Coronary heart disease is a disorder caused by the inhibition of arteries that drain blood to the heart muscle. This disease is a non-communicable disease which often results in death directly to the victims. This study aims to design an artificial neural network architecture using the backpropagation method that can predict a person having coronary heart disease by inputting cholesterol levels, blood pressure, and blood sugar levels, and body mass index. This research is research and development. The research methods used in making this prototype, namely: (1) problem analysis, (2) needs analysis, (3) literature study, (4) prototype design, and (5) prototype testing. The patient data used to test the prototype were 20. The results show that the neural network model used has an average error value of 0.792% with 5000 training times. A diagnostic prototype for coronary heart disease using backpropagation was successfully built with good results.