Computational Approach for the Real-Time Diagnosis of Attention Level and Focus State

Zack Dvey-Aharon, Or Cohen, Nathan Intrator

Abstract: The study of cognitive attention has become a major area of interest in the fields of education, neuroscience and psychiatry during the past several decades. Although the motivation behind these studies originated from the need to better understand mental disorders, such as ADHD, a growing desire to study attention and identify tools to assess the focus state has emerged. As the interest in this area has grown rapidly, computational academic tools are still lacking, the state-of-the-art tools that are currently available focus solely on either general retroactive detection of the focus state or discrimination between subjects diagnosed with various mental illnesses (mostly ADHD) and healthy subjects. To date, there is no reliable tool available to make a real-time diagnosis of the focus state. Due to the volatile nature of attention and unlike the common usage of general discrimination tools, real-time diagnostic tools enable actionable and practical event-driven responses. These event-driven responses include neurofeedback to tackle specific brain diseases and live notifications for helping subjects regain focus while carrying out critical daily tasks that require high levels of attention, such as flying a plane, supervising or even driving, to name a few. In this paper, we present the WBV (Weighted-Band and Volatility) procedure, a novel methodology for real-time diagnosis of the attention level and focus state. The methodology showed excellent results in real-time focus state prediction, revealing the attention level of the subjects using less than 30 seconds of data and limited EEG input from only two frontal electrodes.

Keywords: ADHD, ADD, Focus, Attention, Learning, Game, Machine Learning, Neuroscience