Dennis Wall, PhD, is an Associate Professor of Pediatrics and Psychiatry at Stanford University and is Co-Founder of Cognoa. Cognoa is dedicated to creating an unparalleled standard of care in pediatric behavioral health that ensures equitable access to early intervention by delivering high quality products to improve lifelong outcomes for all children and families.
You’ve long had a personal relationship with someone who has autism. Can you describe this relationship and what you have learned from it?
My sister-in-law has severe autism. I have actually known her since I was a teenager. I went to Boston College High School with her cousin and my friend, but I lived in Cape Cod, which was often a commute of 2 hours each way. After a while, my friend insisted that I start staying at his house during the week, and I thankfully obliged. I became close with their family. That’s how I came to know Becky and the highs and lows of autism.
I went on to complete training in Integrative Biology at UC Berkeley and Computational Genetics at Stanford and then moved back to Boston with my wife, Abby, and started a new faculty appointment at Harvard. I then had the chance to commit my lab and academic research to understanding autism and finding ways to put my training to work to find solutions for families like Becky’s. That led me down a number of roads, most important of which was a fresh look at the realities and challenges with today’s system for autism diagnosis. My personal experience with autism taught me that new innovations and change is greatly needed for many families in how we diagnose and treat autism.
You’ve often been described as the “bad boy” of autism research, do you feel that this reputation is deserved?
I don’t think the description is a universally shared opinion. I was recently just inducted as a fellow into the American College of Medical Informatics and I was named ninth among the world’s top autism researchers. I’ve spent the last 16 years working towards finding better solutions to diagnose and treat autism.
The fact that parents must often wait one to three years to receive an autism diagnosis and begin treatment for their children resulting in many missing a critical, early neurodevelopment window (limiting the effectiveness of treatment), prompted me to use AI and machine learning to develop a solution that would improve outcomes and the current system. Waiting times continue to worsen due to growing shortages of behavioral health specialists. COVID-19 is further exacerbating access challenges to diagnosis. It is clear that we must innovate and embrace new solutions that can help children and their families access diagnosis and care, faster and more efficiently. Cognoa is doing this and my commitment to driving innovation and creating solutions that reach all families in an equitable way and at the right time is my primary concern. Innovation and collaboration across the healthcare community is desperately needed – and it is possible.
You often re-visit Charles Darwin’s seminal book “On the Origin of Species”. What is it about this seminal book that speaks to you so highly?
I love biodiversity. I love to understand the world. I love that Darwin challenged the system with a heretical theory about the way species came and continue to come to be. His book also articulates an algorithm for categorizing organisms through physical, behavioral and morphological characteristics that launched a rigorous field of mathematical foundations to quantitatively define forms of life. The fields of systematics, phylogenetics and population genetics, while seemingly far from the field of medicine, are actually not. Like with complexity of species, conditions such as autism have many forms that require careful description through mathematical foundations like machine learning. This makes the process more objective, data-driven, and also brings us back to quantities and numbers that we can use in very practical ways.
You are the founder behind Cognoa, could you share the genesis story behind this company?
Due to my personal experiences with autism and my professional research, I founded Cognoa with the aim of improving the current system of care for children and families living with autism in ways that will improve their lives. During my training and by shadowing doctors, I witnessed firsthand the immense challenges families face in navigating the current system and particularly how the diagnosis process is misaligned with families’ needs. Far too many families face a long and arduous journey to receive a diagnosis. As a result, children miss a critical neurodevelopmental window when early interventions have the greatest potential to improve lifelong outcomes for children and families living with autism.
I started my faculty position at Harvard with express focus on the molecular basis of autism. We knew (and know), as a field, that autism is genetic and heritable. Yet, the genes responsible were not clear. I collaborated with Boston Children’s Hospital (then the Children’s Hospital of Boston), and after many comprehensive analyses, I began to wonder how the labeling – the confirmation of the class or diagnosis for the child – was done.
