The use of neural networks in the diagnosis of heart failure via the analysis of medical data

Document Type : Original Article


1 Univ Buenos Aires, Fac Agron, 4453 San Martin Ave, Buenos Aires, DF, Argentina

2 Fac Reg San Nicolas FRSN UTN San Nicolas, Dept Ingn Elect, Buenos Aires, DF, Argentina


The crux of effective management is in the process of decision-making, which is contingent upon the availability of information and effective communication. The fundamental responsibility of executives is to provide the necessary information to facilitate sound management decisions. This study seeks to utilize hospital managers' outcomes from data mining of hospital information systems to develop an intelligent model using machine learning techniques. The objective is to enhance the accuracy of predictions and facilitate more effective decision-making in patient treatment, recognizing the significance of hospital managers' decision-making approaches in advancing hospital goals and addressing patients' treatment challenges. The dataset used in this research pertains to the demographic and clinical information of 297 individuals. This data was obtained from the UCI website's data warehouse and encompasses 14 distinct variables. The three models, namely "k-means, support vector machine, and neural network," are extensively used classification methods in the domains of data mining and machine learning. These models have been applied to forecast cardiac disease, and their predictive performance has been evaluated and compared. The findings demonstrate that the neural network model, characterized by a multi-layered perceptron architecture, achieved a classification accuracy of 89.9% when applied to the test dataset. However, the support vector machine using the radial basis function kernel demonstrates enhanced accuracy, achieving a level of 93%.