Fig. 7: Factors associated with prediction performance of CAT-DI severity scores.
From: Personalized mood prediction from patterns of behavior collected with smartphones
Correlation between prediction accuracy of an individual (metrics on the y-axis) and the number of CAT-DI assessment available in the training and test set, the difference in median CAT-DI severity between the training and test set, the number of the unique CAT-DI categories (normal to severe) observed (total and in training and test sets), age, sex, wave, and treatment group (a proxy for depression severity). MAPE: mean absolute percentage error.
