Show simple item record

dc.contributor.advisorPassino, Kevin
dc.contributor.advisorNandi, Arnab
dc.creatorKillian, Jackson A.
dc.date.accessioned2018-04-30T13:05:26Z
dc.date.available2018-04-30T13:05:26Z
dc.date.issued2018-05
dc.identifier.urihttp://hdl.handle.net/1811/84842
dc.descriptionArts and Sciences Undergraduate Research Scholarshipen_US
dc.descriptionDenman: First Place Award in Math, Computation, and Analyticsen_US
dc.description.abstractExcessive alcohol consumption is an avoidable health risk, yet it causes a significant percentage of yearly deaths and injuries on college campuses. Recent work showed that weekly mobile-based interventions can effectively reduce alcohol consumption in students. However, few studies investigate delivering mobile interventions in real-time during drinking events where interventions could reduce risks like drunk driving, alcohol poisoning, and violence. Such studies require measuring real-time intoxication levels outside of a lab setting at scale. Some technologies exist for this purpose but are impractical or expensive. To address these shortcomings, we built an intelligent system capable of passively tracking smartphone accelerometer data to identify heavy drinking events in real time. We collected smartphone accelerometer readings and transdermal alcohol content (TAC) readings from 20 subjects participating in an alcohol consumption field study. The TAC readings served as the ground-truth when training the system to make classifications. The TAC sensors and smartphone accelerometers both provided noisy readings which were cleaned with the MATLAB signal processing toolbox. We then segmented the data into 10 second windows and extracted features known to change when humans lose control of their center-of-mass (i.e. become intoxicated). Additionally, we experimented with some feature extraction methods from sound recognition tasks and show that they provide a significant improvement in this task (up to 8% absolute accuracy gain in our case.) Finally, we built and trained several classifiers to call each window as a "sober walk" or "intoxicated walk", the best of which achieved a test accuracy of 75.04%. This result has promising implications for making classifications on noisy accelerometer data in this space and also offers multiple avenues for improvement. We plan to use our classifiers to build a mobile sobriety tracking application that ultimately will serve as a free, reliable, and widely adoptable application that tracks intoxication in real-time, enabling development of effective real-time mobile-based interventions which can later be delivered via the application to reduce unnecessary alcohol-related injury and death. The results and application will also benefit future studies as new sensor-bearing technologies become widely adopted.en_US
dc.description.sponsorshipArts and Sciences Undergraduate Research Scholarshipen_US
dc.language.isoen_USen_US
dc.publisherThe Ohio State Universityen_US
dc.relation.ispartofseriesThe Ohio State University. Department of Computer Science and Engineering Honors Theses; 2018en_US
dc.subjectmachine learningen_US
dc.subjectmobile AIen_US
dc.subjectalcohol consumptionen_US
dc.subjectsobriety sensoren_US
dc.titleSmartphone-Based Intelligent System: Training AI to Track Sobriety Using Smartphone Motion Sensorsen_US
dc.typeThesisen_US
dc.description.embargoNo embargoen_US
dc.description.academicmajorAcademic Major: Computer and Information Scienceen_US
dc.description.academicmajorAcademic Major: Physicsen_US


Files in this item

Thumbnail

Items in Knowledge Bank are protected by copyright, with all rights reserved, unless otherwise indicated.

This item appears in the following Collection(s)

Show simple item record