The use of medical datasets has attracted the attention of researchers worldwide. Data mining techniques have been widely used in developing decision support systems for diseases prediction through a set of medical datasets. In this paper, we propose a new knowledge-based system for diseases prediction using clustering, noise removal, and prediction techniques. We use Classification and Regression Trees (CART) to generate the fuzzy rules to be used in the knowledge-based system. We test our proposed method on several public medical datasets. Results on Pima Indian Diabetes, Mesothelioma, WDBC, StatLog, Cleve-land and Parkinson's telemonitoring datasets show that proposed method remarkably improves the diseases prediction accuracy. The results showed that the combination of fuzzy rule-based, CART with noise removal and clustering techniques can be effective in diseases prediction from real-world medical datasets. The knowledge-based system can assist medical practitioners in the healthcare practice as a clinical analytical method. (C) 2017 Elsevier Ltd. All rights reserved.