Extended Data Fig. 4: Machine learning comparisons, cross validation and repeatability. | Nature Medicine

Extended Data Fig. 4: Machine learning comparisons, cross validation and repeatability.

From: Human postprandial responses to food and potential for precision nutrition

Extended Data Fig. 4

Machine learning comparisons, cross validation and repeatability. a, Bland-Altman plots comparing predicted and measured postprandial responses in TG, glucose and C-peptide using UK and US data. Sample sizes used were (n = number of meals) triglyceride UK: n = 958 US: n = 91; C-peptide UK: n = 957 US: n = 93; Glucose UK: n = 11550 US: n = 1200. b, Leave-one-out cross-validated Pearson R scores in PREDICT UK. 5-fold cross validation for triglyceride6h-rise on n = 958 meals, for lgucoseiAUC0-2h on n = 11,550 meals, p-values shown for two-sided t-test c, comparison of models using repeat meals vs not using them.

Source data

Back to article page