Table 7.

Performance of deep learning optimized with integrated evolutionary learning in predicting cases of ODD using multimodal and neural-only feature types.

Age of case determination	Accuracy (%)	Precision (%)	Recall (%)	AUROC
(A)
New onset at age 11–12 years	96.8	96.8	93.5	0.996
All cases at age 11–12 years	96.2	94.8	95.5	0.988
(B)
New onset at age 11–12 years	74.2	69.4	64.5	0.792
All cases at age 11–12 years	56.1	53.9	27.2	0.567

Performance statistics of accuracy, precision, recall, and the AUROC are shown for the most accurate model obtained with deep learning optimized with integrated evolutionary learning using (A) multimodal features and (B) only neural features. We used features obtained at 9–10 years of age to predict new onset cases of ODD at 11–12 years of age, as well as all prevailing cases at 11–12 years of age. Corresponding ROC curves may be viewed in Supplementary Figures 1, 2.