AP26197917 – Development of a recommendation system based on machine learning for early diagnosis of respiratory diseases using lung sound analysis
Objective of the project – To develop and implement a machine-learning recommendation system for early diagnosis of respiratory diseases by analyzing acoustic respiratory data, which will improve the accuracy and availability of diagnostic services in medical practice.
Relevance: The relevance of the project is determined by the persistent challenge of early diagnosis of respiratory diseases, especially in the context of increasing incidence rates and growing pressure on healthcare systems. Timely detection of lung pathologies remains limited, leading to delayed treatment and worsening patient conditions. In this regard, the development of non-invasive diagnostic methods, such as lung sound analysis, is of particular importance. The use of machine learning technologies and deep analysis of acoustic signals opens new opportunities to improve the accuracy and speed of diagnosis. The creation of an intelligent recommendation system will enable automation of pathology detection and provide physicians with an additional decision-support tool. Overall, the project contributes to improving the accessibility and quality of healthcare services, as well as reducing treatment costs through earlier intervention.
Scientific supervisor: Ph.D., Professor, Zhekambaeva Maygul Nesipaldievna
Expected and achieved results: Within the framework of the project, extensive datasets of lung sounds have been collected and preprocessed, including both normal and pathological conditions, ensuring high-quality data for model training. Various deep learning architectures have been developed and tested, including CNN, RNN, LSTM, as well as hybrid and ensemble models with attention mechanisms. The obtained results confirmed the effectiveness of the proposed approach for early diagnosis of respiratory diseases based on acoustic signal analysis. The scientific novelty of the project lies in the integration of modern sound processing methods with deep learning algorithms for medical diagnostics. Further development of the system is expected to enable highly accurate real-time classification of lung pathologies. The developed system will serve as an additional decision-support tool for physicians and improve the quality of diagnostics. The implementation of the technology in medical institutions of the Republic of Kazakhstan will reduce diagnostic time and ensure earlier initiation of treatment. It is also expected to reduce healthcare costs through improved diagnostic efficiency and prevention of complications. The commercialization of the project includes the deployment of the system in clinics and its adaptation for mobile platforms, enabling its use for home monitoring of patients. Overall, the project results will contribute to the development of scientific and technical capacity, enhance the competitiveness of research organizations, and foster new research directions in biomedical engineering and digital medicine.
List of publications with links to them
- Zhekambayeva M., Akylzhan P., Nazarova A., Yussupova G., Mailybayev E., Katayev N. Real-time augmented reality-enabled sports exercise monitoring system with personalized recommendations // Retos. – 2025. – Vol. 73. – P. 909–922. – DOI: https://doi.org/10.47197/retos.v73.117751. – ISSN 1579-1726