Artificial Intelligence in Autism Detection
A.I. in Autism is the ultimate "A.I for Good" R&D in healthcare.
As you know I care a lot about the topic of A.I. in healthcare, and the future of healthcare.
Still with regards to Autism, even women with Autism or on the spectrum, often go mostly (or late) undiagnosed.
Do you care about the A.I. for good movement? I most recently covered in the A.I. intersection of Healthcare the following topics:
The Future of A.I. in Healthcare. ~ HERE.
Artificial Intelligence is Taking on Parkinson's Disease. ~HERE.
Artificial Intelligence Helps Cut Miss Rate of Colorectal Polyps. ~ HERE.
A.I. Advances in Treatment Of Spinal Cord Injuries and Surgery. ~ HERE.
Future of A.I. in Neurosurgery. ~ HERE.
Artificial Intelligence Could Help Detect Onset of Cardiovascular Disease. ~ HERE.
Can A.I Improve our Breast Cancer Screening?~ HERE.
Artificial Intelligence is Changing the Future of Radiology. ~ HERE.
So let’s get into it.
The painful reality for Autism is simple: Diagnostic criteria are developed using white boys and men, failing to serve many neurodivergent girls and women. However A.I. is bringing something new to the table.
Artificial Intelligence Understands Autism from Brain Fingerprints
A new AI algorithm can predict whether a person is on the autism spectrum by examining their brain scans. The algorithm can also predict the severity of symptoms and could be used as an early detection tool for ASD.
Patients with autism who are diagnosed early and definitively may benefit from earlier therapies and better results. The breakthrough is brought to you out of Stanford.
How Prevalent is Autism?
Think about it, Over 1 million children under the age of 17 in the US are on the autism spectrum. These children often times fail to recognize basic facial emotions, which make social interactions and developing friendships even more difficult to sustain. Gaining these skills requires intensive behavioral interventions that are often expensive, difficult to access, and inconsistently administered.
In Canada, the “neuro-divergent” are seen as a potential help cure to the labor demand crisis. Our understanding and treatments for autism are still moderately poorly understood.
CDC researchers reported that autism rates in the United States increased from 1 in 150 children in 2000 to 1 in 54 in 2016, and the rate now stands at 1 in 44 children. So is Autism increasing or is just becoming diagnosed better? I’m no expert, but I think it’s possibly a bit of both.
Back since about 2017, the rise in the rate has sparked fears of an autism ‘epidemic.’ But experts say the bulk of the increase stems from a growing awareness of autism and changes to the condition’s diagnostic criteria.
Stanford’s AI on Autism “Fingerprint”
New AI-Driven Algorithm Can Detect Autism in Brain “Fingerprints”
What this suggests is that an early, definitive mechanism of detection of autism in patients could lead to timelier interventions and better outcomes. A.I. is likely to offer such a method. As our understanding of neuroscience and A.I. converge, early detection in a wide range of conditions could become possible.
As Professionals those with Autism must fight just to live a normal life. According to a 2012 Canadian Survey of Disabilities, 83 per cent of adults with autism reported no employment income. In fact, according to Jill Farber, Autism Speaks Canada's executive director, the 14.3 per cent employment rate for people with autism is well below the 45 per cent employment rate for people of all disabilities, and they are often compensated “well below minimum wage.”
Do you care about this issue or topic?
Now in 2022, Stanford scholars have created an algorithm that uses functional magnetic resonance imaging scans to find patterns of neural activity in the brain that indicate autism. A.I. can use fMRIs to detect Autism more easily.
The novel algorithm, driven by recent advances in artificial intelligence (AI), also successfully predicts the severity of autism symptoms in individual patients. With further honing, the algorithm could lead to earlier diagnoses, more targeted therapies, and broadened understanding of autism’s origins in the brain.
