Diagnostic Classification for Human Autism and Obsessive-Compulsive Disorder Based on Machine Learning From a Primate...

Area of Science:

• Neuroscience
• Psychiatry
• Computational Biology

Background:

• Psychiatric disorders like autism spectrum disorder (ASD), obsessive-compulsive disorder (OCD), and attention deficit hyperactivity disorder (ADHD) often present with overlapping symptoms, complicating accurate diagnosis and understanding of shared neural mechanisms.
• Identifying reliable neural markers is crucial for improving diagnostic accuracy and developing targeted treatments for these complex conditions.

Purpose of the Study:

• To identify conserved neural markers across species that can aid in the diagnosis and severity assessment of psychiatric disorders.
• To investigate the utility of nonhuman primate models in discovering neuroimaging biomarkers for human psychiatric conditions.

Main Methods:

• Resting-state functional connectivity data from wild-type and methyl-CpG binding protein 2 (MECP2) transgenic monkeys were used to build diagnostic classifiers.
• These monkey-derived classifiers were applied to human datasets for ASD (ABIDE-I, ABIDE-II), ADHD (ADHD-200), and OCD.
• Stepwise linear regression analyzed associations between functional connections and dimensional symptom severity in human cohorts.

Main Results:

• A classifier based on nine core frontal and temporal cortical regions in monkeys successfully informed diagnostic classification for human ASD and OCD cohorts.
• Specific functional connections, such as right ventrolateral prefrontal cortex to left thalamus, predicted communication scores in ASD.
• Connections involving right prefrontal polar cortex predicted compulsivity scores in OCD patients.

Conclusions:

• The identified core brain regions show potential as a foundation for developing diagnostic markers and symptom severity measures for ASD and OCD.
• These findings suggest that cross-species neuroimaging can inform the development of machine learning models for psychiatric disorders, potentially enhancing clinical assessment speed and accuracy.