CC BY
1Информационно-вычислительные технологии, искусственный интеллект и обработка больших данных в медицине
SCD RISK STRATIFICATION IN PATIENTS WITH HYPERTROPHIC CARDIOMYOPATHY USING
MACHINE-LEARNING
Dima M.-T.1, Dima M.1, Dima S.N.2
1Hyperion University of Bucharest, Bucharest, Romania 2Privolzhsky Research Medical University, Nizhny Novgorod, Russia Scientific advisors: Mihailescu M., lect. univ. dr., Dima M.-O., Ph.D., prof.
Research Relevance. Machine learning is becoming an important component of clinical medicine, particularly In risk assessment - and to varying degrees in diagnosis also. Compared to "classical" risk assessing methods for sudden cardiac death (SCD) in patients with hypertrophic cardiomyopathy (HCM) - such as HCM Risk-SCD, that use just a few clinical variables, machine learning (ML) takes into consideration a significantly wider range of variables. Goal: study of risk stratification of SCD in patients with HCM using ML.
Materials and Methods. Analysis of bibliographic sources from PubMed, Scopus and Elsevier for the last 10 years. Results. Smole T. et al. developed the "HCM-RSS" model for risk stratification in patients with HCM based on a broad set of methods: random forests, boosting trees, support vector machine and neural networks (with clinical, genetic, imaging and laboratory data) [1]. The data reported better ventricular tachycardia prediction (which may be the cause of SCD), with an AUC of 0.90.
The study [2] - from 2021, also selected 4 ML approaches, death due to SCD being one of the outcomes. Data from 183 patients with HCM were analyzed using a reference model (based on logistic regression of 8 previously established clinical parameters) and an ML model with 20 predictors. The ML prediction accuracy was 85 %, compared to that of the reference model of just 73 %. Similar results were demonstrated by other studies [3, 4].
Conclusion. Currently, clinical data embodies also a large amount of laboratory data, the processing of which can be vast. ML is capable of analysing this data and presenting it in a human-decision format, having strong future potential in the analysis of data from different sources and in risk modeling.
References
1. Smole T. et al. A machine learning-based risk stratification model for ventricular tachycardia and heart failure in hypertrophic cardiomyopathy. Computers in biology and medicine. 2021 ;135:104648.
2. Kochav S. M. et al. Predicting the development of adverse cardiac events in patients with hypertrophic cardiomyopathy using machine learning. International Journal of Cardiology. 2021 ;327:117-124.
3. Bhattacharya M. et al. Identifying ventricular arrhythmias and their predictors by applying machine learning methods to electronic health records in patients with hypertrophic cardiomyopathy (HCM-VAr-risk model). The American journal of cardiology. 2019;123(10):1681 -1689.
4. Alis D. et al. Assessment of ventricular tachyarrhythmia in patients with hypertrophic cardiomyopathy with machine learning-based texture analysis of late gadolinium enhancement cardiac MRI. Diagnostic and Interventional Imaging. 2020;101(3):137-146.
ПРОБЛЕМА ФИЛЬТРАЦИИ МАТЕМАТИЧЕСКИМИ МЕТОДАМИ ЭЛЕКТРОКАРДИОСИГНАЛОВ С УСТРОЙСТВА ДЛЯ ПОСТОЯННОГО МОНИТОРИНГА БИОСИГНАЛОВ
Внуков Е.В.
Саратовский государственный технический университет, Саратов, Россия Научный руководитель: Барулина М.А., д-р физ.-мат. наук
Введение. Разработка и внедрение систем постоянного мониторинга физиологического состояния сердца человека для использования самим пациентом в домашних условиях в настоящее время является актуальной. Эта актуальность обусловлена тем, что смертность от сердечно-сосудистых заболеваний достигает 31 % от всех смертей в год в России, а догоспитальная смертность при остром инфаркте миокарда достигает 60 %. Большую часть этих смертей можно было бы предотвратить при постоянном мониторинге и своевременной медицинской помощи. Постоянный мониторинг может быть реализован с помощью носимого устройства для круглосуточного съема биосигналов.
