Advancing NeuroDiagnostics with Artificial Intelligence

The combination of Oracle’s advanced technology, free Oracle cloud credits (yes…free!) and leading researchers is delivering a project to evaluate children for various cognitive disorders. Our guest blogger, Jon Russell, Innovation Director in the Oracle Startup for Higher Education organization – describes how this collaboration of technology, cloud credits and leading researchers, has led to a breakthrough in the field of neurocognitive analysis and diagnosis.

Helping identify children with cognitive disorders earlier on in life makes treatment of those conditions far more effective than with older people. However, conditions like attention deficit hyperactivity disorder (ADHD) can be hard to diagnose and difficult to differentiate from other disorders.

UT Arlington (UTA) and Yale researchers led by Dr. Fillia Makedon, Dr. Morris Bell and Dr Vassilis Athitsos are working on a project to evaluate children for various cognitive disorders by having them perform cognitively demanding physical tasks while motion capture technology automatically scores their performance.  Capturing cognition in motion may be much closer to how children function in daily life than are static cognitive tests. It is only now possible to use physical cognitive tasks because of the technological advances being made at UTA.  This work is also part of the iPerform, a NSF-funded Industry-University research center directed by Makedon, that focuses on assistive technologies to enhance human performance.

Essentially, the children are asked to perform a series of physical or computer-based tasks which can then be analyzed using software algorithms. There are 32 different physical activities.  For example, one cognitively demanding task is called the “Opposites Game”. Children are asked to do the opposite movement of what the administrator is saying. When the child is asked to touch her hips, she must touch her shoulders, and when she is told to touch her shoulders, she should touch her hips instead.  Such a task requires self-regulation and working memory, important components of executive functioning.  The child’s movements are recorded using motion capture software, with each video frame then analyzed using specialized algorithms which are continually refined using machine learning to spot patterns of behavior and assess how the child processes and acts on each set of instructions. The resulting data gives a far quicker indication of whether the child has a diagnosable condition like ADHD.

“Our mission is to innovate human performance and safety, produce assistive technologies and services, and to promote iPerform’s industry members, researchers, and students."

Dr. Fillia Makedon & Dr. Ovidiu Daescu - iPerform Directors

Oracle Startup for Higher Education gave iPerform $100k of cloud credits to accelerate research projects like this one. Dr Athitsos is using his project’s cloud credits to examine how much faster they can train their algorithms for analyzing the massive amount of data being captured as well as demonstrate how they can save money by using the elastic nature of the cloud to grow or shrink their computing capacity on demand. Oracle’s Cloud Infrastructure can also provide a highly secure, HIPAA-compliant way for the UT Arlington and Yale scientists to more easily share project data in the future. Two birds with one stone, so to speak. 

"From supporting scientific research and experimentation, to applying research results in creating a product, to enabling entrepreneurs to bring the product to market, Oracle Startup for Higher Education aims to play an active role in this journey, with the ultimate goal of helping to deliver solutions that positively impact the human experience and our world at large."

Jenny Tsai-Smith - Vice President,Oracle Startup for Higher Education

The work by the Athitsos and Makedon team will be considered a breakthrough in the field of neurocognitive analysis and diagnosis and has potential for these innovations to be developed into future applications. What they’re learning could also be of great help to any technology involving automated motion detection and analysis, such as security or other scientific applications.