Embedded Files
Based on Local Field Potentials
Project Overview
Project Overview
Our team developed a model for classifying REM sleep through analysis of local field potential data. We used a convolutional neural network to identify the awake, non-REM, and REM states based on scalograms of the data. Our final model was trained on 583 scalograms for each state, and has an accuracy of 97.33%.
The Four Commonly Classified Stages of Sleep
The Four Commonly Classified Stages of Sleep
Generally, sleep stages are classified by frequency of brain wave activity.
Our model is trained with Local Field Potential (LFP) data taken of rat subjects over time.
Applications And Motivations
Applications And Motivations
Classification of sleep stages is beneficial in numerous sleep-related, medical applications. For example, classification of a study's sleep could result in an ability to diagnose and treat sleep disorders, and improve health of subjects. Classification of these stage can also aid scientific studies relating sleep to different characteristics such as gender and age. Additionally, it can help with discovering effects of sleep, including adolescent development and cognitive function. This classification technique could also be personalized to track one person's sleep.
Questions?
Questions?
Contact acheslek@umich.edu, vvizza@umich.edu, or yixa@umich.edu for more information on the project
Page updated
Report abuse