Table 6.
Final predictors of cases of ADHD at ages 11–12 years.
| (A) Age of case determination | Ranked final predictors | Importance |
|---|---|---|
| New onset at age 11–12 years | Disorder of excessive somnolence | 0.2038 |
| Does very well at school | 0.1423 | |
| Parent reports some conflict with child | 0.0796 | |
| Youth gets excellent grades at school | 0.0567 | |
| Mean | 0.1206 | |
| All cases at age 11–12 years | Child has received MH/SU services in last 6 months | 0.1638 |
| Parent total behavioral problems | 0.1061 | |
| Parent reports some conflict with child | 0.0725 | |
| Disorder of sleep-wake transition | 0.0665 | |
| Youth performs very well in school | 0.0617 | |
| Disorder of excessive somnolence | 0.0375 | |
| Youth receives excellent grades in school | 0.0234 | |
| Prosocial behaviors mean score | 0.0147 | |
| Parent externalizing problems | 0.0091 | |
| Mean | 0.0617 | |
| (B) Neural data type | Ranked final predictors | Importance |
| New onset at age 11–12 years | T1 intensity in brain stem ROI | 0.0890 |
| Correlation between ventral attention network and right ventral diencephalon ROI | 0.0532 | |
| SST any stop vs. correct go contrast in left pars opercularis ROI | 0.0411 | |
| SST incorrect stop vs. correct go contrast in left lingual ROI | 0.0364 | |
| T1 intensity WM for left lateral occipital ROI | 0.0309 | |
| Cortical thickness in mm of right transverse temporal ROI | 0.0304 | |
| MID loss anticipation vs. neutral contrast in right supramarginal ROI | 0.0271 | |
| Average FA in GM right caudal ACC ROI | 0.0147 | |
| Mean | 0.0400 | |
| All cases at age 11–12 years | GM FA in right caudal middle frontal ROI | 0.0446 |
| Cortical area in mm2 of left inferior parietal ROI | 0.0259 | |
| MID small loss vs. neutral contrast in right inferior temporal ROI | 0.0192 | |
| SST incorrect go vs. incorrect stop contrast in left lingual ROI | 0.0134 | |
| Cortical thickness in mm of left pars triangularis ROI | 0.0129 | |
| SST incorrect stop vs. correct go in left lingual ROI | 0.0046 | |
| Mean | 0.0200 |
Final predictors of all prevailing cases of ADHD at 11–12 years, as well as new onset cases only at 11–12 years of age, are shown for the most accurate models obtained using deep learning optimized with IEL obtained with (A) multimodal features and (B) only neural features. Final predictors are ranked in the order of importance, where the relative importance of each predictor is computed with the Shapley additive explanation technique and presented here averaged across all participants in the sample. Features in red indicate an inverse relationship with ADHD verified with the Shapley method. MH, mental health; SU, substance use; SST, Standard Stop Signal task; MID, Monetary Incentive Delay task; ROI, region of interest; FA, fractional anisotropy; WM, white matter; GM, gray matter.