Özet:

Abstract In the human body, the heart is the most significant organ. The major role of the heart is to circulate blood, nutrients, along with oxygen all over the body. The heart’s function might get affected owing to a range of reasons. For identifying along with preventing sudden cardiac death, anomalous heart conditions must be detected in the earlier stage itself. To identify heart abnormalities, a non-invasive methodology termed Electrocardiogram (ECG) is utilized. The Electrical Activity (EA) of the blood circulatory system’s (heart) center is handled by this. For automatic Heart Disease Prediction (HDP), various Machine Learning (ML) and Deep Learning (DL) methodologies have been employed. Nevertheless, the diagnostic accurateness is affected owing to the influence of the external environment like signals’ poor quality along with inappropriate features. Furthermore, the classification of the sorts of cardiac disease was not done successfully by the prevailing methodologies. Thus, regarding ECG signals, an effectual HDP System (HDPS) has been proposed here by utilizing Patch wise Logistic Tanh SegNet (PLT-SegNet) along with Sinusoidal Chaotic JellyFish search hinged Softswish Artificial Neural Network (SCJFSANN) methodologies. Initially, from the openly accessible dataset, the input ECG signals are extracted. Then, the pre-processing is performed. After that, by utilizing Cosine Similarity adapted Hilbert Transform (CSHT), the signal peak values are identified. Next, the PLT-SegNet is utilized to partition the signals. Subsequently, by employing Shannon Entropy adapted Linear Discriminant Analysis (SELDA), the features are extracted along with the dimensionality is mitigated. Lastly, for classification, the SCJFSANN is provided with the dimensionality reduced features. The experiential outcomes displayed that the highest performance was achieved by the proposed methodology than the prevailing methodologies.

Anahtar Kelimeler: