Artificial Intelligence and Machine Learning

On the system of predictive criteria for cardiac arrhythmias in newborns based on data analysis

D. V. Akishev, E. N. Tiagusheva, A. O. Syromyasov, T. I. Vlasova, T. E. Badokina

Ogarev Mordovia State University, Saransk

Abstract: The paper examines machine-learning algorithms for detecting cardiac arrhytmias of newborns. Main cardiography intervals are analized. Medical histories containing indicators of newborn children for the first, third and the tenth days of life act as the source of data. The authors pay main attention to statistical analysis of this data and to construction of regression models; the results of such analysis are examined as well. Application of constructed models together with calculation of the main metrics shows that explanatory quality of regression algorithms in general is not enough good. Besides that, MARS has more predictive power than other models if the volume of sample group is small enough. To improve diagnostics and to make prevention of cardiac arrhytmias more effective it is necessary to provide further optimization, to enlarge training sample and to increase data diversity.

Keywords: artificial intelligence in pediatrics, prediction of cardiac arrhythmias, machine learning, regression modeling, newborns.

UDC: 519.25:616.127

Received: 24.03.2025
Revised: 25.05.2025
Accepted: 06.10.2025

DOI: 10.26456/vtpmk736

Bibliographic databases:

•