In 2019, cardiovascular diseases were responsible for approximately 20.5 million lives, and 22 million deaths are expected in 2030 (World Heart Vision, 2025). This problem is partly caused by the lack of early and accurate diagnosis. This paper aims to develop an AI diagnostic tool that predicts the likelihood of heart disease. The model was based on the Random Forest Classifier, a machine learning method, and trained with pre-processed data of 303 patients. The SHAP method was applied to make the model interpretable and trustworthy. In order to achieve 94% accuracy, it was optimized through hyperparameter tuning. We integrated this model into a website to build a trustworthy and accessible prediction tool.