A pioneering method has demonstrated its potential in a large study, showing that metabolic fingerprints from blood samples could render important new knowledge on the connection between food and health. The study finds that diet is one of the strongest predictors of type 2 diabetes risk in older women.