So I shadowed clinicians and clinical technicians who performed autism assessments, including observations behind one way glass/mirrors, as the clinical practitioners conducted interviews with parents and children using standard of care assessments, the Autism Diagnostic Interview-Revised (ADI-R) and ADOS (Autism Diagnostic Observation Schedule). I learned that the diagnosis process, while well designed, is extremely long and very difficult to repeat and that subjectivity inevitability crept into the equation. All of these elements motivated me to look for faster ways to perform diagnostic evaluations as quality of life and clinical outcomes of the care system hinges on accuracy and time-to-diagnosis.
I refocused my effort on the phenotype – and on how to build digital quantitative tools to phenotype children with high accuracy, inter-rater reliability, and greater efficiency. I was focused on whether we could reduce the complexity of the autism diagnosis without appreciable loss of accuracy. The answer is definitely yes.
With this, I was able to begin the journey to raise funding for a new company that would start by focusing on improving the journey to diagnosis. Currently, pediatricians refer most children with suspected developmental delay to specialists to diagnose and prescribe treatment. This often results in children and families waiting months or even years before their child receives an initial autism diagnosis and can start life-changing therapy. Cognoa is positioned to fundamentally change this reality by empowering pediatricians to make a diagnosis in the primary care setting and supporting specialists by allowing them to focus on those children with more complex presentations – streamlining the entire system.
Cognoa is highly reliant on machine learning, what initially attracted you to this technology?
Machine learning (ML) is excellent at finding patterns in data. A lot of data is complex and multidimensional – including observational data made on children during an assessment. For diagnosis of behavioral health conditions like autism, there are social measurements, sensory measurements, time dependence and independence, as well as variations in emotions, motions, gestures, responses, and more. All of these components need to be factored into a decision, and ML can help clinicians figure out which particular features correlate to others and which ones don’t. It can help identify the most relevant features that ultimately drive a decision.
We can use ML to cut through the noise to find salient features and produce a mathematical model to generate a digital phenotype for a child. But this is really just a start. What needed to happen next is to determine how to score those features in a way that is accurate, scalable and repeatable.
Cognoa is first and foremost a company committed to improving health outcomes, in particular those concerning child developmental health. We are not a company committed to a particular technique, per se. Machine learning can help us make decisions faster, can help eliminate errors in the process, and critically help to dramatically increase reach to diverse socioeconomic populations, to equal amounts of boys and girls and to all cultures and races. But for that to succeed, the machine’s learning needs to be coupled with the right vehicle for data flow. So Cognoa is also committed to making sure the ML – which has demonstrated to be highly clinically accurate — is accessible through ubiquitous technologies (such as the smartphone) to make the process fast, efficient, and as available as possible. In addition, it is abundantly clear – for all – that technology’s role in addressing healthcare disparities and inefficiencies have only been heightened given COVID-19.
Could you discuss how Cognoa is harnessing the latest AI and Machine Learning technology to introduce a new paradigm of care for autism?
By bringing autism diagnosis into the digital age, Cognoa is pioneering a new approach to autism diagnosis to improve health outcomes, quality of life for families, and by doing so, improving what is today a very complex system. As discussed, currently pediatricians refer most children with suspected developmental delay to specialists to diagnose and prescribe treatment. Many families face a long diagnostic process, often waiting months or even years before their children can receive the diagnosis required to initiate appropriate, life-changing care.
Cognoa’s approach introduces a new paradigm of care for autism by empowering pediatricians to make an accurate diagnosis in the primary care setting. This significantly benefits specialists and children alike as it allows specialists to focus on those children with more complex presentations, as opposed to having long waiting lists for children who are less complex to diagnose, thus enhancing and streamlining the system. The goal is to enable more children to begin autism-specific early intervention sooner, during the early, critical neurodevelopmental window, when interventions have the greatest potential to improve lifelong outcomes for children and families living with autism.
