context aware computing for personalised healthcare
TRANSCRIPT
Saurav Gupta
CONTEXT AWARE MOBILE AGENT FOR REDUCING
STRESS AND OBESITY BY MOTIVATING PHYSICAL
ACTIVITY
Problem
March 2016 2
o People, at large, are suffering from stress and obesity. With various
studies showing strong correlation between stress and obesity.
o After studying the physical activity patterns amongst people in India, it
was found that majority of them are inactive.
o Also, people, while being mobile, their operating environment/context
changes frequently, which limits the duration and degree of physical
activity.
Objectives
March 2016 3
o To enable adoption of active and healthy lifestyle amongst
the people, while being ‘on-the-fly’
o To bring about a behavioral change amongst the people to
shift from curative care to preventive care
To achieve these objectives, a ‘Context-Aware Mobile Agent’
was seen as the solution
Literature Study
March 2016 4
Distributed
implicit HCI
Context-
Aware
I
Autonomous Intelligent
Virtual Environments
Physical Environments
HCI
(Cooperate)
HCI
(Compete)
Human Environments
ICT
UbiComp
System ICT
CCI
HCICPICPI(Sense,
Adapt)
The UbiCom System Model (source: Ubiquitous Computing, Wiley)
EVOLUTION:
Human to Human
Human to Computer
Computer to Human
Computer to Physical
Environment
Literature Study
March 2016 5
‘Context is any information that can be used to characterize the situation of
an entity. An entity is a person, place, or object that is considered relevant to
the interaction between a user and an application, including the user and
applications themselves.’
‘A system is context-aware if it uses context to provide relevant in-formation
and/or services to the user, where relevancy depends on the user’ s task’
(Source: Dey and Abowd)
Literature Study
March 2016 6
Literature Study
March 2016 7
Context aware System capabilities:
• Contextual Sensing: It is the retrieving the contextual data in a user-friendly
format.
• Contextual Adaptation: It is the ability of the context aware system to adapt to
the changing environment.
• Contextual Resource Discovery: The context aware system identifies the
additional resources that it would require to present an improved adaptation by
the user.
• Contextual Augmentation: In this, the context aware system is able to co-relate
the contextual information with existing factual digital information.
Abstract Layered Architecture
March 2016 8
Abstract Layered Architecture
March 2016 9
Challenges in Context Aware Systems:
• Data acquired may not provide a holistic view of the operating environment
of the user
• The quality of data depends on the quality of sensors utilized in acquiring
the contextual information.
• The data acquired may not provide the actual information desired by the
system. Hence, appropriate interpretation mechanisms need to be
deployed to understand the data acquired.
• The measurement units involved while acquiring the data maybe multiple.
For example, the data maybe in centimeters, meters, kilometers or miles.
Step 1a
Identify the ‘subjects’ who were obese
To quantify obesity, Body Mass Index was used. It is defined as:
Out of the 97 people surveyed, 33 subjects having BMI of 25 and above were identified
Methodology
March 2016 10
Methodology
March 2016 11
2
31
51
13
Obese Overweight Normal Underweight
BMI Index
Series1
Step 1b
Identify the ‘subjects’ who were stressed
To identify subjects suffering from Stress, a questionnaire was designed. It comprised of 05 Indicators:
Physical Indicator (comprised of 21 questions)
Sleep Indicator (comprised of 05 questions)
Behavioral Indicators (comprised of 17 questions)
Emotional Indicators (comprised of 21 questions)
Personal Habits (comprised of 09 questions)
Each question was based on a 5-point Likert scale
A tablet-based mobile application was designed to capture the user inputs.
Methodology
March 2016 12
Methodology
March 2016 13
Step 1b (cont’d)
Methodology
March 2016 14
Step 1b (cont’d)
0%6%
26%
37%
31%
PHYSICAL STRESS
Very Low
Medium
High
Very High
Danger
8%
38%
20%
14%
20%
SLEEP STRESS
Very Low
Medium
High
Very High
Danger
2%
48%
40%
7%3%
BEHAVIOURAL STRESS
Very Low
Medium
High
Very High
Danger
5%
13%
15%
67%
EMOTIONAL STRESS
Very Low
Medium
High
Very High
Danger
2%6%
17%
34%
41%
PERSONAL STRESS
Very Low
Medium
High
Very High
Danger
Methodology
March 2016 15
Step 1b (cont’d)
97 people were surveyed in random control trial
The people were selected in the age group of 25-35 years of age
All the subjects who were overweight/obese were also stressed
Out of these, 33 (n=33) were identified as subjects who had either form of stress and were either overweight/obese.
Methodology
March 2016 16
Step 2
System Design
M.Tech Thesis |Saurav Gupta |CDAC
Methodology
March 2016 17
Step 2 (Cont’d) For Location
Google Fused Location API was integrated
Uses data from WiFi, Mobile towers and GPS available on smartphone.
Location determined based on latitude and longitude coordinates. Accuracy up to 50 meters.
