Decoding Autism: AI Foundation Models Promise Precision Pediatric Care

MOUNTAIN VIEW, Calif. and SAN DIEGO – October 15, 2026 — For decades, the treatment of autism spectrum disorder has been defined by a frustrating, albeit necessary, clinical reality: trial and error. Two children sharing the exact same diagnosis might present with identical behavioral challenges, yet respond entirely differently to the same therapeutic intervention. Without objective biological markers to guide them, clinicians have been forced to rely on observational adjustments, a process that costs critical developmental time and exacts a heavy emotional toll on families.

Today, a newly announced strategic collaboration aims to fundamentally change that paradigm. Neumarker, Inc., a venture-backed techbio company known for its artificial intelligence-native brain health platforms, has partnered with Cortica, Inc., the nation’s leading physician-led provider of integrated neurodevelopmental care. Together, they are embarking on an ambitious initiative to apply large-scale electroencephalogram (EEG) foundation models to Cortica’s vast repository of de-identified clinical data. Their shared goal is to discover predictive biomarkers that can definitively guide autism treatment from day one.

From Epilepsy to Autism: Decoding Brain Wave Chaos

The scientific community has long suspected that the key to understanding autism’s clinical heterogeneity lies in the brain's electrical activity. Quantitative EEG, which involves the digital analysis of brain waves to identify atypical patterns, has consistently shown that individuals with autism often exhibit altered neural connectivity and oscillatory power. However, translating these broad research findings into reliable, individual-level predictive biomarkers has proven extraordinarily difficult. The data is simply too messy, and the spectrum of the disorder too wide.

This is where Neumarker’s technology enters the equation. Originally focused on epilepsy—a condition where electrical misfires are relatively distinct and identifiable—Neumarker has been developing its lead application, NMK-001, to optimize clinical workflows and treatment decisions. Expanding this technology into autism requires a leap in computational sophistication. Instead of looking for the obvious spikes of a seizure, the AI must find deeply hidden, dynamic patterns within chaotic datasets.

To achieve this, Neumarker is utilizing AI foundation models. Unlike traditional machine learning algorithms trained for a single, narrow task, foundation models are trained on massive, broad datasets, allowing them to learn the fundamental language of brain activity before being fine-tuned for specific conditions.

“Children with autism can respond differently to the same interventions, yet clinicians today have limited objective tools to help predict treatment response,” said Ken Wang, Ph.D., MBA, co-founder and CEO of Neumarker. “By combining Neumarker’s AI-native brain intelligence platform with Cortica’s deep clinical expertise and longitudinal data, we hope to identify biomarkers that can support a more personalized approach to care. We believe this collaboration represents an important opportunity to extend our platform into autism and other neurodevelopmental conditions.”

Ending the Trial-and-Error Cycle in Pediatric Care

The human impact of this technological leap cannot be overstated. Currently, a child diagnosed with autism may be prescribed a combination of Applied Behavior Analysis (ABA), speech therapy, occupational therapy, or various medications. Finding the right mix and intensity is an iterative process. Months can pass before clinicians and parents realize a specific approach isn't yielding the desired developmental progress.

Cortica was founded in 2017 to address the fragmented nature of this exact system. Operating under a patient-centered medical home model, the organization employs over 2,300 clinicians and serves more than 20,000 patients annually. This scale provides something incredibly rare and valuable in the techbio space: a massive, centralized, longitudinal dataset of clinical outcomes.

“Cortica has built an integrated, whole-child model of care for children with autism and other developmental differences,” said Neil Hattangadi, M.D., co-founder and CEO of Cortica. “We are excited to partner with Neumarker and combine our clinical expertise and de-identified longitudinal data with their advanced AI and EEG technologies. This collaboration has the potential to accelerate the development of precision medicine in autism, helping clinicians deliver the right interventions to the right children at the right time.”

By feeding Cortica's ground-truth clinical outcome data into Neumarker's EEG foundation models, the system can begin to correlate specific baseline brain wave patterns with eventual success in specific therapies.

“Children with autism have highly diverse clinical profiles and responses to treatment, creating an important opportunity to identify biomarkers that can guide more individualized care,” added Kevin Shapiro, M.D., Ph.D., Co-Director of the Cortica Innovation Network. “Our goal is to discover measurable patterns of brain activity that help explain this variability and predict treatment response, bringing greater scientific precision to how we understand and treat autism.”

Data Mining in the Clinic: Privacy and Value-Based Care

While the clinical promise is profound, the partnership also highlights the shifting business dynamics of modern healthcare, where real-world patient data has become an invaluable currency. Training AI foundation models requires millions of data points. Cortica’s ability to provide this data positions it not just as a healthcare provider, but as a critical research engine.

However, mining pediatric health records for commercial AI development necessitates rigorous privacy governance. The collaboration will operate through the Cortica Innovation Network and must adhere strictly to the Health Insurance Portability and Accountability Act (HIPAA). To ensure compliance, Cortica will utilize de-identification protocols—likely stripping the data of specific identifiers such as names, birth dates, and medical record numbers—before it ever reaches Neumarker’s servers.

Ethical data sharing is paramount, particularly when dealing with vulnerable populations. As one digital health policy analyst observed regarding the broader industry trend, the line between quality improvement and commercial data mining is thin, requiring healthcare networks to maintain absolute transparency with patients and parents about how their de-identified data is being utilized to train proprietary algorithms.

Beyond privacy, the economic implications of this partnership are deeply tied to the concept of value-based care. Unlike traditional fee-for-service models that reimburse based on the volume of therapies provided, value-based arrangements reward providers for improving patient outcomes while controlling costs. Cortica already partners extensively with health plans through these arrangements, actively working to reduce costs associated with ABA overutilization, unnecessary medications, and avoidable hospitalizations.

If Neumarker’s AI can successfully generate predictive EEG biomarkers, it would serve as the ultimate tool for value-based care. By predicting exactly which therapy will work for a specific child, Cortica could eliminate the waste of ineffective treatments, thereby dramatically improving their margins under value-based contracts while simultaneously accelerating the child's developmental progress.

The road ahead involves significant regulatory and commercial hurdles. Translating electrophysiological research into a clinically actionable biomarker that insurance companies will actually reimburse requires exhaustive prospective validation. Payers will demand definitive proof of clinical utility—evidence that these AI-driven EEG insights actually change clinical decision-making and lead to measurably better, more cost-effective health outcomes. Yet, if Neumarker and Cortica can successfully navigate these challenges, their collaboration may finally bring the elusive promise of precision medicine to the complex, vital world of pediatric neurodevelopmental care.