As I mentioned, Cognoa’s AI was also consciously built to embrace gender, racial, ethnic, and socioeconomic origins thereby eliminating the inherent human biases which have historically plagued autism diagnosis. In doing so, Cognoa is democratizing autism diagnosis for all children, to help create equal care for all children.
We are tackling the issue of access to care in two distinct ways. Generally speaking, pediatricians are far more accessible to children and families than autism specialists, of which there is a significant shortage. By enabling and encouraging pediatricians to make diagnoses (in line with American Academy of Pediatrics’ guidelines for pediatricians to begin diagnosing autism in a primary care setting), children and families gain greater access to care. Secondly, the digital nature of our device means that pediatricians will be able to use it to accurately capture inputs remotely via telemedicine. This makes autism diagnosis possible even with geographical or other logistical challenges, such as in the middle of a pandemic.
Can you discuss some of Cognoa’s efforts to use machine learning to make autism diagnosis more objective and efficient?
Research has continually shown that gender, racial, ethnic, and socioeconomic disparities are widely prevalent in autism diagnosis. Girls, for example, are diagnosed with autism an average of 1.5 years later than boys. Furthermore, 1 in 4 children under the age of 8 living with autism, majority of whom are Black or Hispanic, are not being diagnosed at all. This is due to a historical lack of understanding of how autism manifests in girls and minority populations as well as barriers affecting these childrens’ access to care.
While there is often concern that AI perpetuates such biases due to underlying prejudice in the data it is fed, at Cognoa, we have consciously built our AI algorithms to embrace gender, racial, ethnic, and socioeconomic origins so as to tackle these longstanding inequalities. This has involved the use of historical patient data belonging to thousands of children across diverse backgrounds. This data includes a variety of conditions, presentations, and comorbidities, and represents both genders across the supported age range.
Using this wide data-set, Cognoa’s AI assesses thousands of human traits and features enabling it to make the subtle connections that inform current and future behavioral health (conditions) with an accuracy and speed that clinicians simply cannot match, while eliminating inherent human biases.
Could you describe what is Cognoa’s ASD Therapeutic and how it will help improve social-emotional reciprocity?
Cognoa’s autism therapeutic is in clinical development, so I cannot disclose exactly how it will work just yet. I can however share that Cognoa is developing therapeutic solutions that target core deficits of autism such as social emotional recognition and engagement. Cognoa’s digital therapeutic solution for autism has also received breakthrough designation status from the FDA.
I am extremely excited about the use of digital behavioral health solutions to drastically enhance autism treatment and therapy and the lives of children and families living with autism.
Prior to licensing Superpower Glass to Cognoa, in my lab at Stanford, we took the solution from proof of concept to a rigorously tested computer vision tool. The system shows a very significant treatment effect in comparison to controls after only six- weeks of use. The results were published in JAMA Pediatrics last year.
We’ve also invented a new tool (guesswhat.stanford.edu) that, similar to Superpower Glass, augments the child’s reality to encourage prosocial behavior with his or her playing partner. We plan to test it in a randomized controlled trial and ultimately bring it to the general population of children living with autism.
Is there anything else that you would like to share about Cognoa?
In September 2020, Cognoa announced that the company would be submitting its autism diagnostic to the FDA for clearance, following a successful pivotal study in which the device surpassed all FDA benchmarks. This represents an exciting milestone in Cognoa’s mission to improve the lives of children and families living with autism. We are also currently preparing the study results for publication in a peer-reviewed journal and look forward to sharing further progress in the future.
Thank you for the great interview, readers who wish to learn more should visit Cognoa.
- Pexip Collaborating with NVIDIA to Create Immersive Video Meeting Experiences
- Sean Byrnes, Co-founder and CEO at Outlier – Interview Series
- Microsoft Buys Nuance For $19.7 billion
- Deep Neural Network Can Screen for Skin Disease on Laptop
- AI Systems Might Prefer Human Language Instead of Numerical Data