HARNESSING ARTIFICIAL INTELLIGENCE FOR PREDICTIVE ANALYTICS IN SILENT CARDIOMYOPATHIES: A PARADIGM SHIFT IN EARLY DETECTION

Abstract

Silent cardiomyopathies are heart muscle diseases with no symptoms on early onset, and the delayed diagnosis may result in higher morbidity. In this paper, we explore the use of artificial intelligence (AI) predictive analytics for early detection of the silent cardiomyopathies. Based on a quantitative crosssectional design, 400 subjects aged 18–80 were recruited. Data on elecrocardiograms, echocardiograms, genetic markers, and clinical risk factors had been extracted and then applied to machine learning algorithms-svm, random forests, and neural networks.

The performance of the neural network model was superior (accuracy 90.0%, sensitivity 84.0%, specificity 93.0%, AUC-ROC 0.95) to conventional diagnostic methods and other AI models. Hypertension, family history of heart disease and diabetes were the most important predictors, but risk was also partially influenced by life style, such as smoking and physical inactivity. The results of the analysis suggest that the AI-based predictive models can diagnose silent cardiomyopathies with high sensitivity, thereby helping to intervene promptly, and offer personalized support. Incorporating AI into routine screening could potentially lead to a substantial benefit by mitigating complications and improving patient outcomes.