Applying machine learning to learning disabilities and autism

Table of Contents

  1. Introduction
  2. Assisting those with learning disabilities
  3. Predicting a child’s special needs as a newborn
  4. Helping those with autism recognize facial expressions
  5. Providing assistance when needed
  6. Conclusion


  1. Introduction

Machine learning can be a tremendous help in the education space. The technology can offer tailored lessons to students so that they can be adapted to their personal learning styles. The concept is that there is an automated feedback loop such that the lesson is given to the student and then sensing data is collected such as a visual feed from a laptop camera to determine attentiveness. The lesson is then adapted accordingly until the student’s attention is brought back into focus. This approach is particularly invaluable for children with learning disabilities who may need extra instruction and attention in order to get them through the course material. Machine learning is also. beginning to be used more frequently for those with autism as well. Those with learning disabilities and autism spectrum disorder (ASD) can benefit greatly from some of the new approaches that we are beginning to see with machine learning that I’ll get into now.

  1. Assisting those with learning disabilities

Private organizations and universities alike are developing machine learning based tools that can make content more available to people with learning disabilities. IBM is creating a tool called the Content Clarifier that is a branch off of their bigger computing project, Watson. The Content Clarifier is able to take parts of speeches and break them into more simple and understandable terms. An example that was presented in MIT’s Technology Review was the phrase “raining cats and dogs,” where it doesn’t mean exactly what it sounds like. IBM is partnering with the University of Massachusetts medical school to test the technology in a real classroom setting. Another superstar cover in MIT’s article is Ineke Schuurman who leads a mobile app project that is designed to simplify text for people with learning disabilities. By creating tools automated with machine learning, text, images, and videos, can be more broadly enjoyed by all.

  1. Predicting a child’s special needs as a newborn

Machine learning is fast-tracking the timeline of detecting autism spectrum disorder.  Early detection of autism spectrum disorder is extremely important since it can allow the parents to get the child the increased. According to Dr. Dawson who leads Duke University’s Center for Autism and Brain Development, early intervention of ASD could save the child and their family over a million dollars in medical costs by intervening early. In order to detect the disorder as early as possible, researchers at Duke University are turning to machine learning approaches.

The researchers developed a mobile app that could play a movie and use the camera to track the child’s facial movements for example. The scale of the research was able to be expanded significantly due to the automation and machine learning framework. The study involved over 1,700 participants that had autism spectrum disorder. The researchers hope to make autism screening much more widely accessible and cheaper due to the ability to scale up with cloud computing services.

Other companies are working on solving this problem too. A California-based startup company called Cognoa is transforming the autism services industry. They are creating a software platform that allows parents and teacher to identify autism earlier. Similarly, a company called Brain Power is working on creating a way for those with autism to learn to better control their own emotions through a series of wearable devices.

  1. Helping those with autism recognize facial expressions

A team at Stanford is working on modifying Google Glass in order to assist in identifying facial expressions. The modified Google Glass called Superpower glass uses artificial intelligence to analyze the facial expressions of whoever the user is looking at and feed them back an emoticon indicating their emotion. One amazing direction that the developers brought the invention is not only towards identifying facial expressions but an in-app review feature. The device records all the interactions that took place throughout the day and can be reviewed later with parents or teachers in order to understand what happened or how a situation should have proceeded differently.

  1. Providing assistance when needed

Lastly, machine learning is helping autistic children when they need help in the classroom. One research team provided autistic children with a robot nicknamed KiWi and this robot served as a coach to the children while they tackled mathematics problems. The robot recorded the children’s attention throughout lessons and after a month training period where able to train an algorithm to recognize when they were paying attention. The model that was trained reached 90% accuracy in determining when the child was attentive during a lesson. KiWi served in some cases to bring the children back into focus by speaking when needed.  More development in this area is very exciting since we know that everyone learns differently.

  1. Conclusion

Learning disabilities and autism are difficult disorders that can be tough on parents and teachers. There are strides being made in machine learning that may make it easier to predict what type of special needs their child will need. Once that is determined, machine learning tools can be used right through the child’s schooling to provide them personalized education and recognize when they need help the most.

Show More

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button