Comparison of the left atrium diameter estimations with principal components regression, partial least squares regression and artificial neural networks metods

In this study, it was aimed to estimate of diameter of the left atrium with Principal Component Regression, Partial Least Square Regression and Artifical Neural Networks. For this purpose, firstly, multiple linear regression analysis, Least square method that is commonly used for parameter estimates and its assumptions were briefly described. Secondly, multicollinearity problem that occur in case of failure of assumption of independence between the explanatory variables was examined. Principal components regression and partial least square regression that used to overcome this problem were described. Finally, artificial neural network was examined. In the part of application, echocardiography reports of 127 hypertensive patients who came to Cardiology Polyclinic of Medicine Faculty of Firat University were collected prospectively. The obtained data were analyzed by all of methods described above and the results were compared.