Ella Swanson-Hysell is a product manager at EarliTec Diagnostics, where she works to revolutionise early autism detection through cutting-edge technology. Her diverse background in both the social sciences and technical fields, combined with a passion for child development, drives her work in creating diagnostic tools that aid clinicians in identifying autism traits in young children. With a rich history in data science and a recent master’s degree in software engineering, Ella’s expertise allows her to bridge the gap between technical innovation and clinical application.
“Data science can make a tangible difference in the lives of children with autism, and I’m thrilled to be part of a team making that possible.”
A passion for interdisciplinary work
My academic journey began at Lewis & Clark College in Portland, Oregon, where I was fortunate enough to study a combination of math, computer science and psychology, focusing particularly on child development and education. Even then, I knew I didn’t want to choose one discipline over another. I wanted to explore how they could work together. While I joked that my ideal job would involve teaching calculus to second graders, I found that merging my interests in the sciences and education would lead me to a career where I could truly make a difference.
Early exposure to research
After graduating in 2015, I was fortunate to join the Marcus Autism Center in Atlanta, which opened my eyes to the world of academic research. Working on a longitudinal study of autism development in infants and toddlers, I was able to combine my technical skills with my passion for child development. I learnt from some of the brightest minds in the field, and my experience there helped me realise that you don’t have to limit yourself to one area of interest. You can find ways to apply your diverse skill set to a cause you care deeply about.
From data science to product management
In 2021, I joined EarliTec Diagnostics as a data scientist, returning to the project I had worked on during my time at the Marcus Autism Center. My role initially focused on refactoring a research software tool into a more accessible product — a 15-minute diagnostic assessment for autism, cleared in the US by the FDA [Food and Drug Administration] for children aged 16 to 30 months. Over time, my responsibilities expanded and I transitioned into a product management role. Today, I oversee the refinement and expansion of our diagnostic tool, working closely with clinicians, business teams and developers to ensure the product meets the needs of those who rely on it.
Balancing motherhood and career
In 2023, I completed a master’s degree in software engineering with a certificate in data analytics, all while balancing my career and raising two young children. It was challenging, but also incredibly rewarding. I spent many evenings and weekends learning, sometimes while holding a toddler or nursing a baby. It’s this balance that’s shaped my perspective — I’ve learnt to manage my time effectively, but I’ve also come to appreciate the flexibility and support that technology can offer in achieving work-life balance.
A life-changing product
At EarliTec Diagnostics, our EarliPoint Evaluation for Autism is a unique diagnostic aid that uses eye-tracking technology to assess a child’s social-visual engagement with various videos of children interacting. It’s a 15-minute tablet-based assessment that can deliver an accurate diagnosis of autism, providing clinicians with severity measures and visual data that they can use to better understand a child’s behaviours. What excites me most is that we are making autism diagnosis accessible and efficient, helping children receive the support they need at an earlier stage in their development.
The future of early diagnosis
I’m excited about the potential for our EarliPoint Evaluation to make a difference in the lives of children and families. Autism diagnosis has historically been a lengthy and uncertain process, but we are changing that. By integrating technology into the process, we can offer a faster, more objective way of identifying autism, providing clinicians with a tool that supports them in their work and ultimately helping children get the support they need at a critical age.
The story of autism that many people are familiar with now is that the earlier you can diagnose it, the earlier you can intervene and the less challenging it might be, as autism is highly variable. However, in many cases, it presents significant challenges. What we’d really like to see is early diagnosis and intervention for every child, everywhere, to optimise the best possible outcome for each individual.
Unfortunately, right now, the gold standard diagnostic process in the US requires about nine hours of a clinician’s time. Often, this process spans multiple appointments for families and there are not enough qualified clinicians available in every geographic area or under every health insurance plan. This results in multi-year long waiting lists, long drives and extreme delays. It is not accessible and it is not fast. (See Autism in Numbers and Stories).
The mission for our company and its founders, Dr Ami Klin and Dr Warren Jones, is to make this process efficient, objective, consistent and accessible. If there is an EarliPoint device in an under-resourced area where qualified clinicians are not available, it can greatly enhance accessibility for those children. This assessment is much faster than the gold standard and is as accurate, as shown in our JAMA and JAMA Network Open publications. It also avoids the potential inconsistencies in training for traditional assessments, which can have so much variability that it leads to disagreements between two clinicians for the same child.
Additionally, many traditional assessments are not focused on children as young as 16 to 30 months. Many otherwise qualified clinicians are not equipped to diagnose or lack the right assessments for this age range. This creates a floor effect for the age of diagnosis, where they only see children when they are older, like three or four years of age, and many children are not flagged or diagnosed until they are school-aged.
Though very few clinicians feel confident and qualified to diagnose at these younger ages, scientific literature shows that autism is already developing and causing differences in behaviour in measurable ways as early as infancy. Having a tool like EarliPoint can help to lower the age at which children are diagnosed, which can be very meaningful for their outcomes.
