Emerging Role of AI in ASD From a High School StudentWritten

Written by Anand Ramabadran

Image courtesy of Pexels

𝐀𝐛𝐨𝐮𝐭 𝐀𝐮𝐭𝐢𝐬𝐦 𝐒𝐩𝐞𝐜𝐭𝐫𝐮𝐦 𝐃𝐢𝐬𝐨𝐫𝐝𝐞𝐫

Autism is a neurological disorder.

It is a spectrum disorder, which means it can affect a person in many different ways, hence the term Autism Spectrum Disorder (ASD). Signs of ASD begin during early childhood and typically last throughout a person’s life. People with ASD can have problems with social, emotional, cognitive and communication skills.

According to the CDC data cited by Autism Speaks, ASD is on the rise in America, with 1 in 54 children identified with ASD.

There is currently no cure for ASD. The wide range of outcomes makes treatment especially difficult. However, research shows that early intervention treatment services can significantly improve a child’s development.

Early intervention services help children from birth to 3 years old learn important skills. Services can include therapy to help the child talk, walk, and interact with others.

Unfortunately, diagnosing ASD is difficult since there is no medical test, like a blood test, to diagnose the disorder. Doctors observe the child’s behavior and development over a period of time to make a diagnosis.

The average age of those diagnosed with ASD is around 4 years, 4 months. So, the opportunity to make a significant impact on the child’s development through early intervention during the first 3 years is lost or under-utilized due to the long time to obtain a proper diagnosis.

The causes of ASD remain unknown.

However, there may be many different factors that make a child more likely to have an ASD, including environmental, biologic and genetic factors.

ASD takes a toll on the afflicted children, their parents and the society at large.

Afflicted children face the challenge of adapting to uncomfortable environments and routines. As they age, integrating with society can pose challenges during teen years, as well as learning life skills needed to become independent.

Parents deal with the anxiety and emotional toll that comes with an ASD diagnosis for their children and being able to provide them the care needed. The economic burden associated with ASD is substantial and estimated to be tens of billions of dollars annually.

As medical and education costs rise, and as ASD children become parents themselves, the number of children with ASD and the economic costs can rise exponentially.

𝐂𝐨𝐦𝐩𝐥𝐞𝐱 𝐏𝐫𝐨𝐛𝐥𝐞𝐦𝐬 𝐑𝐞𝐪𝐮𝐢𝐫𝐞 𝐒𝐨𝐩𝐡𝐢𝐬𝐭𝐢𝐜𝐚𝐭𝐞𝐝 𝐓𝐨𝐨𝐥𝐬: 𝐀𝐈 𝐇𝐞𝐥𝐩𝐬 𝐓𝐚𝐜𝐤𝐥𝐞 𝐭𝐡𝐞 𝐀𝐒𝐃 𝐂𝐡𝐚𝐥𝐥𝐞𝐧𝐠𝐞

Every facet of ASD, ranging from causation, screening, detection/diagnosis and treatment is a complex problem with multiple variables.

Given the complexity, the need for analytical tools that can handle large amounts of data and help observe patterns, develop insights and narrow areas to focus on for more detailed research is acute. AI has already helped make significant inroads in various facets of our response to the challenges of ASD.

𝐀𝐈 𝐏𝐨𝐰𝐞𝐫𝐞𝐝 𝐃𝐢𝐚𝐠𝐧𝐨𝐬𝐢𝐬

In her article titled “Can AI Help With Autism Diagnosis?”, Jun Wu profiles Cognoa, a pediatric behavioral health company. In June 2021, the U.S. FDA authorized marketing of Cognoa ASD Diagnosis Aid to help diagnose ASD. The FDA made the following comments about the Cognoa product in their press release to announce the authorization:

“Cognoa integrates AI and machine learning together with a deep understanding of child behavioral development to help pediatricians diagnose autism in a timely fashion so that more kids can get the right treatments years sooner, when they can have the greatest impact.

The Cognoa ASD Diagnosis Aid is a machine learning-based software intended to help health care providers diagnose ASD in children 18 months through 5 years of age who exhibit potential symptoms of the disorder.

The Cognoa ASD Diagnosis Aid is a software as a medical device that uses a machine learning algorithm to receive input from parents or caregivers, video analysts and health care providers to assist physicians evaluate a patient at risk of ASD.

The device consists of three main components: a mobile app for caregivers and parents to answer questions about behavior problems and to upload videos of their child; a video analysis portal that allows manufacturer-trained and certified specialists to view and analyze uploaded videos of patients; and a health care provider portal that is intended for a health care provider to enter answers to pre-loaded questions about behavior problems, track the information provided by parents or caregivers and review a report of the results.

After processing the information provided by parents, caregivers and healthcare providers, the ASD Diagnosis Aid reports a positive or negative diagnosis if there is sufficient information for its algorithm to make a diagnosis. If there is insufficient information to render a “Positive for ASD” or “Negative for ASD” result to help determine a diagnosis, the ASD Diagnosis Aid will report that no result can be generated.”

