AI for Disability by Dr. Sonia Goel

AI for Disability



Worldwide more than 1 billion people have been identified as disabled in some way according to the recent World Health Organisation (WHO) Statistics. Just within the Indian Subcontinent alone, the census data indicates 2 %  of the population has some kind of disability. Of course, besides permanent disability, there is a whole spectrum of ranges of special needs to consider.

For example, people might become temporarily disabled perhaps with a broken limb or situationally disabled like driving your car and not able to use your visual attention.

Here I want to share some overview about some key topics areas inaccessibility that we can begin to think about as challenge areas to address with AI technologies.

From a technological point of view there were five different categories of disability:

  • Visual Disabilities
  • Hearing Disability
  • Speech and Communication Disability
  • Motor and Mobility Impairments
  • Mental Health
The above infographics are a somewhat simplified taxonomy of disability. So, many people might have conditions across several of these different categories such as certain kinds of illness and medical conditions might result in different combinations of disabilities, for example, people with ALS like Late physicist Stephen Hawking have this condition that causes motor disability but then because of the impact of that motor disability that also impact their ability to produce speech. So, you would have a combination of motor and speech needs. And I think these are really the rich areas for thinking about the application of AI.

AI for Visual Impairment 

The development of technologies for people who are blind seems like a grand challenge for computer vision research. I think what is more common to computer vision as a discipline than a natural application that reproducing in software, the perceptual ability is afforded by the human eye. There’s has been a lot of research in this area.

But sometimes it's become a bit of afterthought like OK!! I get this cool new computer vision software I wonder if a blind person can use it or not. So, a few years ago a survey was done by Jeff Bigham, a Professor at Carnegie Mellon University and his fellow students where they surveyed blind peoples through an app called ‘Vizwiz’. The participants have to take a picture and ask questions and in return, they could get their answers. Questions are of wide variety like how the sky looks like and how to cook a recipe. They packed this up into a data set of about 50000 answers and questions and the hope was that their everyday life may inspire computer vision researchers to work on this problem and develop a better app for visual impairment.

There are also other apps that are used by visually impaired people like Alexa, Google Assistants which are already helping them to complete day to day tasks without the technological barriers. In the field of development of AI for Vision impaired, I would like to see some more work.

AI for hearing disability 

Educational institute and big corporate offices often use software like Microsoft translators as an additional communication tool which can help hearing-impaired students and workers on their journey towards their success. Microsoft translators use AI to provide another really strong bridge for the gap that has been there for so long. Various Presentation software used in Schools like National Technical Institutes for Deaf reads the content and everything that you have within the Powerpoint system.  The Cognitive Services Custom Speech Recognition (CSCSR) is critical for capturing vocabulary words that wouldn’t be necessarily conventional in everyday life. Through these technologies, the user can pick any language that they choose to receive the information. For example, if the presenter has chosen English, then the user can choose whatever language they learn in best. Users can also use this kind of AI-driven apps to initiate a conversation with others.

AI for Speech and Communication Disability  

An estimated about a million people in India, a bit less than one out of 50 people would be considered as severely communication impaired. So, there are many different technologies like ACC (Augmentative and Alternative Communication) which can be used to improve their daily life. ‘Augmentative’ means to supplement or enhance the communication methods that a person already has, and ‘alternative’ means instead of the communication methods a person already has. AAC is becoming a great solution that helps people to communicate effectively and more efficiently. The AAC users are of wide range. People using AAC are those who have physical disabilities (i.e., cerebral palsy and ALS), linguistic disabilities like aphasia, autism spectrum disorders and many more.  These technologies may also be applicable to people outside the disability spectrum e.g. an adult who might be learning a new language.

AI for Motor and Mobility Impairments

Users with motor impairments have a wide range of conditions including spinal cord injuries, multiple sclerosis, ALS, Cerebral Palsy, essential tremor and Parkinson’s disease. These users may find it difficult to use a touch screen particularly if your app relies on small touch targets or complex gestures. In recent years big tech companies like Google has developed the accessibility services like Switch Access and Voice Access in order to provide these users with alternate means of operating these devices.

AI for Mental Health

This last category is something which I struggle to come up with a good name for it, so I articulated it as AI assistance for people with mental distress and this topic is just beginning to come to many people’s awareness. Now, these neural conditions such as post-traumatic stress disorder, anxiety, depression etc are acquiring a new perspective for treatment with AI. Even though still in development, social robots such as Sophia and Han created by Hanson Robotics are gaining popularity for becoming a personalized psychiatrist for those with mental disorders. This concept is getting more popularised because machines have the ability to give much more unbiased opinions and logical reasoning. In many areas of the world, using AI in this field is still a deceitful topic. 

Comments

Popular Posts