Many autistic people live with avoidant or restrictive eating patterns: intense sensory aversions to certain foods, very limited diets, or little appetite. This pattern, called avoidant/restrictive food intake disorder (ARFID), can disrupt nutrition, social participation, and daily life.

Using SPARK surveys and DNA from over 5,000 participants, we produced the first genetically informed study of ARFID in a large autism cohort. We estimated that roughly 21 percent of autistic individuals meet criteria for ARFID, a figure at the upper end of previous clinical estimates, suggesting widespread under-diagnosis. The study also found that about 45 percent of the variation in ARFID risk comes from genetics, and it identified a signal near the gene ZSWIM6, which is linked to other neurodevelopmental conditions. We also found that ARFID shares genetic roots with autism, anxiety-related personality traits, and metabolic conditions.

Read the full paper in Frontiers in Psychiatry (Koomar et al., 2021)

Most autistic people struggle with sleep. Poor sleep compounds challenges with attention, emotion regulation, and daily functioning, yet the biological roots of sleep problems in autism remain poorly understood.

We grouped thousands of SPARK families into shared "sleep types" by sorting participants into distinct profiles based on their sleep patterns. By linking these profiles to participants' DNA, we are uncovering the genetics behind different sleep difficulties and pointing toward treatments matched to each type. The goal: move from "you have a sleep problem" to "you have this kind of sleep problem, and here is what is most likely to help."

Read the sleep archetypes preprint (Brueggeman et al., 2025)

Which treatments help, and for whom? Most medications prescribed for challenges associated with autism were not developed specifically for autistic people. Whether a given medication helps a given individual often comes down to trial and error.

We are working to change that. Using SPARK data and machine learning, we developed a method for ranking the medications most likely to benefit a given individual based on their genetic profile. The approach uses deep learning (a branch of AI) to find genetic patterns that predict which participants respond well to which medications, a step toward medicine tailored to the individual.

This work was presented as a poster, "Polygenic Signatures of Medication Prioritization Using Deep Learning: Application in SPARK Autism Study," at the ACNP Annual Meeting in Tampa (December 2023), and continues.

A strong sense of identity and personal strengths protects mental health across the lifespan. In autism, clinical attention has historically centered on deficits, leaving the strengths side of the picture poorly mapped.

In a study of over 1,000 clients from a specialty clinic for high-ability students, we found that autistic individuals with exceptional cognitive ability are nearly 15 times more likely to reach adulthood undiagnosed. Self-reported sense of inadequacy and anxiety both rose with IQ in autistic clients, the opposite of the pattern seen in non-autistic individuals. A characteristic cognitive profile, slower processing speed paired with very high verbal comprehension, was linked to autism, inattention, and internalizing problems. That profile also had a significant tie to genetic risk for autism. We replicated these findings in SPARK (N=17,634) and ABCD (N=10,602).

The takeaway: high ability does not protect autistic people from mental health challenges, and recognizing strengths requires understanding the full cognitive and emotional picture. Related work was presented as an invited talk at the Kennedy Krieger Institute / CASSI Autism Acceptance Month Miniseries (April 2025).

Read the preprint (Michaelson, Doobay, et al., 2022); the manuscript is currently in peer review.

Many SPARK participants are "twice-exceptional" (2e): they have both disabilities and profound creative or intellectual abilities. These individuals are often underserved because their abilities can mask their challenges and their challenges can overshadow their abilities.

We are studying how creativity appears in autistic children and what genetic factors contribute to it. Using SPARK data and related assessments, we are measuring creativity as a trait that can be scored and compared, then linking it to genetic differences across participants. The long-term aim is to help educators and clinicians support creative development in 2e youth.

This work was first presented at the Scanlan Center for School Mental Health Research (Iowa City, February 2023): "Links between behavior and creativity in autistic children, and their genetic predictors." We are preparing a paper for peer-reviewed publication.

What helps twice-exceptional students thrive in school and beyond? Many autistic individuals with high cognitive ability do well academically but face distinct challenges: social isolation, burnout, and difficulty accessing support systems designed for students who struggle in more visible ways.

We are studying talent, learning differences, and long-term outcomes in twice-exceptional individuals, with the aim of helping families and educators support high-ability autistic children as they grow. This work draws on SPARK data and connects to a broader program examining how autism and high ability interact.

One published finding: autism combined with high cognitive ability is associated with elevated suicidal ideation. High IQ alone does not guarantee good outcomes.

Read the paper in Neurobiology of Learning and Memory (Casten et al., 2023)

This work has been presented at the Wallace Research Symposium (Storrs, CT, May 2024), as a keynote at the Belin-Blank Twice-Exceptional Summit (May 2021), and at the 2e22 Conference (February 2022).