The algorithm pores over data gathered through functional magnetic resonance imaging (fMRI) scans. These scans capture patterns of neural activity throughout the brain. By mapping this activity over time in the brain’s many regions, the algorithm generates neural activity “fingerprints.” Although unique for each individual just like real fingerprints, the brain fingerprints nevertheless share similar features, allowing them to be sorted and classified.
The lead of the paper is Kaustubh Supekar. He’s a Clinical Associate Professor, Psychiatry and Behavioral Sciences at Stanford.
Will A.I. Help us Diagnose Autism at Scale?
As described the new study (Feb 15th, 2022) published in Biological Psychiatry, the algorithm assessed brain scans from a sample of approximately 1,100 patients. With 82% accuracy, the algorithm selected out a group of patients whom human clinicians had diagnosed with autism.
“Although autism is one of the most common neurodevelopmental disorders, there is so much about it that we still don’t understand,” says lead author Kaustubh Supekar, a Stanford clinical assistant professor of psychiatry and behavioral sciences and Stanford HAI affiliate faculty. “In this study, we’ve shown that our AI-driven brain ‘fingerprinting’ model could potentially be a powerful new tool in advancing diagnosis and treatment.”
It’s weird that Autism is such a common neurodevelopmental disorder that we don’t fully understand, I think in the coming years A.I. has a big possibility to help us in understanding it better.
It’s believed that with further honing, the algorithm could lead to earlier diagnoses, more targeted therapies, and broadened understanding of autism’s origins in the brain.
AI Supremacy is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
A.I. Neuro Fingerprinting
Beyond Autism this ability of A.I. to fingerprint our brains is interesting and could likely lead to other kinds of applications. Think about it, the algorithm pores over data gathered through functional magnetic resonance imaging (fMRI) scans. These scans capture patterns of neural activity throughout the brain. By mapping this activity over time in the brain’s many regions, the algorithm generates neural activity “fingerprints.” Although unique for each individual just like real fingerprints, the brain fingerprints nevertheless share similar features, allowing them to be sorted and classified. To some, it might even be scary.
However Neuroscience intersecting with A.I. is one of the most exciting fields I think in the 21st century and likely the key to unlocking AGI in the technological singularity.
The Problem of Tracking Autism in Particular
Unlike many other diseases, autism lacks objective biomarkers—telltale measurements that reveal a medical condition’s presence and sometimes severity—meaning there is no simple test for the disorder. Instead, diagnosis is based on observing patients’ behaviors, which are naturally highly variable and thus make diagnosis a challenge. (Common signs of autism include difficulty navigating everyday social interaction, deficits in communicating and learning, and repetitive speech and motions.)
“We need to create objective biomarkers for autism,” says Supekar, “and brain fingerprints get us one step closer.”
Apparently boys are 4x more likely to be diagnosed with ASD (autism spectrum disorder) than girls. This could be also due to bias.
Read this account in Scientific America about the problem of female diagnosis of Autism and ASD. It is pretty enlightening. I wonder if A.I. could remove some of the bias. (The Author is a professor, a screenwriter, producer, mother and a woman who has autism.).
How A.I. uses fMRIs is also going to improve.
Scientists have long searched for biomarkers via fMRI scans. Yet studies to date with small populations have reported conflicting results, stemming from a natural variability in patients’ brains and confounded further by differences in fMRI machines and testing methods.
The Big Data of Healthcare and Our Health Data is Coming
Like many scientific fields, autism research has embraced the big data approach, Supekar says, where previously unobtainable insights emerge from analyzing large, statistically powerful samples. The new study is a case in point, pooling brain scans from medical centers worldwide into a mammoth, demographically and geographically diverse dataset.
People with ASD are Chronically Underemployed
"People with autism are very much capable of working and they are some of the best employees," said Neil Forester who, along with his business partner Xavier Pinto, created the Spectrum Works Job Fair that ran Friday. This CBC article really caught my attention on this topic (April 10th, 2022).
It’s heart breaking but Only 1 in 3 people with autism are employed. But many more of those on the spectrum want to work.
