A new study published in JAMA Cardiology on Jul. 15 shows that a simplified machine learning version of the Martin-Hopkins equation for calculating low-density lipoprotein (LDL) cholesterol matches the accuracy of the original method, making it more accessible for widespread use in laboratories. The research analyzed millions of blood samples from both adults and children across the United States.
Accurate LDL cholesterol assessment is increasingly important as current guidelines recommend lowering these levels to reduce cardiovascular risk. Seth Martin, one of the study authors, said underestimation using some equations can lead to missed treatment opportunities, and that "a lipid profile with low cholesterol and high triglycerides is the ultimate stress test of the LDL cholesterol calculation." He added, "It's these types of on-the-cusp examples that benefit most from more accurate results."
The researchers developed and tested their machine-learning formula using data from 4.9 million U.S. children and adults drawn from the Very Large Database of Lipids, which had a median LDL cholesterol level of 114 mg/dL. They compared its performance against existing methods such as the Sampson-NIH and Friedewald equations by referencing ultracentrifugation measurements—a gold standard in research settings.
Results showed that both versions of the Martin-Hopkins equation correctly classified 90% of samples within appropriate treatment categories, while other common equations like Sampson-NIH and Friedewald performed less accurately at 86% and 83%, respectively. In patients with triglycerides between 200 mg/dL and 399 mg/dL and LDL below 70 mg/dL—considered high-risk—the new model accurately classified up to 84% compared to only 40% for Friedewald.
Mark Marzinke, another study author who oversees testing at Johns Hopkins Hospital Core Laboratories, said, "This updated equation is not only highly accurate, but it's transparent and can be easily adopted by laboratories." He emphasized avoiding a 'black box' approach so users can understand how calculations are made.
The authors note this open-access calculation could help implement national dyslipidemia guidelines recommending preferential use of Martin-Hopkins calculations for LDL assessment. The equation has no patent or intellectual property restrictions.
