USING GOOGLE’S ANDROID TO CREATE A HEALTH GAME By: Elie ElChartouni, Diami Goudiaby, Brice Sorrells 7/29/2009
Feb 13, 2016
USING GOOGLE’S ANDROID TO CREATE A HEALTH GAME
By: Elie ElChartouni, Diami Goudiaby, Brice Sorrells7/29/2009
Motivation Childhood Obesity
According to the American Academy of Child & Adolescent Psychiatry:
Between 16 and 33 percent of children and adolescents are obese
Annual cost to society for obesity is nearly $100 billion
7/29/2009
An obese child between 10 and 13 has an 80% chance of being an obese adult
Background Info
7/29/2009
Current state of Health games
This represents a first step in developing a health game for DexterNet
DexterNet Wearable body sensor network
Multiple wireless sensor nodes and one wearable base station
Action recognition
Automatic communication with doctors
7/29/2009
G1 Features• Android-based Mobile Phone• Built-in receiver and digital compass• Built-in Accelerometer & GPS• 3.2’’ Touchscreen
7/29/2009
What is ANDROID?
A complete set of software for mobile devices(an operating system, middleware and key mobile application).Is the first free, open source, and fully customizable mobile platform.
7/29/2009
Our Game Multi-purpose:
Played for fun by children on their own Will be tested at children’s obesity clinics to
promote exercise
Uses single accelerometer on G1 Mobile Phone
Targeting children (6 - 10 yrs old)
7/29/2009
Original Goal A “Simon Says” game
Use audio to prompt player to do an action Requires reorientation of the sensor’s x,y,z-
axis Requires a classification algorithm to
determine the actions done
7/29/2009
Data Collection Actions to Recognize:
Jump, Squat, Shake, Slide, Spin left, Spin right..
Collected accelerometer data from 10 different individuals doing each action four times
7/29/2009
Data Collection Two different people jumping:
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39-505
1015202530
Z AxisY AxisX Axis
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39-10-505
101520253035
Z AxisY Axis
7/29/2009
Huge variety in how the action looks
Classification Allen's algorithm
1. vector representation of data2. Filtration and compression of vectors 3. sample action is compared to a sets of
training data4. best match is used
KNN algorithm
7/29/2009
New Game “GO CRAZY” Altered game to measure general
activity level As a fun children’s game:
shake & “go crazy” level of activity
As a medical tool in an obesity clinic: A way of measuring the level of effort put
into doing a specified activity
7/29/2009
Future Development Best classification algorithm
More data for the “Go Crazy” game
Testing on children needed
7/29/2009
Questions?
7/29/2009