Time-Lagged Prediction of Food Craving with Elastic Net Regression and BISCWIT: Prediction Performance of Qualitative Distinct Predictor Types
Keyword(s):
Food craving (FC) peaks are highly context-dependent and variable. Accurate prediction of FC might help preventing disadvantageous eating behavior. Here, we examine whether data from two weeks of Ecological Momentary Assessment questionnaires on stress and emotions alongside smartphone sensor data would be able to predict food cravings approximately 2.5 hours into the future in N = 46 individuals. We compared two prediction approaches (logistic BISCWIT and Elastic Net Regression (ENR). While overall prediction accuracy was low, passive sensing data alone was superior to predictions based on psychometric data and BISCWIT was superior to ENR for binary classification of FC peaks.
Keyword(s):