About the lab
A research lab, not a product company.
AutismAI Lab is a small research group building AI tools that bring research-grade assessment within reach of families, clinicians, and researchers using devices they already own. We treat published methods as the input and clinically-grounded, openly documented tools as the output.
Why we exist
Assessment-grade tools, in the browser.
Autism spectrum disorder affects 1 in 36 children in the US,[Maenner et al., 2023] and motor and cardiorespiratory differences extend the same population into adulthood and aging. Yet most assessment tools are subjective, time-consuming, and inaccessible to those who need them most. They sit behind specialist clinics, expensive hardware, and long waitlists.
We believe AI can help bridge these gaps, not by replacing human expertise, but by augmenting it with objective, scalable, and accessible tools that run on devices people already own.
The lab was founded to make that vision concrete. We translate peer-reviewed methods into deployed tools, validate them against clinical references, and publish the methodology so peers can scrutinise and extend our techniques.
- →Browser-native assessment tools that work without specialist hardware.
- →Methods documented publicly so peers can scrutinise and extend them.
- →Privacy by construction: inference on-device, raw data stays local.
- →Continuous statistical layer (GAMMs, GP normative modeling) shared across every tool.
Our story
A small lab, with a personal origin.
Founded 2025 · University of Kansas
AutismAI Lab was started by a parent of a child on the spectrum who also happened to be an AI architect. The original motivation was unglamorous: the tools the family needed (for adaptive learning at home, for objective motor screening, for low-friction in-home monitoring) were not yet available in a deployable form. So they had to be built.
The lab is small and intentionally so. A lead researcher (PhD candidate in AI at the University of Kansas) carries the build work, drawing on the published literature in motor neuroscience, special education, signal processing, and disability-inclusive design, and on the peer review that each manuscript passes through.
What has changed in the last five years is the methods stack. Computer vision, large language models, modern signal processing, and rigorous statistical modelling have matured to the point where research-grade assessment can run on devices people already own. The lab's job is to translate that maturity into tools that families, clinicians, and researchers can actually use, and to publish the methodology so peers can scrutinise it.
Two tools, Grovalin and Volitiq, are the practical output of that translation so far; WAVE and PULSE are proposals that have not started data collection. They are released as research instruments, not products: built with peer-reviewed methods, validated against clinical references, and shared with the community that motivated them in the first place.
People
The lab + collaborators.
The lab is one researcher deep by design. The work that needs more than one person is done in the open: through peer review, through published methods, and through collaborations formed around a specific study.
PhD (ABD) in AI at the University of Kansas; dissertation on attention-guided motion profiling for autism detection and intervention personalisation. Enterprise AI architect with 20+ years building production ML and LLM systems across healthcare and life-sciences data; AWS-certified (Solutions Architect + ML Specialty). Parent of a child on the spectrum, the original reason the lab exists.
LinkedIn →Every research line here depends on domain knowledge the lab does not hold on its own. Four areas in particular shape the work:
- →Motor and oculomotor neuroscience — what a kinematic difference actually means clinically.
- →Special education and autism intervention — whether a learning design is worth a child's time.
- →Signal processing and applied ML — whether a measurement is real or an artifact of the pipeline.
- →Lived experience — autistic self-advocates, caregivers, and clinicians who use these tools.
Named collaborators and advisors will be listed here once those collaborations are formally confirmed.
Work with the lab →How we got here
Timeline
Early development of two research prototypes at the University of Kansas: E-MotionSpec for video-based motor analysis, and ContextLearn for AI-assisted contextual learning. Both would later move under the AutismAI Lab umbrella.
Founded by Shailesh Pandey (PhD candidate at the University of Kansas, EECS) to consolidate the research-grade tooling under a single brand and bring assessment-grade tools to anyone with a device.
First deployed tool: quantitative motor-movement analysis (rebrand and refactor of E-MotionSpec) launched at volitiq.com.
AI-powered contextual learning platform (rebrand of ContextLearn) moves into active development at grovalin.com.
The lab's work extends from the clinic into the home, with two complementary passive-sensing modalities: WAVE (WiFi channel-state inference) and PULSE (RGB rPPG vitals).
GAMM trajectory analysis, GP normative modeling, and individual deviation scoring formalised as the shared statistical layer across Volitiq, HEARTH, and Grovalin.
Contact
Get in touch
Questions, ideas, collaboration inquiries: reach out through any of the channels below, or fill out the collaboration form.
- Apps →Grovalin, Volitiq, WAVE, PULSE: what's live and what's proposed.
- Research →Active research lines and the statistical foundation shared across all programs.
- Publications →Peer-reviewed references our tools build on, and the lab's own papers.
- HEARTH report →The full long-form report on the WAVE + PULSE program.
- Collaborate →How developers, researchers, clinicians, and families can contribute.
Sources
References
- Maenner, M.J., Warren, Z., Williams, A.R., Amoakohene, E., Bakian, A.V., Bilder, D.A., et al. (2023). Prevalence and Characteristics of Autism Spectrum Disorder Among Children Aged 8 Years — Autism and Developmental Disabilities Monitoring Network, 11 Sites, United States, 2020. MMWR Surveillance Summaries 72(2), 1–14. doi:10.15585/mmwr.ss7202a1