Table 3 A) Linear mixed model results for effects of diagnosis, ROI and the top performing SNPs on BPND. B) Binomial logistic regression results for classification of diagnosis.

From: Machine learning classification of ADHD and HC by multimodal serotonergic data

A) Mixed model

DF numerator

DF denominator

F value

p-value

Group ∗ ROI ∗ HTR2A rs1328684

49

1666

1.61

0.001

Group ∗ ROI ∗ HTR1B rs130058

49

1666

9.49

<0.0001

Group ∗ ROI ∗ HTR2A rs6313

49

1666

2.38

<0.0001

Group ∗ ROI ∗ TPH2 rs1843809

49

1666

2.08

<0.0001

B) Logistic regression model

Estimator

Std. Error

Z value

p-value

Temporal Pole ∗ HTR1B rs130058

−21.76

9.52

−2.29

0.022

PCG ∗ HTR1B rs130058

−17.93

8.38

−2.14

0.032

  1. Models were built for each of the top performing ROI and, respectively, for each interaction with performing SNPs, according to the machine learning classification model. Interaction effects are marked with ∗. Only significant results are shown (uncorrected). Mixed model results with p ≤ 0.001 stayed significant after correction for multiple testing. None of the logistic regression results withstood correction.
  2. ROI region of interest, SNP single nucleotide polymorphisms, BPND non-displaceable binding potential, DF Degrees of Freedom, PCG Posterior Cingulate Gyrus.
Back to article page