Cognoa

Early Autism Detection

Author

Andrew Kuzemczak
September 26, 2025

In partnership with

Typing is a thing of the past

Typeless turns your raw, unfiltered voice into beautifully polished writing - in real time.

It works like magic, feels like cheating, and allows your thoughts to flow more freely than ever before.

With Typeless, you become more creative. More inspired. And more in-tune with your own ideas.

Your voice is your strength. Typeless turns it into a superpower.

Welcome to the 9 new subscribers who have joined Future Human since our last edition! Join 259 other leaders learning about the future of human health by subscribing here:

TL;DR

  1. Cognoa is redefining autism diagnosis by using AI, caregiver input, and video analysis to make evaluations faster, more accurate, and more equitable, collapsing years-long waitlists.

  2. Canvas Dx isn’t just technology—it’s a system that amplifies clinical expertise, providing multidimensional insights while maintaining humility and prioritizing patient-centered care.

  3. Early, scalable diagnosis has massive impact: it improves developmental outcomes for children, reduces unnecessary healthcare spending, and allows families and clinicians to focus on meaningful interventions rather than prolonged uncertainty.

Hi friend,

Welcome back to Future Human! It’s been another whirlwind week at FH HQ (which, to be clear, is not a real place). I’m writing this intro still in physical shambles after running a 50-mile ultramarathon. Not trying to virtue signal here, but… what were you doing last Saturday?

The race went better than I could’ve hoped—my friends and family made up the best support crew imaginable, and I crossed the finish line well ahead of my projected time. Nothing like 10 hours and 48 minutes of running to start (and end?) your day. On the bright side, I’m equally thrilled that my workouts no longer require multiple hours—what am I going to do with all this free time?

On a more professional note: we took last week off to finalize and polish our 30th newsletter. It’s a special one, built around a fantastic interview with Dr. Sharief Taraman, CEO of Cognoa. That conversation was made possible by our research associate Isabelle, who first connected us after an inspiring summer internship with the company. I’m grateful she opened that door, and I’m excited to share this deep dive with you.

So with that, let me ask you:

How could AI and multidimensional data reshape the way we diagnose and treat neurodevelopmental conditions, and what would it mean for families if years-long waitlists were reduced to weeks?

The Story

Autism has long been one of medicine’s most complex frontiers—difficult to study, even harder to diagnose. The condition is defined not by a single biomarker but by a wide constellation of behaviors, leaving researchers to depend on diagnostic “phenotypes” that are often inconsistent and biased. In the early 2010s, Dr. Dennis Wall, a Harvard professor working at the crossroads of genetics, computational biology, and autism research, grew convinced that this foundation was unstable. Large-scale genetic studies were failing to replicate, not because of faulty science, but because the tools used to classify patients were deeply flawed. Gold-standard instruments like the Autism Diagnostic Observation Schedule (ADOS) were designed for research rather than day-to-day clinical care, and they were validated almost exclusively on white, male, affluent children—systematically overlooking girls, minority children, and those from lower-income families. For Wall, the conclusion was stark: the very labeling system used to define autism was broken.

In 2013, Wall founded Cognoa to pursue a radical alternative. Rather than relying on narrow, biased frameworks, he proposed using AI and bioinformatics to surface the most predictive behavioral features of autism from much larger and more diverse cohorts. The vision was clear: make autism diagnosis earlier, faster, and more equitable.

Nearly a decade later, Cognoa’s story intersected with that of its current CEO, Dr. Sharief Taraman. We were fortunate to connect with him for this deep dive thanks to our research associate Isabelle, who had previously interned at Cognoa and was deeply impressed by his leadership and the impact he was making. A couple of weeks ago, I sat down with Dr. Taraman to make newsletter #30 truly special—and I was not disappointed.

An engineer and physician by training, Dr. Taraman described himself as “always the techy kid.” Before medicine, he worked at Ford Motor Company programming databases and solving engineering problems. In medical school and residency, he carried that same inventive streak forward, securing patents for novel medical devices and digital interfaces. Eventually, as a pediatric neurologist at the Children’s Hospital of Orange County, he confronted the reality of autism from the clinic side: endless waitlists for evaluations, with families waiting a year or longer just to get answers for their children.

