Alcohol abuse is the third leading lifestyle-related cause of death for individuals in the United States, causing 88,000 deaths each year in the United States from 2006-2010. Existing smartphone applications allow users to manually record their alcohol consumption or take cognitive tests to estimate intoxication levels; however, no smartphone application passively determines one's level of intoxication. After gathering smartphone sensor data from 34 "intoxicated" subjects, we generated time and frequency domain features such as sway area (gyroscope) and cadence (accelerometer), which were then classified using a supervised machine learning framework. Other novel contributions explored include feature normalization to account for differences in walking styles and automatic outlier elimination to reduce the effect of accidental falls by identifying and removing the top and bottom of a chosen percentage of the data. Various machine learning classifier types such as Random Forest and Bayes Net were compared, and J48 classifier was the most accurate, classifying user gait patterns into BAC ranges of [0.00-0.08), [0.08-0.15), [0.15-0.25), [0.25+) with an accuracy of 89.45%. This best performing classifier was used to build an intelligent smartphone app that will detect the user's intoxication level in real time.
Identifer | oai:union.ndltd.org:wpi.edu/oai:digitalcommons.wpi.edu:etd-theses-1387 |
Date | 27 April 2016 |
Creators | Aiello, Christina Jane |
Contributors | Lane T. Harrison, Reader, Emmanuel O. Agu, Advisor, |
Publisher | Digital WPI |
Source Sets | Worcester Polytechnic Institute |
Detected Language | English |
Type | text |
Format | application/pdf |
Source | Masters Theses (All Theses, All Years) |
Page generated in 0.0018 seconds