Identifying People Based On Machine Learning Classification Of Foods Consumed In Order To Offer Tailored Healthier Food Options

Using food recall data collected from over 8000 Americans, we used Machine Learning to classify a person's demographic characteristics (age, gender, race, and income) based on the foods they consumed in a 24-h period. The best-performing models predicted gender correctly 61%, race 44%, and age 43% of the time on independent validation data. The model was subsequently used to provide tailored recommendations for healthier food options based on low-calorie and low-fat options for specific food groups that are typically consumed by other Americans that match the person's demographics. This system is part of a larger Smart Human-Centered System that assists users in recording the foods they consume, recognizes nutritional content of the food, offers tailored recommendations for consuming healthier foods, and tracking behavior change over time.