The accurate diagnosis of heart diseases, is one of the most important biomedical problems whose administration is imperative. In the proposed work, decision support system is made by three data mining techniques namely Classical Random Forest, Modified Random Forest and Weighted Random Forest. The classical random forests construct a collection of trees. In Modified Random Forest, the tree is constructed dynamically with an online fitting procedure. A random forest is a substantial modification of bagging. Forest construction is based on three step process . 1. Forest construction 2. The polynomial fitting procedure 3.The termination criterion In Weighted Random Forest, The Attribute Weighting Method is used for improving Accuracy of Modified Random Forest. There are Two Techniques are Used in Attribute Weighting: 1. Averaged One-Dependence Estimators (AODE) 2. Decision Tree-based Attribute Weighted Averaged One-dependence Estimator( DTWAODE).