Dr. Taraman first encountered Cognoa on an AI panel where he met Wall. The technology struck him immediately as a potential breakthrough. If AI could help reframe how autism was identified, it could collapse those waitlists and spare families years of uncertainty. He joined Cognoa first as Chief Medical Officer, where he worked to bridge the science with clinical practice. By late 2022, he stepped into the CEO role.

Cognoa sits at the intersection of Dr. Taraman’s deepest passions—where medicine meets data, and innovative technology can transform patient care. For him, Cognoa is more than a company—it’s an opportunity to finally build scalable solutions that bring timely, accurate autism care to every child who needs it.1

The Tech

At the heart of Cognoa’s work is Canvas Dx, the first FDA-authorized AI diagnostic for autism. Unlike traditional tools that rely on hours of specialist observation, Canvas Dx brings together multiple perspectives and data streams to create a faster, more equitable path to diagnosis.

The process begins with inputs from those who know the child best. Parents or caregivers complete a structured questionnaire about developmental history and everyday behaviors. Physicians—importantly, not just subspecialists but also pediatricians and primary care providers—add their own clinical impressions through a parallel survey. Families also upload short videos of the child at home, in natural environments. This avoids the “white coat” effect of clinical visits, where kids may mask behaviors under observation.

These videos are analyzed by trained human experts, who code subtle behaviors that have been shown to be maximally predictive of autism. Those data are then fed into Cognoa’s AI algorithm, which cross-references patterns against DSM-5 diagnostic criteria. The result is a detailed developmental report outlining the child’s strengths, challenges, and likelihood of autism. Crucially, the turnaround is measured in days, not the months—or even years—that families often face on specialist waitlists.

The performance metrics are striking: a positive predictive value around 95% and a negative predictive value around 97%. Just as important, the system is designed with humility. If the model cannot reach a confident conclusion, it outputs “I don’t know” rather than forcing a potentially misleading result—a safeguard that clinicians and families can trust.

Cognoa’s technology rests on key design principles. It combines perspectives from parents, clinicians, and objective video analysis, reducing dependence on narrow, single-signal tests. This multidimensionality makes the diagnostic process both more robust and more inclusive across gender, race, and socioeconomic lines.

From a regulatory standpoint, Canvas Dx also breaks new ground. Cognoa worked closely with the FDA to secure a “predetermined change control plan,” which means its algorithm can continue to learn and evolve under FDA oversight without requiring full re-approval for every update. This regulatory flexibility ensures the tool stays state-of-the-art while remaining safe and compliant.

Looking ahead, the company is pushing further into advanced computational techniques. Computer vision models are being trained to automate the video annotation process, reducing the need for manual review. Emerging features include eye-tracking, gesture recognition of movements like hand flapping or rocking, and analysis of micro-movements that often elude human observers. Cognoa is also exploring generative AI to synthesize insights into accessible guidance for both families and clinicians.

While autism is the starting point, the platform is designed to scale. The same framework could extend to a wide range of neurodevelopmental and behavioral conditions, broadening access to early, accurate, and equitable diagnoses.

The Market

Cognoa operates in a market defined by overwhelming need, entrenched bottlenecks, and outdated tools. Autism diagnosis today is dominated by the ADOS (Autism Diagnostic Observation Schedule), a gold standard that has become increasingly impractical. ADOS is highly subjective, rooted in norms derived from small populations, and demands hours of specialist time to administer. Other legacy psychometric tools face the same problems: labor-intensive, biased, and slow.

The bottleneck is not just methodological—it is structural. Dr. Taraman highlighted for me the fact that the United States has only about 1,500 pediatric neurologists and 750 developmental pediatricians, the specialists most often tasked with diagnosing autism. With such a limited workforce, families routinely face waitlists of six months to two years just to secure an evaluation. During that time, children miss critical windows for early intervention, which is known to significantly improve long-term outcomes.

