Atherosclerotic cardiovascular disease (ASCVD) is a major cause of death in Türkiye and worldwide. Because it progresses silently, early risk prediction is critical. Current risk scoring systems are based on standard factors and may overlook metabolic problems and insulin resistance. New data suggest that novel metabolic markers and machine learning, particularly in primary care settings, can help better predict risk.This research sought to create a better model for predicting the risk of ASCVD in men aged 40–65. It did this by including new factors related to metabolism and insulin resistance and by testing how well machine learning models could predict risk. The study checked if including HbA1c, TSH, SPISE, and AIP, along with standard risk factors, made the prediction of ASCVD risk more accurate.This case–control study looked at 350 men between 40 and 65 years old who visited the Family Medicine clinic at Ankara Bilkent City Hospital. Of the participants, 175 had ASCVD and 175 served as controls. Lifestyle and laboratory data were collected, and predictive modeling was conducted using ROC analysis and machine learning algorithms, including SVM, decision trees, and random forest models.Older age, family history, obesity, high systolic blood pressure, high triglycerides, low HDL-Cholestrol (HDL-C) and high HbA1c were associated with atherosclerotic cardiovascular disease. ROC analysis showed that AIP and SPISE better discriminated cases from those with normal lipid markers. The Support Vector Machine (SVM) model had the best among machine learning algorithms.The study results suggest that metabolic and insulin resistance factors greatly improve ASCVD risk prediction compared to standard risk factors. The strong performance of SPISE and AIP shows they are clinically relevant for early risk classification. Machine learning models, specifically SVM, showed better discrimination by assessing many interacting variables at once, which addresses some limits of standard scoring systems.Incorporating novel metabolic markers such as SPISE, AIP, HbA1c, triglycerides, physical activity level, and smoking exposure into machine learning–based models significantly improves ASCVD risk prediction. The SVM model showed high accuracy and discriminatory power, suggesting strong potential for personalized, data-driven ASCVD risk assessment tools that can be effectively integrated into primary care practice.