TY - JOUR T1 - Heart disease forecast using neural network data mining technique TT - پیش بینی بیماری قلبی با استفاده از تکنیک داده کاوی شبکه عصبی JF - sjimu JO - sjimu VL - 25 IS - 1 UR - http://sjimu.medilam.ac.ir/article-1-2376-en.html Y1 - 2017 SP - 20 EP - 32 KW - Data mining KW - Neural network KW - Heart-disease N2 - Abstract Introduction Introduction Data mining refers to the study and analysis of large amounts of data for discovering meaningful patterns and rules. Mainly through the models and algorithms, data mining puts the inputs in a specific order. Data mining techniques sometimes lead to the identification of meaningful algorithms which can use available and low-cost data in order to provide us with areas of infection, prevention, and treatment of diseases and help the physicians in timely and accurate diagnosis. Materials & Methods The present paper aimed to study the use of the results of data mining of hospital information systems by hospital managers for more accurate prediction and more effective decision-making about treatment of patients. The data used in this study included the information of 270 patients (14 variables) extracted from the database of UCI website. A “neural networks” model was used for the prediction of affliction with heart disease and its accuracy was measured and compared. Research findings According to the results, it can be observed that Multilayer Perceptron Neural Networks Model has classified the set of test observations with an accuracy of 83.33%. Discussion & Conclusion The results showed that the accuracy of “neural networks” model in classification of records in terms of heart disease response is 87.75% for the set of modeling records and 83.33% for the set of test records. In addition, the findings revealed that the variables of the number of large vessels (Nbr-ves), stress reduction (ST-dep), defect, chest pain, stress peak (Peak-ST), heart rate, angina, gender, age, static ECG (Res-elec), blood pressure (Blood-press), blood sugar, and serum cholesterol (Serum-chol), respectively, have the highest importance in “Multilayer Perceptron Neural Networks” model for the prediction of heart disease response. M3 10.29252/sjimu.25.1.20 ER -