Cognoa’s strategy is to shift this paradigm entirely. Rather than relying on a tiny pool of subspecialists, Canvas Dx empowers primary care physicians, nurse practitioners, and physician assistants to conduct autism evaluations directly. This not only accelerates time-to-diagnosis but also allows scarce specialists to focus on complex cases where their expertise is most needed. The model transforms autism care from a bottlenecked, specialist-driven process into a scalable, distributed one.

In terms of direct competition, Cognoa is not alone in seeking to modernize autism diagnostics. Its most notable competitor is EarliTec, a spinout of the Marcus Autism Center and Georgia Tech. EarliTec’s product is also FDA-authorized and relies on eye-tracking technology to detect autism as early as 15 months—slightly younger than Cognoa’s current minimum of 18 months. However, EarliTec’s approach is narrow, depending on a single modality (eye-tracking), whereas Cognoa’s system is multi-modal, integrating questionnaires, video analysis, and clinician input. This broader foundation may ultimately prove more adaptable across diverse populations and additional conditions.

The real competition, though, is not just other startups but the inertia of the status quo. ADOS and legacy psychometrics remain deeply embedded in clinical practice, insurance policies, and academic training. Cognoa’s challenge is not merely technical but cultural—persuading physicians, payers, and regulators that AI-augmented diagnostics can be both accurate and trustworthy at scale. Its FDA authorization and “predetermined change control plan” provide a critical advantage in this respect, allowing the technology to evolve under regulatory oversight.

Beyond autism, the market opportunity is vast. The same diagnostic bottlenecks exist across other neurodevelopmental and behavioral conditions, from ADHD to language disorders. By establishing a platform that extends AI-powered evaluation into primary care, Cognoa has positioned itself not just as an autism company, but as a potential category leader in next-generation behavioral health diagnostics.

The Sick

For families, the journey to an autism diagnosis has long been one of the most painful parts of the condition itself. In the current system, the path usually begins when parents notice something different about their child—delays in speech, lack of eye contact, or unusual behaviors. Their pediatrician refers them to a developmental pediatrician or neurologist, where the average wait can stretch from six months to two years. During that limbo, parents often bounce between irrelevant specialists—ophthalmologists, neurologists, even geneticists—racking up bills and anxiety without getting closer to answers. All the while, their child misses the most important window for early intervention, between ages zero and five.

Canvas Dx fundamentally rewrites that story. Instead of years of uncertainty, families can now complete the diagnostic process in a matter of weeks within primary care. Parents are not passive bystanders but active participants: they fill out questionnaires, record videos of their child at home, and contribute observations that become central to the evaluation. This engagement does more than streamline the process—it helps families reach acceptance earlier, which is critical for engaging with therapy rather than remaining stuck in cycles of doubt and delay.

The impact extends beyond timing. Canvas Dx produces customized developmental reports that point families toward the right therapy at the right time. A child with sensory challenges may be directed toward occupational therapy, while a child with social communication differences may benefit from behavioral interventions. And because the system has been validated on diverse populations, it reduces the historical bias that left many children underdiagnosed or misdiagnosed.

Clinicians also see a tangible difference. Instead of spending hours collecting data and scoring assessments, they receive a structured, AI-generated developmental profile. That efficiency allows psychologists and pediatricians to shift their energy from paperwork to meaningful conversations with families. It also expands their capacity: a psychologist who might traditionally see four to six children a week could potentially evaluate multiple children in a single day using Cognoa’s platform.

For children themselves, the stakes could not be higher. Decades of research show that early intervention profoundly changes developmental trajectories. One study in JAMA that Dr. Taraman and I discussed found that 37% of children who received early therapy no longer met the criteria for an autism diagnosis by the time they reached elementary school. Early support also reduces long-term risks such as malnutrition, accidental injuries, and behavioral crises.2

In short, Canvas Dx is not just a diagnostic tool—it is an accelerator of care, a reducer of anxiety, and a bridge to better outcomes. For parents, it replaces uncertainty with clarity. For children, it replaces lost years with opportunities to thrive.

