Bayesian Networks-Based Interval Training Guidance System for Cancer Rehabilitation

benhurspicyΤεχνίτη Νοημοσύνη και Ρομποτική

7 Νοε 2013 (πριν από 3 χρόνια και 10 μήνες)

68 εμφανίσεις

+

Bayesian Networks
-
Based Interval Training Guidance System

for Cancer Rehabilitation

Myung
-
kyung Suh, Kyujoong Lee, Alfred Heu, Ani Nahapetian,

Majid Sarrafzadeh


University of California, Los Angeleås


+

Intro


Over 53.9% of cancer patients survive more than 5 years after surgeries .


Many of these patients have a chronic illness.


Cancer fatigue
is seen most frequently.


Results from muscle weakness, pain or sleep disruption.


Causes disruptions in physical, emotional, and social functions.




Many researchers and physicians recommend
interval training


Interval training helps


improve aerobic capacity


restore physical functions


cardiovascular systems


Interval training has been shown to decrease fatigue, and somatic complaints in
recovering cancer patients [1].

Cancer Rehabilitation


[1] Diemo FC. 1999. Effects of physical activity on the fatigue and psychological status of cancer patients during chemothera
py

+

Intro



Consists of interleaving high intensity exercises with rest periods







Other Benefits


weight loss


general fitness


the reduction of heart diseases

Interval Training

3

+

Intro


Programmed treadmills and cycles


Without them, there is almost no way to imitate a given exercise
protocol.


Without strong
motivation
, an individual can be discouraged
from following an interval training protocol.

Interval Training

4

+

iPhone

Interval Training Guidance System


Our behavioral cueing system developed for the iPhone uses
music, sensor readings, and social networking


5

Customized Input

+

iPhone

Interval Training Guidance System


Our behavioral cueing system developed for the iPhone uses
music, sensor readings, and social networking


6

Interval Training
Game

+

iPhone

Interval Training Guidance System


Our behavioral cueing system developed for the iPhone uses
music, sensor readings, and social networking


7

Music
Recommendation

+

iPhone

Interval Training Guidance System


Our behavioral cueing system developed for the iPhone uses
music, sensor readings, and social networking


8

Social
Networking

Email Sent

+

Interval Training Motivations


Reduce space and cost
restrictions compared
with traditional fitness
equipment


iPhone’s easy interface


3.5 inch multi
-
touch
display


480
-
by
-
320
-
pixel
resolution



9

Light
-
Weight Wireless Smartphone


Factors influencing mobile handheld device use and adoption

+

Interval Training Motivations

10

Light
-
Weight Wireless Smartphone


Factors influencing mobile handheld device use and adoption


Network connection


Modalities of mobility


HSDPA (High
-
Speed Downlink
Packet Access) to download
data quickly over UMTS
(Universal Mobile
Telecommunications System)


Using 3G network


When not in a 3G network
area, the iPhone uses a GSM
network for calls and an
EDGE network for data.


According to the market
research group NPD, Apple's
iPhone 3G topped the sales
charts



+

Interval Training Motivations



11

Music Motivation


Situational factors

Personal factors

Rhythm response
M
usicality


Improved mood

Arousal control

Dissociation

Reduced RPE

Greater work output

Improved skill
acquisition

Flow state

Enhanced performance

Terry, Peter C. and Karageorghis, Costas I., Psychophysical effects of music in sport and exercise: an update
on theory, research and application
,
Joint Conference of the Australian Psychological Society and the New
Zealand Psychological Society. 2006

+

Interval Training Motivations

Subscale

Ranking

Affiliation

2

Appearance

12

Challenge

4

Competition

1

Enjoyment

3

Health pressures

14

Ill
-
health avoidance

13

Nimbleness

8

Positive health

7

Revitalization

5

Social recognition

9

Strength and endurance

6

Stress management

10

Weight management

11

12

Competitive Group Exercise




Exercising together



Maintain affiliation with friends and promote more exercise



Related to social network


Ranking of exercise motivation

Kilpatrick, M., College Students' Motivation for Physical Activity: Differentiating Men's and Women's Motives for Sport Parti
cip
ation and
Exercise. Journal of American college health, 2005


+

Related Works


Music Recommendation Systems


Pandora


MusicSurfer


iPod Exercise Applications


Nike + iPod Sport Kit


Nike+ Shoes


Social Network Systems


FaceBook


MySpace

13

+

System Design


Using the user input, the system comes up with a customized
interval training protocol.


By comparing the schedule with the exercise data collected
from the 3
-
axis accelerometer, the accuracy or score of the
exercise is calculated.

14

Game

Scheduled interval training (a)

and the accelerometer data for the
exercise (b)


+

System Design


Content
-
based filtering


Selects songs based on the correlation
between the content of the items and the
user’s preferences.

15

Music Recommendation

+


Collaborative filtering


Chooses songs based on the correlation
among people with similar preferences.


Uses Bayesian networks in our system.


16

System Design

Music Recommendation

+


In collaborative filtering



The system classifies users based on age, gender, and residential location, etc.



Songs are selected by using Bayesian networks.















17

System Design

Music Recommendation

Sources of variation in music preference

LeBlanc,

A
.
,

Tempo

Preferences

of

Different

Age

Music

Listeners
.

Journal

of

research

in

music

education,

1988



+


How Bayesian Networks Work?


Based

on

the

assumptions,

a

Bayesian

network

model

is

obtained

and

is

used

to

calculate

the

probability

that

the

given

song

is

recommended

by

people

sharing

similarities

with

the

user
.


When

the

value

is

above

the

threshold,

the

song

is

recommended

to

the

user
.


18

System Design

Music Recommendation

+


Context
-
aware filtering


Provide a user with relevant information and services based on
one’s current context such as exercise intensity.


19

System Design

Music Recommendation

+

System Design


E
-
mails containing the accuracy of the exercise sessions, exercise
session time, and the amount of calories burned, etc. are sent to
other members in the user’s social networking group

20

Social Network

Contact List
System
User
Group
Friends who compete
with the user
User Databse
+

Experimental Results

Individual 1

Individual 2

Individual 3

Individual 4

Individual 5

Individual 6

Individual 7

Individual 8

Gender

Female

Male

Male

Female

Female

Male

Male

Female

Age

25

24

27

28

25

29

27

25

Weight

(kg)

51

61.4

73

49.5

50.5

62

70

51

Height

(cm)

158

170

175

163

164

172

175

158

Residential
District

Los Angeles,
CA

Los Angeles,
CA

Los Angeles,
CA

Los Angeles,
CA

Los Angeles,
CA

Los Angeles,
CA

Los Angeles,
CA

Los Angeles,
CA

21

0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
90.00%
100.00%
The accuracy of the exercise

The effectiveness

of the guidance system

Without any command
with commands
+

Experimental Results

22


Each song in the web database was annotated more than 8 times
by 8 users.


Compared with the method which recommends music preferred
by people who share the same conditions, Bayesian networks
-
based recommendation method is better for selecting suitable
exercise music.


The number of refused songs among 10 recommendations for a 30 years old, 180cm, and 80kg
individual living in Los Angeles, California.

+

Conclusion

23

+

Questions??

24