A Novel Multi-Class Classification System Improving Heart Disease Diagnosis With Combined ECG and PCG Signals
Research Shows Integration of Electrocardiogram and Phonocardiogram Signals Leads to Unprecedented Diagnostic Accuracy
Heart diseases remain one of the leading causes of death worldwide, but recent advancements in medical technology are changing the way these conditions are diagnosed. A groundbreaking study published on March 9, 2025, introduces a multi-class heart disease classification model utilizing both Electrocardiogram (ECG) and Phonocardiogram (PCG) signals. This innovative approach achieves remarkable results, boasting accuracy rates of 97% with minimal error, significantly outperforming traditional methods.
The heart is the core organ responsible for circulating blood throughout the body, and its optimal function is imperative to overall health. Cardiovascular diseases (CVDs), which include various disorders related to the heart, are often marked by the buildup of plaque within arteries, leading to severe complications including heart attacks. Identifying these conditions early can be lifesaving, emphasizing the importance of effective diagnostic methods.
Current diagnostic techniques tend to rely on either ECG or PCG signals. ECG focuses on the electrical activities of the heart, measuring heartbeats through electrochemical signals, whereas PCG hones in on the sound made by the heart as it pumps blood. While each method has its strengths, reliance on one signal type often results in missed data. ECG may overlook characteristic heart murmurs associated with valvular heart diseases, and PCG might miss the subtle changes reflective of electrical activity. To address these limitations, the researchers combined both signals for comprehensive analysis and diagnosis.
To automate heart disease detection, the proposed model employs a series of advanced technologies, starting with data preprocessing to remove noise and improve signal clarity. The integration process utilizes the Algebraic Integer-quantized Stationary Wavelet Transform (AI-SWT) to effectively localize heartbeats, ensuring no important information is lost. Once both signals are clean and fused, they undergo analysis through the Pan-Tompkins Algorithm for waveform detection. Next, the study introduces Low-rank Kernelized Density-Based Spatial Clustering of Applications with Noise (LK-DBSCAN) for classifying signals as normal or abnormal.
The classifier, based on the C squared Pool Sign BI-power-activated Deep Convolutional Neural Network (CP-SBI-DCNN), processes the normalized signals to predict the likelihood of specific heart disorders, which include Aortic stenosis, Mitralstenosis, and Atrial Fibrillation, among others. The implementation of this integrated system not only leads to significant detection rates but also minimizes false positives—148 errors for every 4,950 predictions made by the model.
Results verified through various metrics yielded impressive performance figures: accuracy of 97%, precision of 95%, recall of 96%, and sensitivity of 98%. This data suggests the model is adept not only at identifying conditions but also at minimizing missed cases, which is often the case with standalone diagnostic tools.
Together, the combination of ECG and PCG signals presents clinicians with powerful tools for diagnosing various heart diseases. Beyond improving the reliability of heart disease diagnoses, this method could also pave the way for the development of wearable health devices capable of early detection, enhancing proactive healthcare.
Although the current findings are promising, the authors highlight areas for future research. The integration of additional biomarkers, such as biochemical indicators related to heart health, may provide even more comprehensive insights. By advancing these methodologies, the medical programming community could significantly improve cardiovascular diagnostics for both patients and practitioners.
Modern technological interventions like the one presented here underline the significant role machine learning plays within healthcare. The utilization of AI techniques pushes beyond traditional limitations, creating opportunities for faster and more accurate diagnosis and treatment of heart-related illnesses. The adoption of such systems could lead to improved patient outcomes and more efficient healthcare processes, fulfilling the urgent need for effective heart disease management in today's clinical settings.
Conclusively, leveraging advanced computational models alongside traditional medical practices establishes new pathways for improving patient care and outcomes in the battle against cardiovascular diseases.