Prediction Model for Sensory Perception Abnormality in Autism Spectrum Disorder

Area of Science:

• Neuroscience
• Genetics
• Biomarker Discovery

Background:

• Autism spectrum disorder (ASD) presents diverse phenotypes, often including sensory perception abnormalities impacting quality of life.
• Current diagnostic methods lack specific biomarkers for ASD-related sensory issues, hindering early detection.

Purpose of the Study:

• To identify potential biomarkers for ASD sensory perception abnormalities.
• To construct a predictive model for early ASD identification and screening.

Main Methods:

• Screened differentially expressed genes from the Gene Expression Omnibus database for sensory perception relevance.
• Utilized enrichment analysis and random forest for gene selection.
• Developed an artificial neural network prediction model and validated it in an ASD mouse model.

Main Results:

• Identified 16 genes associated with sensory perception abnormality in ASD, linked to actin, cholesterol metabolism, and tight junctions.
• Selected 15 disease-characteristic genes for model construction.
• Achieved high accuracy (AUC=0.999 for training, 0.711 for validation) in the prediction model, confirmed in BTBR mice.

Conclusions:

• Developed a highly accurate prediction model for ASD sensory perception abnormality using 15 identified genes.
• This model offers a novel approach for early identification and diagnosis of sensory issues in ASD.