Predicting and diagnosing cardiac conditions is a major challenge in medicine because of the many variables involved, including the results of a physical examination and the patient's symptoms and indicators. According to data from the World Health Organi-zation (WHO), heart disease is the leading cause of death worldwide, taking the lives of 18 million people annually. Machine Learning (ML) algorithms are essential to modern medicine, particularly when it comes to using medical databases to diagnose illnesses. In this paper, a novel ML model called Heart Disease Detection Model (HD2M) is presented. The suggested HD2M has four phases: (i) data collecting and preprocessing, which in-cludes converting non-numerical values into numbers, eliminating outliers, and filling in missing values. (ii) feature selection using Archimedes Optimization Algorithm (AOA). Actually, AOA is used to select the most important features. (iii) Patient detection through this phase the selected features are used to fed various ML classifiers. These classifiers are Extreme Gradient Boost (EGB), Random Forest (RF), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN). (iv) Evaluate Model and Disease Prediction. Experimental results indicate that HD2M outperforms its rivals in terms of F1-measure, precision, accuracy, and recall.