In Vitro Method for Predicting Glycemic Index of Foods Using Simulated Digestion and an Artificial Neural Network

The glycemic index (GI) is an indicator of the relative human glycemic response to dietary carbohydrates in a food. It is determined using a costly and time-consuming in vivo method. We describe an in vitro analytical method that allows the accurate prediction of the GI of a food product. The method involves digestion of the food product using HCl and enzymes, followed by HPLC analysis of sugars and sugar alcohols. Data from the HPLC analysis combined with the product's compositional information are treated using an artificial neural network to produce a predicted value for the GI of the food product. For the sample set examined (n = 72) consisting of a variety of food types, r2 = 0.93 and the root mean square error of correlation (RMSEC) = 5 GI units. Twenty-fold cross-validation yields CVR2 = 0.89, indicating good predictive ability for samples outside the calibration set. The relative standard deviation of the method is 6.6%. This method is rapid and low cost relative to in vivo testing. Due to good ability to predict in vivo GI, it may be a valuable screening tool for determining the relative effect of food ingredients on the glycemic index of a food product.