[Paper Review] iCardo: A Machine Learning Based Smart Healthcare Framework for Cardiovascular Disease Prediction

Nidhi Sinha, Teena Jangid+2 more|arXiv (Cornell University)|Dec 7, 2022

Healthcare Systems and Public Health4 citations

TL;DR

This paper proposes iCardo, a machine learning-based smart healthcare framework for early cardiovascular disease (CVD) prediction using multiple feature selection techniques and ensemble model evaluation. It identifies SVM with Recursive Feature Elimination (RFE) as the optimal classifier, achieving 92.31% accuracy on a combined heart disease dataset, and integrates the system with the Internet of Medical Things (IoMT) for real-time cloud-based patient monitoring.

ABSTRACT

The point of care services and medication have become simpler with efficient consumer electronics devices in a smart healthcare system. Cardiovascular disease is a critical illness which causes heart failure, and early and prompt identification can lessen damage and prevent premature mortality. Machine learning has been used to predict cardiovascular disease (CVD) in the literature. The article explains choosing the best classifier model for the selected feature sets and the distinct feature sets selected using four feature selection models. The paper compares seven classifiers using each of the sixteen feature sets. Originally, the data had 56 attributes and 303 occurrences, of which 87 were in good health, and the remainder had cardiovascular disease (CVD). Demographic data with several features make up the four groups of overall features. Lasso, Tree-based algorithms, Chi-Square and RFE have all been used to choose the four distinct feature sets, each containing five, ten, fifteen, and twenty features, respectively. Seven distinct classifiers have been trained and evaluated for each of the sixteen feature sets. To determine the most effective blend of feature set and model, a total of 112 models have been trained, tested, and their performance has been compared. SVM classifier with fifteen chosen features is shown to be the best in terms of overall accuracy. The healthcare data has been maintained in the cloud and would be accessible to patients, caretakers, and healthcare providers through integration with the Internet of Medical Things (IoMT) enabled smart healthcare. Subsequently, the feature selection model chooses the most appropriate feature for CVD prediction to calibrate the system, and the proposed framework can be utilised to anticipate CVD.

Motivation & Objective

• To develop a smart healthcare framework that enables early and accurate prediction of cardiovascular disease (CVD) using machine learning.
• To evaluate the performance of multiple machine learning classifiers across various feature selection methods to identify the most effective combination for CVD prediction.
• To integrate the optimized model with the Internet of Medical Things (IoMT) for real-time, cloud-based access to patient health data.
• To improve early diagnosis and reduce premature mortality through automated, data-driven CVD risk assessment.

Proposed method

• The framework uses a hybrid dataset combining Cleveland, Hungarian, Swiss, and Long Beach VA heart disease datasets, totaling 303 instances with 56 original attributes.
• Four feature selection methods—LASSO, Tree-based, Chi-Square, and Recursive Feature Elimination (RFE)—were applied to generate 16 distinct feature subsets (5, 10, 15, 20, 25 features each).
• Seven classifiers (including SVM, Naive Bayes, Random Forest, XGBoost, etc.) were trained and evaluated on each of the 16 feature sets, resulting in 112 unique model configurations.
• Performance was evaluated using standard metrics including accuracy, precision, recall, and F1-score, with confusion matrices used to validate model behavior.
• The best-performing model (SVM with RFE) was validated on a combined dataset, achieving 82.10% accuracy, outperforming the baseline SVM without feature selection (78.00%).
• The final system architecture enables cloud storage of EHR data and real-time access via IoMT-enabled devices for patients, caregivers, and clinicians.

Experimental results

Research questions

• RQ1 Which combination of feature selection method and machine learning classifier yields the highest accuracy in predicting cardiovascular disease?
• RQ2 How does the inclusion of demographic, clinical, ECG, and lab-based features impact the predictive performance of CVD models?
• RQ3 Can a machine learning-based framework integrated with IoMT enable real-time, remote, and accurate CVD prediction for improved patient outcomes?
• RQ4 What is the comparative performance of SVM, Naive Bayes, and other classifiers across multiple feature subsets derived from different selection techniques?

Key findings

• The Support Vector Machine (SVM) classifier achieved the highest overall accuracy of 92.31% when combined with feature sets selected by the Chi-Square and RFE methods.
• SVM with RFE-based feature selection produced the most robust performance, achieving 82.10% accuracy on the combined dataset, outperforming the baseline SVM without feature selection (78.00%).
• The Chi-Square feature selection method was found to be the most effective, consistently yielding high-performing feature subsets across multiple classifiers.
• Naive Bayes achieved 100% recall on several feature sets, but this was attributed to its assumption of feature independence, which does not hold in clinical data (e.g., age and hypertension are correlated).
• The confusion matrix analysis confirmed that SVM with RFE-based features had the highest overall accuracy and best balance of precision and recall.
• The proposed iCardo framework demonstrates strong potential for integration with IoMT platforms to enable real-time, cloud-based CVD risk prediction and remote patient monitoring.