November 1, 2025National Journal of Community Medicine1 citationsOpen Access

Assessment of Risk Factors of Cardiovascular Diseases Using Artificial Neural Network (ANN): A Pilot Study

PSP. Reddy Gurunatha SwamyRajiv Gandhi University of Health Sciences KRKanika RaiChitkara University

Structured PICO

Does an Artificial Neural Network (ANN) model improve cardiovascular risk prediction compared to traditional methods in an urban population?

Population

Urban population in India, including 35 healthcare professionals for qualitative focus groups and 60 participants for quantitative Artificial Neural Network model training and testing.

Intervention

Artificial Neural Network (ANN) model (multi-layer perceptron) for cardiovascular risk prediction

Comparator

Traditional risk assessment tools (e.g., Framingham Risk Score, Systematic Coronary Risk Evaluation, Reynolds Risk Score)

Outcome

Predictive performance of the ANN model (accuracy and Area Under the Curve)

An Artificial Neural Network model demonstrated strong predictive performance for cardiovascular risk assessment, potentially outperforming traditional risk scores.

Abstract

Background: Cardiovascular diseases (CVDs) are the leading cause of death globally, particularly in urban populations. In India, an estimated 62.5 million lives are lost prematurely due to CVD. Traditional risk assessment tools like the Framingham Risk Score, Systematic Coronary Risk Evaluation, and Reynolds Risk Score are widely used. However, Artificial Neural Networks (ANNs) may provide improved prediction of cardiovascular risks. This study aims to assess cardiovascular risk factors among urban populations using ANN models and compare their effectiveness with traditional methods. Methodology: A sequential exploratory mixed-method approach was used. The qualitative phase included focus group discussions with 35 healthcare professionals to identify perceived cardiovascular risk factors. Analysis via QDA Miner Lite highlighted health education, screening, and training as crucial roles of health workers, especially concerning diabetes, diet, and hypertension. The quantitative phase used a multi-layer perceptron model to analyze data from 60 participants (pilot phase), divided into 77% training and 23% testing datasets. Results and Conclusion: The ANN model achieved 100% training accuracy and 85.7% testing accuracy, with an AUC of 0.969, showing strong predictive performance. The model effectively identified moderate-risk individuals, suggesting that ANNs outperform traditional methods in cardiovascular risk prediction.

Ask AI

Helpful

Bookmark

View Full Paper