Medical Records or Electronic Health Records refers to the collection of patient health information in a digital format. The problem of classifying medical record data involves high-dimensional features. This raises a problem in determining which features have a correlation with the predicted results. Embedded technique uses learning model construction and feature selection simultaneously. The Wrapper technique performs feature evaluation by utilizing machine learning algorithms. The experimental results produce accuracy values for each embedded and wrapper technique. In this research, it is proposed to develop a hybrid technique that aims to find feature significance by applying machine learning techniques to increase the accuracy of predictions for disease classification. The proposed hybrid model combines the results of feature weighting and compares feature performance with several known classification techniques. The test results resulted in an increase in the accuracy value according to the disease classification dataset through the hybrid feature weight evaluation (HFWE) model.