𝐀𝐈 𝐏𝐨𝐰𝐞𝐫𝐞𝐝 𝐋𝐢𝐟𝐞 𝐒𝐤𝐢𝐥𝐥 𝐓𝐨𝐨𝐥𝐬

In her article titled “Artificial Intelligence and Autism”, Jun Wu takes the view that AI will transform the lives of people with ASD and profiles the products of two companies, Brain Power and LuxAI. Jun makes the following comments:

“Brain Power is a company that designs AI solutions that respond to emotions. Their flagship product, Empower Me teaches social and cognitive skills to people with autism through emotional recognition. Empower Me runs on any smart glasses. When a child or an adult wears the glasses, it sees and hears special feedback geared toward the situation. The digital coach helps the wearer to interpret facial expressions of emotions, when to look at people and gives feedback of the wearer's own state of stress or anxiety.

Empower Me allows someone with ASD to function better independently in the world. It is one of the first products to allow a person with ASD to interact effectively in a traditional workplace setting.

LuxAI is a company that aims to address the shortage of special education services worldwide. Their flagship product, QTrobot, is a humanoid social robot designed to help children with ASD learn social skills. Therapists can easily program the robot for a special set of skills and instructions to tailor learning for the individual patient. Children with ASD have oversensitivity issues when interacting with other human beings. Since they are unable to interpret the nonverbal communication that is often central to interactions with another human being, they are often overwhelmed. Learning from a robot instructor who does not display all the skills that children with ASD can not interpret can be a very soothing experience. Children with ASD can focus on learning the materials presented by the robot instructor without distractions.”

𝐀𝐈 𝐏𝐨𝐰𝐞𝐫𝐞𝐝 𝐀𝐒𝐃 𝐂𝐚𝐮𝐬𝐚𝐭𝐢𝐨𝐧 𝐑𝐞𝐬𝐞𝐚𝐫𝐜𝐡

In her article titled “Deep Learning Detects Mutations Behind Autism Spectrum Disorder”, Jessica Kent describes a breakthrough 2019 Princeton study. Jessica makes the following comments:

“Previous studies aimed at identifying genetic causes of ASD focused on mutations in the 20,000 known genes and the DNA surrounding the genes, which only makes up a little over one percent of the 3.2 billion chemical pairs in the human genome. In 2019, a Princeton team developed a deep learning algorithm to examine the other 99 percent of genomic data.

Prior to the study, it was impossible to look across the entire genome for bits of DNA that regulate genes and to predict how likely mutations in regulatory DNA caused complex diseases. This study is the first to prove that mutations outside the genes themselves can result in conditions like autism.

Researchers analyzed the genomes of 1,790 families in which one member had autism, but others didn’t. The deep learning model sorted through 120,000 mutations and found those that affect the behavior of genes in people with autism with a particular eye on mutations predicted to have a high functional impact. Researchers then examined which genes were affected by these mutations and found that they were genes strongly associated with brain functions.

While the results did not reveal the exact causes of autism, the team did find thousands of possible contributors for other researchers to study. Among the sample used in the study, less than 30 percent of people affected by ASD had a previously identified genetic cause, but researchers expect that the newly identified mutations will increase that percentage.

This information will help researchers understand all the factors involved in autism. The findings may also help doctors better diagnose the condition and recognize how autism differs from person to person. This new method can speed up the process significantly, allowing scientists to recognize the most pressing cases. The deep learning system also calculates a disease impact score, which assesses how likely a given mutation will affect disease.

Researchers have used deep learning to improve many areas of healthcare. Recently, a team at Google developed a deep learning tool that can detect lung cancer as well as or better than human radiologists, showing that the technology could potentially be used in clinical settings to augment decision making.

Using this new deep learning method, the team at Princeton University are planning to analyze the genetic causes of different types of cancer, heart disease, and other complex conditions.”

𝐂𝐨𝐧𝐜𝐥𝐮𝐬𝐢𝐨𝐧

AI is emerging as a key tool in the fight against ASD.

AI powered devices are being deployed in the diagnosis and early detection of ASD. Persons afflicted with ASD are able to better adapt to society using AI enabled life skills tools.

Lastly, but importantly, AI is enabling researchers to take on the daunting task of understanding genetic links to ASD causation. Advances in the use of AI to understand genetic links are enabling research of ASD as well as other areas of healthcare such as cancer.

𝐒𝐨𝐮𝐫𝐜𝐞𝐬 𝐮𝐬𝐞𝐝

""CDC estimate on autism prevalence increases by nearly 10 percent, to 1 in 54 children in the U.S."" Autism Speaks. Last modified March 26, 2020. https://www.autismspeaks.org/press-release/cdc-estimate-autism-prevalence-increases-nearly-10-percent-1-54-children-us.

""FDA Authorizes Marketing of Diagnostic Aid for Autism Spectrum Disorder."" U.S. Food and Drug Administration. Last modified June 2, 2021. https://www.fda.gov/news-events/press-announcements/fda-authorizes-marketing-diagnostic-aid-autism-spectrum-disorder.

Kent, Jessica. ""Deep Learning Detects Mutations Behind Autism Spectrum Disorder."" Health IT Analytics. Last modified May 28, 2019. https://healthitanalytics.com/news/deep-learning-detects-mutations-behind-autism-spectrum-disorder.

Wu, Jun. ""Artificial Intelligence and Autism."" Towards Data Science. Last modified May 13, 2019. https://towardsdatascience.com/artificial-intelligence-and-autism-743e67ce0ee4.

Wu, Jun. ""Can AI Help With Autism Diagnosis?"" Forbes. Last modified Mar 19, 2020. https://www.forbes.com/sites/cognitiveworld/2020/03/19/can-ai-help-with-autism-diagnosis/?sh=2b19a13564a3."