amsys 16(10): e3

Research Article

Personalized neuroscience: User modeling of cognitive function and brain activity in the cloud

Download591 downloads
  • @ARTICLE{10.4108/eai.28-9-2015.2261443,
        author={Teresa Nick and Laura Berman and Arye Barnehama},
        title={Personalized neuroscience: User modeling of  cognitive function and brain activity in the cloud},
        journal={EAI Endorsed Transactions on Ambient Systems},
        volume={3},
        number={10},
        publisher={EAI},
        journal_a={AMSYS},
        year={2015},
        month={12},
        keywords={cloud, wearable, electroencephalography, eeg, cognitive game},
        doi={10.4108/eai.28-9-2015.2261443}
    }
    
  • Teresa Nick
    Laura Berman
    Arye Barnehama
    Year: 2015
    Personalized neuroscience: User modeling of cognitive function and brain activity in the cloud
    AMSYS
    EAI
    DOI: 10.4108/eai.28-9-2015.2261443
Teresa Nick1,*, Laura Berman1, Arye Barnehama1
  • 1: DAQRI, LLC
*Contact email: teresa.nick@daqri.com

Abstract

Reliable detection and prediction of neural activity and behavior requires a user model of brain activity that dynamically adapts based on known time-dependent physiological processes, as well as unknown traits of the user. We have applied wireless electroencephalography (EEG) sensors, edge devices with feedback capability, and cloud-assisted data acquisition to real-time and longitudinal brain monitoring and alerting. Toward a user model of brain function, we collected neural and behavioral data from humans in the field. The data replicate previous findings that were obtained under tight laboratory control, suggesting that the methods that we describe will be useful for user modeling of human brain activity under more natural conditions. Specifically, we report that frontal cortex oscillations reorganized with age. Focusing on time-varying aspects of behavior, we then found that performance on memory-intensive cognitive tasks declined during the day. Next, we examined interactions between neural activity and behavioral performance. We report that neural activity and performance co-varied and that this co-variation depended on the cognitive task in ways that were, again, consistent with previous laboratory studies. Lastly, we report the foundations of an adaptive model based on this system that will enable dynamic personalization tailored to each user.