Heart Disease Prediction using Multimodal Data with Multi-Layer Perceptron

Cardiovascular diseases (CVD) continue to be an increasing worldwide health issue, requiring specialized diagnostic instruments for early identification and care. The proposed technique seeks to improve cardiac disease prediction by using a multi-modal approach that combines patient’s demographic data with raw ECG signals. This approach combines signal processing, feature extraction, and selection methods to enhance the predicted accuracy of the system. Thus, here we develop an application which can predict the vulnerability of a heart disease by giving details like age, gender, height, weight, etc., along with 12 lead ECG signal images provided in the PTB-XL dataset. This system method employs advanced Fast Fourier Transform (FFT) feature extraction technique, to extract informative features from ECG signals. Additionally, random forest classification algorithm is utilized for feature selection to identify the most discriminative attributes for prediction. The extracted features are then inputted into a Multi-Layer Perceptron (MLP) model, which is trained on a comprehensive dataset comprising patient demographics and ECG signals.