The Economy

The economic burden of today’s autism diagnostic system is staggering—not only for families but for healthcare systems at large. Claims data that Dr. Taraman and I discussed shows that each undiagnosed child generates an additional $330 per month in excess healthcare costs. Across the roughly 1 in 36 children affected by autism in the U.S., this translates into tens of millions of dollars annually in avoidable spending. Those dollars are not buying meaningful care; they are consumed by redundant referrals and unnecessary workups—trips to ophthalmologists, neurologists, or geneticists that rarely bring families closer to answers. Multiply that figure across hundreds of thousands of children stuck on waitlists, and the waste is enormous.

Canvas Dx offers a direct path to cost savings by collapsing that inefficient journey. By delivering a reliable diagnosis within weeks in primary care, it eliminates the merry-go-round of specialist visits and expensive testing. Families move more quickly into evidence-based therapies, where dollars translate into meaningful developmental gains rather than diagnostic dead ends. Studies suggest that early intervention, like the one Canvas Dx allows, can help children eventually use less occupational, physical and speech therapy and fewer intervention services leading to $19,000 in cost savings per year for each child.3

The downstream financial implications are even more significant. Delayed diagnoses often lead to preventable complications that require acute care. Restricted eating can progress to severe nutritional deficiencies like scurvy; children who elope without supervision face risks of accidents, injuries, or even drowning. These scenarios can easily result in ER visits costing thousands per incident. Faster diagnosis and intervention mitigate these risks, avoiding both financial and human costs.

Cognoa has also begun to generate real-world evidence for its economic value. In a pilot program with CalOptima, the Medicaid health plan for Orange County, Canvas Dx demonstrated effectiveness in both primary and specialty care settings. The results showed not only improved access and operational efficiency but also alignment with the broader health system goals often described as the Quadruple (or Quintuple) Aim: lower costs, better health outcomes, higher family satisfaction, higher clinician satisfaction, and improved health equity. By reducing administrative burden and burnout among clinicians while reaching historically underserved populations, Cognoa’s model addresses every dimension of that framework.

At a macroeconomic level, the opportunity is transformative. Instead of pouring billions into prolonged diagnostic paths, healthcare systems can redirect those dollars into actual care delivery—therapies, educational supports, and family services that produce measurable improvements in outcomes. And because Canvas Dx enables autism evaluations in primary care, it offers a scalable way to expand capacity without needing to dramatically increase the supply of highly specialized clinicians, who are already in critically tight supply. For context, the U.S. has fewer than 8,000 board-certified child neurologists and developmental pediatricians combined, far below what would be needed to meet current demand using traditional diagnostic pathways.

In short, the economics of Cognoa’s model are about efficiency and redirection. Every month shaved off the diagnostic timeline represents not just reduced costs but reinvestment in interventions that can change a child’s life trajectory. For payers, providers, and families alike, the financial logic is as compelling as the human one: spend less chasing a diagnosis, and more on delivering the care that matters.

My Thoughts

Having the opportunity to sit down with Dr. Sharief Taraman offered one of the most thought-provoking conversations I’ve had in a long time, and I’m truly grateful he made the time. 

Our discussion highlighted not just the delays in autism diagnosis, but the deeper structural and epistemological limits of medicine—how our reliance on narrow observational frameworks, small datasets, and biased norms can unintentionally leave families navigating years of uncertainty. What fascinated me most was seeing how a system like Canvas Dx approaches these problems from a fundamentally different angle: it treats data as a multidimensional conversation to surface patterns that human eyes alone might never catch. The elegance also lies in its humility—acknowledging uncertainty with an “I don’t know” rather than overconfidently asserting answers—reminding me that responsible AI in medicine is as much about caution as it is about efficiency. It made me reflect on a bigger question: if technology can extend our capacity to act in ways that humans alone cannot, how do we redesign healthcare so that these new tools serve the patient rather than the system? Sitting with Dr. Taraman reinforced that the future of health tech isn’t about replacing clinicians; it’s about amplifying insight, redistributing scarce expertise, and ultimately making care not just faster or cheaper, but more just and humane.

To more lives saved,

Andrew, Isabelle, and Nicholas

I always appreciate feedback, questions, and conversation. Feel free to reach out on LinkedIn @andrewkuzemczak.

References