Use of machine learning to identify patients at risk of sub-optimal adherence: study based on real-world data from 10,929 children using a connected auto-injector device

Table 2 F1-scores for the four machine learning models evaluated over three time periods

Optimized Machine Learning Model	Mean (SD) F1-score (train set, fivefold cross-validation)
Optimized Machine Learning Model	3 months prediction	6 months prediction	9 months prediction
Logistic regression	0.797 (0.012)	0.785 (0.016)	0.768 (0.012)
Ordinal logistic regression	0.798 (0.011)	0.784 (0.013)	0.768 (0.010)
Random forest	0.807 (0.012)	0.798 (0.018)	0.780 (0.011)
Extra trees	0.801 (0.017)	0.783 (0.021)	0.771 (0.01)

ISSN: 1472-6947