Project Grant R01MH143902

The National Institute of Mental Health (NIMH) awarded the University of Southern California a Project Grant of $855,772 under the Mental Health Research Grants program (CFDA 93.242) on July 15, 2026, with a completion date of April 30, 2031. This research initiative aims to characterize autism spectrum disorder (ASD) heterogeneity through machine learning methods while identifying underlying genetic and environmental drivers. The project will utilize an unsupervised machine learning approach called Discriminative Dimensionality Reduction Tree (DDRTREE) to construct clinically relevant ASD subtypes and severity scores based on core symptom presentation and co-occurring conditions, addressing current limitations in diagnosis and treatment planning that result from substantial clinical heterogeneity in the ASD population.

The research will expand upon preliminary findings from the Simons Foundation Powering Autism Research for Knowledge (SPARK) initiative by analyzing approximately 55,000 autistic individuals across multiple cohorts to link identified ASD subtypes and severity variations to specific genetic and prenatal risk factors. In Aim 1, researchers will construct and validate machine learning-derived ASD subtypes and severity scores using symptom and comorbidity data from approximately 45,000 individuals in SPARK, with external validation in approximately 10,000 additional participants. The deliverables will include dimensional ASD subtype classification systems that better reflect real-world support needs and differential patterns of intervention use, advancing understanding of ASD etiology and potentially enabling more precise intervention and treatment planning strategies.