A Compact LSTM-SVM Fusion Model for Long-Duration Cardiovascular Diseases Detection

Siyang Wu

2024-01-24

Abstract:Globally, cardiovascular diseases (CVDs) are the leading cause of mortality, accounting for an estimated 17.9 million deaths annually. One critical clinical objective is the early detection of CVDs using electrocardiogram (ECG) data, an area that has received significant attention from the research community. Recent advancements based on machine learning and deep learning have achieved great progress in this domain. However, existing methodologies exhibit inherent limitations, including inappropriate model evaluations and instances of data leakage. In this study, we present a streamlined workflow paradigm for preprocessing ECG signals into consistent 10-second durations, eliminating the need for manual feature extraction/beat detection. We also propose a hybrid model of Long Short-Term Memory (LSTM) with Support Vector Machine (SVM) for fraud detection. This architecture consists of two LSTM layers and an SVM classifier, which achieves a SOTA results with an Average precision score of 0.9402 on the MIT-BIH arrhythmia dataset and 0.9563 on the MIT-BIH atrial fibrillation dataset. Based on the results, we believe our method can significantly benefit the early detection and management of CVDs.

Signal Processing,Artificial Intelligence,Machine Learning

What problem does this paper attempt to address?

The problems that this paper attempts to solve mainly focus on the following aspects: 1. **Limitations of existing methods**: Existing machine - learning - and deep - learning - based methods have some inherent limitations in detecting cardiovascular diseases (CVDs), including inappropriate model evaluation and data leakage issues. These methods usually adopt the "intra - patient" paradigm, that is, the training set and the test set come from the same patient, which will lead to model over - fitting and cannot generalize well to new patients' data. 2. **Evaluation problems of unbalanced data sets**: In the MIT - BIH atrial fibrillation data set and the MIT - BIH arrhythmia data set, the proportion of normal and abnormal samples is severely unbalanced. Traditional evaluation metrics (such as accuracy, specificity, sensitivity, etc.) may produce misleading results on such unbalanced data sets. Therefore, more appropriate evaluation metrics are needed to measure model performance. 3. **The need for early detection and management of CVDs**: Cardiovascular diseases are one of the leading causes of death globally, causing approximately 17.9 million deaths each year. Early detection and management of CVDs are crucial for reducing mortality. However, traditional manual feature extraction and QRS - wave detection methods are less efficient and time - consuming, and it is difficult to meet the needs of practical applications. To solve the above problems, the author proposes a new compact LSTM - SVM fusion model (LSF), aiming to: - Ensure that the model can better generalize to new patients' data by strictly following the "inter - patient" paradigm. - Use the average precision score (AP score) as an evaluation metric, which is a more suitable evaluation standard for dealing with unbalanced data sets. - Propose an end - to - end workflow without manual feature extraction or QRS - wave detection, improving the detection efficiency. - Verify on the MIT - BIH arrhythmia data set and the MIT - BIH atrial fibrillation data set, demonstrating the superior performance of the model on long - term electrocardiogram data. In summary, the goal of this paper is to develop a more efficient and reliable CVDs detection method to address the limitations and challenges in existing methods.

A Compact LSTM-SVM Fusion Model for Long-Duration Cardiovascular Diseases Detection

Integrated Approach for ECG Signal Classification Using Fused Slantlet Transform and Autoregressive Features

A Wearable ECG Monitor for Deep Learning Based Real-Time Cardiovascular Disease Detection

Reliable Detection of Myocardial Ischemia Using Machine Learning Based on Temporal-Spatial Characteristics of Electrocardiogram and Vectorcardiogram

Patient-Specific ECG Classification with Integrated Long Short-Term Memory and Convolutional Neural Networks

Efficient Prediction Model for Cardiovascular Disease Using Deep Learning Techniques

A Multi-Task Group Bi-LSTM Networks Application on Electrocardiogram Classification

An ECG Signal Acquisition and Analysis System Based on Machine Learning with Model Fusion

A novel multi-resolution SVM (MR-SVM) algorithm to detect ECG signal anomaly in WE-CARE project

Early prediction of sudden cardiac death using multimodal fusion of ECG Features extracted from Hilbert-Huang and wavelet transforms with explainable vision transformer and CNN models

An efficient cardio vascular disease prediction using multi-scale weighted feature fusion-based convolutional neural network with residual gated recurrent unit

Early detection and prediction of Heart Disease using Wearable devices and Deep Learning algorithms

An Improved Long Short-Term Memory Algorithm for Cardiovascular Disease Prediction

Enhancing Cardiovascular Disease Detection through Deep Multimodal Fusion: Integrating Radiology and Electrocardiogram via Convolutional Neural Network

Detection and Classification of electrocardiography using hybrid deep learning models

CE-SSL: Computation-Efficient Semi-Supervised Learning for ECG-based Cardiovascular Diseases Detection

Comparative analysis of supervised learning algorithms for prediction of cardiovascular diseases

A Novel Deep Learning Approach for Early Detection of Cardiovascular Diseases from ECG Signals

Detecting Atrial Fibrillation from Single-Lead ECG Using Unbalanced Multi-classification Support Vector Machine.

TetraCVD: A Temporal-Textual Transformer based Model for Cardiovascular Disease Diagnosis.

Machine learning-based detection of cardiovascular disease using ECG signals: performance vs. complexity