Hurdles overcome in collecting data
Medical technology for young children presents unique challenges. One of the first challenges we had to consider was designing a system that could work with children who may shift positions during the assessment. Since some of the symptoms that could bring a child in for an autism assessment, like repetitive motor behaviours or hyperactivity, might involve such movements, the system had to be flexible enough to accommodate children who may not sit still in one position throughout the assessment.
We needed to ensure the eye-tracking data collection system was sensitive enough to work with children who might shift positions. This flexibility was built into the design of the product, the placement of the eye tracker, the instructions for use and the analysis process.
In our initial clinical trials of over 1,500 children, 95% of children were able to complete the data collection, and we found similar rates in our first year as a commercial product. We’ve been very successful in terms of data collection and designing a system to work for all children.
Ensuring the system works for diverse populations
As for ensuring the system works for children from diverse backgrounds, this is a crucial question. We are aware that when it comes to AI, it’s important to ask certain questions, particularly regarding bias. We cannot develop a system and claim it will work in a specific population unless it has been trained and tested with that population.
In our clinical trials, we intentionally collected data from a diverse group of children to ensure the system was representative of the populations we aim to serve. We have always been mindful of the need for the system to work across racial, socioeconomic and cultural backgrounds. We are directly addressing the potential for bias in our system and comparing it to known biases in the clinical setting. Our device offers a more objective experience, allowing children to interact without the social pressures that might influence behaviours in a clinical setting.
Diversity of thought in technology development
Diversity of thought is crucial in developing new technologies. It’s easy to focus on accessibility and equity, but without the right people or viewpoints in the room, key design decisions can be overlooked. I once worked on an eye-tracking project in a previous role where we initially used a system with a baseball cap for eye tracking, only to realise that children with afros couldn’t wear the cap. That led to a quick redesign into a visor that worked for more children. It’s a simple example, but it highlights the importance of considering different body types, experiences and sensitivities in the design process.
We have to recognise that one child is not the same as another child, and that families and their needs vary. That’s why having people with experience working with children in different settings has been crucial in making our products accessible and effective.
Since we’ve been intentional about preventing bias in the tool, it helps to provide a more honest reflection of how children see the world. Our system doesn’t have the same biases that may exist in traditional clinical interactions, where cultural differences between the clinician and child might affect how a child behaves. We are proud that our system works well in diverse populations, and we continue to analyse and refine the tool to avoid any potential bias.
Rapid advances in developmental health technology
We are entering an era of rapid innovation in developmental health technology, where tools designed to enhance childhood assessment and intervention are set to transform the landscape in the next decade. While AI plays a significant role in this, at EarliTec, we ensure it’s not just another buzzword. We prioritise interpretable AI, meaning every step of our process is transparent and clinically explainable. There are no black boxes here — just clear, data-driven insights that clinicians can trust and understand.
In the space of autism and behavioural health, the scientific and clinical rigour of tools can vary greatly. Some approaches may not be regulated as medical devices, meaning they don’t have to meet the same rigorous clinical standards as FDA-cleared tools. It’s important to scrutinise the quality and accuracy of AI tools, especially those claiming to be tools for clinical diagnosis or treatment.
For EarliTec, our interpretable AI is medically and scientifically rigorous, and we continue to raise the standards in autism assessment. The understanding of autism and effective treatments is constantly evolving, and we will continue to see changes and new tools in how treatments are delivered, how they are funded, and how they impact children and families.
There’s also a concern that AI might replace human jobs, particularly for psychologists and clinicians. We don’t view this as our goal. At EarliTec, we respect the essential role that clinicians play, and we believe our tools empower providers to help more children in more meaningful ways, efficiently. Our tool is not about replacing human interaction, but about making the diagnostic process more accessible, particularly when current methods are so time-intensive and inaccessible. We are ethically obligated to consider how our tools can improve child outcomes, and we are confident that they can empower, not replace, clinicians’ work.
What’s next?
We’re excited to announce that EarliTec is completing large clinical trials to evaluate EarliPoint as a diagnostic tool for children up through age seven. Some of our current customers struggle to schedule younger children due to long waitlists for older children, even though paediatricians are flagging younger children for assessment. To assist with this backlog and to serve as many children as possible, we are validating our tool for this older age range.
Stay tuned for more updates from us. We also have exciting publications coming soon that will share how EarliPoint is being used in diverse clinical settings today. We’re eager to share these findings and continue to impact more children’s lives through earlier diagnosis and more objective, standardised measurements.
Reference links
- https://www.linkedin.com/in/ella-swanson-hysell/
- https://earlitecdx.com/
- https://www.linkedin.com/company/earlitec/
- https://autism.gatech.edu/
- https://jamanetwork.com/journals/jama/fullarticle/2808996
- https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2808909
- https://www.nature.com/articles/nature12715
