Sparse multiple factor analysis to integrate genetic data, neuroimaging features, and attention-deficit/hyperactivity...

Area of Science:

• Neuroscience
• Genetics
• Psychiatry

Background:

• Attention-deficit/hyperactivity disorder (ADHD) is a complex neurodevelopmental disorder.
• Understanding the neurobiological underpinnings of ADHD requires advanced analytical methods.

Purpose of the Study:

• To apply a multivariate cross-sectional framework combining variable selection and Multiple Factor Analysis (MFA) for identifying biological signals related to ADHD symptoms.
• To investigate the relationships between ADHD symptom domains (hyperactivity and inattention) and neuroimaging and genomic data.

Main Methods:

• A cohort of 135 children from the general population provided genomic and neuroimaging data.
• ADHD symptoms were assessed using DSM-IV criteria-based questionnaires.
• A two-step analytical framework was employed: LASSO-ZINB for feature selection, followed by MFA for signal identification.

Main Results:

• Significant associations were found between ADHD symptoms and specific white matter and gray matter regions.
• Relationships were also observed with cerebellar regions and loci on chromosome 1.
• The multivariate framework successfully identified meaningful biological signals in imaging genetics data.

Conclusions:

• Multivariate methods enhance statistical power for neurobiological characterization of complex disorders like ADHD.
• This approach facilitates the identification of complex biological signals in imaging genetics studies.
• The findings contribute to a deeper understanding of the neurobiology of ADHD.