A Novel Approach for Specification Testing on Heart Disease Detection Using Feed-Forward Neural Network

• Published in: SN computer science 2023-11, Vol.4 (6), p.850, Article 850
• Main Authors: Naresh, E., Darshan, S. L. Shiva, Srinidhi, N. N., Niranjanmurthy, M., Dayananda, P.
• Format: Article
• Language: English
• Subjects: Abnormalities; Accuracy; Algorithms; Artificial neural networks; Back propagation; Cardiovascular disease; Computer Imaging; Computer Science; Computer Systems Organization and Communication Networks; Data Structures and Information Theory; Datasets; Error correction & detection; Fault detection; Freeware; Heart; Heart diseases; Illnesses; Information Systems and Communication Service; Literature reviews; Machine Intelligence and Smart Systems; Machine learning; Multilayer perceptrons; Neural networks; Original Research; Patients; Pattern Recognition and Graphics; Product lines; R&D; Regression analysis; Research & development; Software Engineering/Programming and Operating Systems; Software quality; Test methods; Vision
• ISSN: 2661-8907; 2662-995X; 2661-8907
• DOI: 10.1007/s42979-023-02332-9

The proposed work highlights the importance of testing in machine-learning applications and the ensuing need to increase model quality to decrease the likelihood of errors. The proposed work considers patient health information that may be utilized for decision-making or prediction utilizing various computations, and in this instance, it emphasizes the development of artificial neural networks with the multilayer perceptron method to forecast cardiac abnormalities. The dataset employed in this work includes information about the patient’s demographics, clinical measures, and medical background. Utilizing this labelled dataset, one may estimate if each patient has heart disease or not. We developed a GUI to collect data about a new patient to accomplish the task specified. After the model is generated, effective functionality testing was performed on the model. As a result of the testing report’s assistance in identifying faults and defects, the artificial neural network’s accuracy ultimately improved as a result of its improved accuracy following correction. After undergoing feature testing, the application’s accuracy ascended to 96.69%, which is a higher percentage than was attained using any alternative method.