Title: Prediction of Heart Failure Using Artificial Neural Networks
Authors: Mohammed I. M. Al-Laham and Samy S. Abu-Naser
Volume: 10
Issue: 7
Pages: 77-85
Publication Date: 2026/07/28
Abstract:
Heart failure is one of the leading causes of hospitalization and mortality worldwide. Early identification of patients who are at risk of adverse outcomes can significantly improve medical decision-making and patient management. In recent years, Artificial Neural Networks (ANNs) have demonstrated remarkable capabilities in handling complex medical datasets and discovering hidden relationships among clinical variables. This study presents the development of a neural network model for predicting heart failure outcomes using the Heart Failure Clinical Records dataset obtained from the UCI Machine Learning Repository. The dataset was preprocessed and prepared using Microsoft Excel before being imported into the JustNN neural network software. A feed-forward neural network architecture was created and trained using twelve clinical input variables and one output variable representing the patient's survival status. The dataset consisted of 299 patient records. The neural network architecture generated by the training process consisted of 12 input neurons, 5 hidden neurons, and 1 output neuron. Training was performed using a learning rate of 0.6, momentum coefficient of 0.8, and a target error value of 0.01. The obtained results demonstrate the ability of artificial neural networks to model complex medical relationships and provide useful predictions regarding heart failure outcomes. The developed model achieved an average error value of 0.054126, indicating an acceptable level of predictive performance. The findings support the applicability of neural network techniques in healthcare decision-support systems and encourage further research using larger datasets and advanced neural architectures.