To this day, The DSM doesn’t distinguish between subtypes of autism, including Asperger’s syndrome. That’s in the 2020s not just outdated, it’s barbaric.
So What’s Next?
The next step was to effectively parse and deal with the data complexity and variability. Supekar and colleagues thought a good place to start would be image recognition algorithms, developed by technology companies. These algorithms have grown increasingly sophisticated at handling significant degrees of variability in the images they assess.
Here too we find the AI-explainability problem. “A challenge has been that AI algorithms can be a ‘black box,’ where we can’t explain where the accuracy of the algorithm comes from,” says Supekar.
Sadly the psychology is behind the science and the times.
Psych vs. Neuroscience
According to the current DSM-5 criteria, only two core features make up an autism spectrum disorder (ASD) diagnosis: (1) persistent deficits in social communication and social interaction across multiple contexts; and (2) restricted, repetitive patterns of behavior, interests, or activities (Lai et al., 2014).
The latest A.I. study showed more data on the neuroscience:
Lending credibility to the XAI algorithm’s findings, those three brain regions have been previously implicated in autism pathology. The regions are the posterior cingulate cortex and precuneus, which form part of the default mode network (DMN), notably active during periods of wakeful rest; the dorsolateral and ventrolateral prefrontal cortex, involved in cognitive control; and the superior temporal sulcus, involved in processing the sounds of human voices. In particular, disruptions to the DMN served as strong predictors of autism symptom severity in the studied population.
Posterior cingulate cortex
Like I learned in my Psych degree, a lot of “mental health” issues have a corresponding neurological or neurobiological basis in fact. In fact, many females with autism are misdiagnosed with eating disorders, according to Spek et al. (2020, Journal of Autism and Developmental Disorders). Many women with ASD are experts at “masking” causing them great anxiety while being functionally undiagnosed.
Real Prevalence of Autism and ASD
In 2021, the CDC reported that approximately 1 in 44 children in the U.S. is diagnosed with an autism spectrum disorder (ASD), according to 2018 data. We don’t even know what the real data on this globally is. It’s thought perhaps to be around 1.85% in many Western countries. That’s a lot of people!
How do clinicians diagnose autism?
There is no blood test, brain scan or any other objective test that can diagnose autism—although researchers are actively trying to develop such tests. Clinicians rely on observations of a person’s behavior to diagnose the condition. In 2022, that’s totally shocking! The CDC 2016 prevalence in the U.S. is then is 1 in 42 for boys and 1 in 189 for girls. These rates yield a gender ratio of about five boys for every girl.
When someone says they are on the spectrum, it has a lot of meanings about their life and their soft skills potentially. A systematic review reported the prevalence of autism in Mainland China, Hong Kong, and Taiwan to be 26.6 per 10,000. That’s a lot lower than in the West. How could this be?
My hope is that A.I. will help in solving many of the issues around ASD and Autism diagnosis and giving us clearer data and knowledge as well as enabling more social justice for girls with Autism and ASD labor participation rates in general leading to more empowering lives and life experiences for all of us no matter our deviances from “normal”.
According to PwC’s report on XAI, AI has a $15.7 trillion of opportunity by 2030. However, as AI tools become more advanced, more computations are done in a “black box” that humans can hardly comprehend. While the XAI algorithm performed admirably at this early stage of development, Supekar and colleagues will need to improve its accuracy further still to raise brain fingerprinting to the level of a definitive biomarker. Even at A.I’s speed, that may take more time.
Do you think A.I. could impact our understanding of Autism in the future?
If you want to support me so I can keep writing, please don’t hesitate to give me tips, a paid subscription or some donation. With a conversion rate of less than two percent, this Newsletter exists mostly by the grace of my goodwill (passion for A.I.) & my own experience of material poverty.
Anyways I hope you enjoyed the topic, that’s all for today. Want me to cover a topic related to A.I.? Pitch me a topic in a comment.