Automatic ECG Signal Quality Determination Using CNN With Optimal Hyperparameters for Quality-Aware Deep ECG Analysis Systems

Mondal, Achinta; Manikandan, M. Sabarimalai; Pachori, Ram Bilas

doi:10.1109/jsen.2024.3382720

Cited by 9 publications

(1 citation statement)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Traditional methods [6], for diagnosing MI encompass clinical assessments [7], ECG signal analysis [8], and medical image-based analysis (MIA) [9], utilizing modalities like computed tomography (CT) [10], magnetic resonance imaging (MRI) [11], positron emission tomography (PET) [12], and ultrasonography [13]. However, a group of researchers working in this area [14], and traditional methods are working well but some research gaps are available, aiming to minimize the research gap from the traditional approach.…”

Section: Introductionmentioning

confidence: 99%

Quantum Leap in Cardiac Prognosis: EMIP-CardioPPG’s Pioneering Approach to Early Myocardial Infarction Prediction

Shrivastava,

Kumar,

Srinivas

2024

Preprint

View full text Add to dashboard Cite

Cardiovascular disease (CVD) encompasses conditions affecting the heart and blood vessels, leading to ailments such as coronary artery disease, heart failure, and myocardial infarction (MI), commonly referred to as a heart attack, which occurs when blood flow to the heart is obstructed. Early prediction of MI is vital to prevent severe damage and enhance survival rates. Traditionally, an electrocardiogram (ECG) is employed to detect cardiovascular anomalies, but its requirement for multiple electrodes placed on various body locations makes continuous monitoring difficult. Current research gaps involve the necessity for medical assistance during ECG monitoring, data variability, early symptom prediction, and limited data availability due to the sensitive nature of medical records. To address these issues, we introduce EMIP-CardioPPG, a novel mathematical framework for early MI prediction using CardioPPG, a non-invasive method that utilizes photoplethysmography (PPG) signals to monitor heart rate (HR) and detect cardiovascular abnormalities. Our approach comprises four steps: first, acquiring data from the same individual using two different sources, a self-created IoMT device and a 4-channel BIOAC-MP 36 device; second, preprocessing the data by denoising, filtering, normalizing, and removing motion arti-facts; and third, employing mathematical calculations to determine heart rate variability (HRV) and HR, enhancing PPG signal features for early MI prediction. fourth, Evaluate our model performance using machine learning (ML) algorithms such as ridge regression (RR), support vector classifier (SVC), independent component analysis (ICA), singular value decomposition (SVD), random forest (RF), and XGBoost, achieving high accuracy of 97.91% for HRV from our IomT device and 98.83% for HR from our BOPAC-MP 36. In future studies, we will compare our results with state-of-the-art algorithms and evaluate our model’s performance using cardiac images.

show abstract