A Machine Learning Approach for Accurate Diagnosis of Heart Disease: A Comparative Study

Heart diseases are leading determinants of worldwide morbidity indices, underscoring the critical importance of accurate diagnostic assessments in facilitating timely therapeutic interventions and optimizing clinical outcomes. Traditional methods of diagnosis rely on clinical evaluations and laboratory tests, which may be subjective and prone to errors. In the present study, we suggest implementing a machine learning methodology to construct a prognostic framework for heart disease diagnosis by integrating demographic details, medical records, and laboratory information. In this research, in order to design an intelligent monitoring system for cardiac disease, to increase the accuracy of diagnosis and also reduce the classification error rate, as an innovative aspect of the research, we used a recurrent neural network LSTM with the optimal neuron value of its hidden layer obtained using the particle optimization algorithm. In this study, after preprocessing by normalizing the data and removing missing data, we reduced the number of unimportant features using principal component analysis in the next step so that we could classify the data more accurately using the recurrent neural network. In this research, we attained a precision level of 99.3%, surpassing the proposed technique by 1% when contrasted with the conventional method outlined in the foundational article.