Cardiology, vol. 15, no. 7, pp. 411-429, Jul. 2018, https://doi.org/10.1038/s41569-018-0100-1 .
G. D. Clifford, et al., “Advanced methods for heart rate variability analysis,” Physiological Measurement, vol. 41, no. 8, Art. no. 08TR01, Aug. 2020, https://doi.org/10.1088/1361-6579/ab9b87 .
Z. I. Attia, et al., “An AI-enabled ECG algorithm for atrial fibrillation,” The Lancet, vol. 394, no. 10201, pp. 861-867, Aug. 2019, https://doi.org/10.1016/S0140-6736(19)31721-0 .
E. Yavorska, O. Dozorska, and V. Nykytyuk, “Wavelet-based preprocessing of biomedical signals,” in Proceedings of the International Conference on Computer Science and Information Technologies (CSIT), Lviv, Ukraine, 2020, pp. 188-192.
V. Dozorskyi, L. Dediv, and M. Khvostivskyi, “Methods of stochastic modeling in biomedical engineering,” Visnyk NTUU KPI. Series Radiotechnics, no. 82, pp. 56-64, 2021.
J. Pan, and W. J. Tompkins, “A real-time QRS detection algorithm,” IEEE Transactions on Biomedical Engineering, vol. BME-32, no. 3, pp. 230-236, Mar. 1985, https://doi.org/10.1109/TBME.1985.325532 .