Based on user inputs, 03 locations were classified as:
Home Location
Work Location
Other/Outside Location
Methodology
March 2016 18
Favorable Range Extreme Range
Temperature 18ºC - 35ºC < 17ºC & >35ºC
Humidity Level upto 90% >90%
Forecast
Clear Skies
Sunny
Cloudy
Windy
Rain
Thunderstorm
Hailstorm
Step 2 (Cont’d) For Temperature/ Weather
Open Weather API was used and integrated
Two divisions were done based on temperature: Favorable Range and Extreme Range. The grouping was done as follows:
Methodology
March 2016 19
Step 2 (Cont’d)
Rule Aggregator
Generates a code based on the datareceived.
For ex: for the code ‘H-I-AT’ generated, itsignifies that the subject is at Home, in anideal environment/ temperature and in anacceptable time zone.
Methodology
March 2016 20
Step 2 Application Screenshots
Methodology
March 2016 21
Step 3 Message Database
Researchers believe that a ‘positive and a motivating’ set of messages has a long lasting impact
A database was created for sending alerts to the subjects.
The format of the message was defined as:
‘Positive motivational message + Activity type’
A database of Positive and motivational messages was created
Activity type was determined based on the code generated by the RuleAggregator.
Broadly, outdoor activities were recommended only when the subjectwas at home and the weather was ideal. For the remaining, indoorrelated activities were advised to the subjects
March 2016 22
Results
The mobile application, ‘Let’s Exercise’ was installed on 33 subjects.
The evaluation study was done for a period of 04 weeks in the month of September 2014.
A post study questionnaire was filled by the participants.
Male Female
Series1 26 7
0
5
10
15
20
25
30
Axis
Tit
le
Gender Ratio
March 2016 23
Results
36%
44%
7%
13%
FIRST WEEK
Yes
No
Too busy to do it
Will do but later
42%
46%
2%10%
SECOND WEEK
Let's do it
No, thanks
Too busy to do it
Will do but later
60%
31%
2%
7%
THIRD WEEK
Let's do it
No, thanks
Too busy to do it
Will do but later
59%24%
3%14%
FOURTH WEEK
Let's do it
No, thanks
Too busy to do it
Will do but later
March 2016 24
FINAL OUTCOME
First Week Second Week Third Week Fourth Week
Series1 49% 52% 67% 73%
0%
10%
20%
30%
40%
50%
60%
70%
80%
PO
SIT
IVE
RE
SP
ON
SE
%A
GE
Response Trend
March 2016 25
Results
EFFECTIVENESS AA A N D DD
The technology was effective in understanding the user context 60.6 30.3 9.1 0 0
The technology accurately determines my location 57.6 30.3 12.1 0 0
The technology accurately determines the weather 48.5 36.4 15.1 0 0
The technology was efficient in understanding the user context 66.6 27.3 6.1 0 0
The technology has motivated me to do physical activities 69.7 27.3 3 0 0
USEFULNESS AA A N D DD
The technology provided is useful 54.6 36.3 9.1 0 0
The technology provided is informative 48.5 45.5 6 0 0
I would use this app frequently 48.5 39.4 9.1 3 0
The technology is convenient to use 63.6 30.3 6.1 0 0
SATISFACTION AA A N D DD
I am satisfied with the technology developed 60.6 24.2 15.2 0 0
The technology performed as expected 51.5 42.4 6.1 0 0
I would adopt this app as part of my daily routine 24.2 36.4 24.2 15.2 0
The prompts/ alerts were apt and appropriate 48.5 45.4 6.1 0 0
March 2016 26
Papers Published
S. Gupta and S. P. Sood, “Context Awareness Mobile Agent for reducing Stress and Obesity by
Motivating Physical Activity-a design approach”, 2nd International Conference on Computing for
Sustainable Global Development, IEEE- IndiaCom 2015; paper presented.
Paper also published in the Book ‘Proceedings of the 9th INDIACom, 2015 2nd International
Conference on Computing for Sustainable Global Development’ having ISSN 0973-7529 and
ISBN 978-93-80544-15-1
S. Gupta, S. P. Sood and D. K. Jain, “Let’s Exercise: A Context Aware Mobile Agent for Motivating
Physical Activity”, Third International Conference on Emerging Research in Computing,
Information, Communications and Applications, SPRINGER-ERCICA 2015; paper accepted.
Paper to be also published in Springer Series ‘Advances in Intelligent Systems and
Computing’ having ISSN No. 2194-5357
March 2016 27
Future Work
Application ‘Let’s Exercise’ published on Google Play Store and made available for free
download.
Integration of additional sensors and APIs
Work Calendar / Skype Status/ inclusion of diet plans
Integration of Fitness trackers
Social Media Integration
Gami-fy the system
Score activity based points/ credits
Share on social media
Compete with friends
March 2016 28
Conclusion
Context Aware Computing, with their ability to sense and adapt, can help bring about a
behavioral change amongst the individuals to take up physical activity.
Context Aware system could serve as a potential tool to solve real life health problems
and providing personalized healthcare services to individuals.
At a broader level, Context aware computing can help solve many health-related issues
and thereby help improve healthcare delivery.
