Researchers have found a method using molecular profiling and machine learning to develop blood-based dietary signatures that more accurately assess diet and predict the risk of cardiovascular disease and type 2 diabetes. They say the metabolic snapshot could allow those studying food science to better understand the implications of diet and nutrition on health.