P-NUT: Predicting NUTrient Content from Short Text Descriptions

Assessing nutritional content is very relevant for patients suffering from various diseases, professional athletes, and for health reasons is becoming part of everyday life for many. However, it is a very challenging task as it requires complete and reliable sources. We introduce a machine learning pipeline for predicting macronutrient values of foods using learned vector representations from short text descriptions of food products. On a dataset used from health specialists, containing short descriptions of foods and macronutrient values: we generate paragraph embeddings, introduce clustering in food groups, using graph-based vector representations, that include food domain knowledge information, and train regression models for each cluster. The predictions are for four macronutrients: carbohydrates, fat, protein and water. The highest accuracy was obtained for carbohydrate predictions – 86%, compared to the baseline – 27% and 36%. The protein predictions yielded the best results across all clusters, 53%–77% of the values fall in the tolerance-level range. These results were obtained using short descriptions, the embeddings can be improved if they are learned on longer descriptions, which would lead to better prediction results. Since the task of calculating macronutrients requires exact quantities of ingredients, these results obtained only from short description are a huge leap forward.