Medical Scientific Bulletin of Central Chernozemye (Naučno-medicinskij vestnik Centralʹnogo Černozemʹâ)Medical Scientific Bulletin of Central Chernozemye (Naučno-medicinskij vestnik Centralʹnogo Černozemʹâ)1990-472XФедеральное государственное бюджетное образовательное учреждение высшего образования "Воронежский государственный медицинский университет имени Н.Н. Бурденко" Министерства здравоохранения Российской Федерации1101610.18499/1990-472X-2025-26-3-91-99Diagnostic algorithms for detecting heartbeat abnormalities using machine learning methodsLyalikovaVictoria Gennadevna<p>Candidate of Physical and Mathematical Sciences, Associate Professor of the Department of Cybersecurity of Information Systems at the Faculty of PMM</p>vikalg@yandex.ruБезрядинMichael Mikhailovich<p>Candidate of Physical and Mathematical Sciences, Associate Professor at the Department of Mathematical Software of the Faculty of Applied Mathematics and Mechanics</p>maickel@yandex.ruMarkinaMaria Sergeevna<p>student of the Faculty of Law</p>maria.markina.03@mail.ruVoronezh State UniversityVoronezh Institute of the Ministry of Internal Affairs of the Russian Federation10112025263919902102025Copyright © 2025, Medical Scientific Bulletin of Central Chernozemye (Naučno-medicinskij vestnik Centralʹnogo Černozemʹâ)2025<p>Heart diseases consistently rank as the leading cause of death in developed countries, second only to COVID-19 in recent rankings. The development of diagnostic methods and risk assessment based on heart rate data plays an important role in research at the intersection of mathematics, data analysis, and medicine. The results of such research are being applied, including in the development of applications for mobile smartphones and wearable devices, as there is increasing focus on self-diagnosing heart conditions. The article analyzes heart tones collected by a stethoscope for diagnosing diseases caused by heart malfunction. The methods discussed in the study allow anyone to record heartbeats using a mobile device and use them for further analysis with various applications. The main task is to classify heart tones into 4 categories. The work was done in Python using libraries like tensorflow, librosa, pandas, and scipy. The results show that with the use of deep neural networks, a classification accuracy of 95% can be achieved on the test dataset. Additionally, this work can serve as a basis for further research, such as integrating the model with other layers of deep learning networks to reduce training time, which is important for using the results in mobile and wearable devices.</p>Classification of heart sounds, random forest, gradient boosting, convolutional neural networks, deep learning, machine learning, audio signal processing.классификация звуков сердца, случайный лес, градиентный бустинг, сверточные нейронные сети, глубокое обучение, машинное обучение, обработка аудиосигналов.1.1. Глобальные оценки здоровья: Основные причины смерти. Смертность по конкретным причинам, 2000–2021 гг. –Режим доступа: https://www.who.int/data/gho/data/themes/mortality-and-global-health-estimates/ghe-leading-causes-of-death - (дата обращения 15.11.2024).2.2. Chen W. Deep Learning Methods for Heart Sounds Classification: A Systematic Review// W. Chen, Q. Sun, X. Chen, G. Xie, H. Wu / Entropy - Basel, 2021, vol. 23(6) – pp.667-685. doi: 10.3390/e23060667.3.3. Zhang, W., Han J., Deng S. Heart sound classification based on scaled spectrogram and tensor decomposition // W. Zhang, J.Han, S.Deng / Expert Systems with Applications – Pergamon, 2017, Vol. 84, pp - 220-231. Doi: 10.1016/j.eswa.2017.05.014.4.4. J.Chen, X. Dang, and M.Li. Heart sound classification method based on ensemble learning // J.Chen, X. Dang, and M.Li. / Proc. ICSP -IEEE, China, 2022, pp 8-13/ doi: 10.1109/ICSP54964.2022.9778383.5.5. T. Tuncer. Application of Petersen graph pattern technique for automated detection of heart valve diseases with PCG signals // T. Tuncer, S. Dogan, R.-S. Tan, and U. R. Acharya / Information Sciences - United Kingdom, 2021, vol. 565, pp. 91–104. Doi: 10.1016/j.ins.2021.01.088.6.6. Zeinali, Yasser & Niaki, Seyed. Heart sound classification using signal processing and machine learning algorithms // Zeinali, Yasser & Niaki, Seyed / Machine Learning with Applications - 2022, №7 (12), pp 100206. DOI: 10.1016/j.mlwa.2021.100206.7.7. P. M. Aparna, G. N. Jayalaxmi, Vishwanath P. Baligar. Heart Sound Classification System Using Deep-Learning Neural Networks // Proceedings of 4th International Conference on Frontiers in Computing and Systems, 2024, pp 647-658. doi:10.1007/978-981-97-2614-1_46.8.8. Yaseen, Son G-Y, Kwon S. Classification of heart sound signal using multiple features // Yaseen, Son G-Y, Kwon S. / Applied Science – Seoul, Korea, 2018, №8(12), pp 2344-2358. doi: 10.3390/app8122344.9.9. Ali Raza. Heartbeat Sound Signal Classification Using Deep Learning // Ali Raza, A. Mehmood, S. Ullah, Maqsood Ahmad, G.Choi, Byung-Won On / Sensors - Basel, Switzerland, vol 19(21), November 2019, pp . doi:10.3390/s19214819.10.10. Prakash A.J., Ari S. A system for automatic cardiac arrhythmia recognition using electrocardiogram signal // Prakash A.J., Ari S. / In book: Bioelectronics and.11.11. Medical Devices. San Francisco: Elsevier, 2019 - pp. 891–911. DOI: 10.1016/B978-0-08-102420-1.00042-X.12.12. Huang J. ECG Arrhythmia Classification Using STFT-Based Spectrogram and Convolutional Neural Network // Huang J., Chen B., Yao B., He W/ IEEE Access, 2019, V. 7. pp. 92871–92880.13.13. Alfaras M., Soriano M.C., Ortín S. A fast machine learning model for ECG-based heartbeat classification and arrhythmia detection // Alfaras M., Soriano M.C., Ortín S. / Frontiers in Physics. 2019., vol. 7 – pp . DOI: 10.3389/fphy.2019.00109.14.14. О.В. Непомнящий. Использование сверточной нейронной сети при анализе электрокардиограмм // О.В. Непомнящий, А.Г. Хантимиров, М.М.И. Аль-сагир, С. Шабир / Нейрокомпьютеры: разработка и применение, т.25, №2, 2023- с.58-65.15.15. Омаров Б.С. Электронный стетоскоп для выявления сердечных аномалий // Омаров Б.С. Айдаров К.А., Султан Д.Р., Байкувеков М.Б. / Вопросы технических и физико-математических наук в свете современных исследований, Т.9-10 (47), Новосибирск, 2022, стр. 5-1616.16. Бейненсон Л.Б. Определение сердечно-сосудистых заболеваний по аудиозаписи биения сердца с помощью нейронной сети // Л.Б. Бейненсон / Тенденции развития науки и образования. №70-1, 2021, стр. 9-12, МИРЭА – Российский технологический университет.17.17. Classifying Heart Sounds Challenge. Режим доступа- https://istethoscope.peterjbentley.com/ heartchallenge/index.html (дата обращения 7.08.2024)18.18. Колесникова, Е.В. Роль свободно циркулирующей днк и свободных нуклеотидов в диагностике, оценке степени тяжести и динамическом контроле пациентов с хронической сердечной недостаточностью: автореф. дисс… к.м.н., 3.1.18. Внутренние болезни. Воронеж, 2024. 22 с.