mining minds: an innovative framework for personalized health and wellness support
TRANSCRIPT
Dr. Oresti Banos
Ubiquitous Computing Lab (UCLab)
Kyung Hee University, South Korea
http://uclab.khu.ac.kr/oresti
9th International Conference on Pervasive Computing Technologies for Healthcare (Pervasive Health 2015)
Istanbul, Turkey
Mining Minds: an innovative framework for personalized health and wellness support
/“The Slow-Moving Public Health Disaster”
Diseases linked to lifestyle choices are currently the biggest cause of death worldwide:• Cardiovascular conditions, cancers, chronic respiratory
disorders, obesity and diabetes, represent more than 60% of global deceases, half of which are of premature nature
• Most of these diseases are fairly associated to common risk factors, namely, tobacco and alcohol use, unwholesome diet and physical inactivity
• This "lifestyle disease" epidemic causes a much greater public health threat than any other epidemic known to man
• Millions of lives could be saved if the world over the next decade invests $1-3 per person on promoting healthier habits
2
Global targets for prevention and control of “lifestyle diseases” to be attained by 2025
Source: WHO, “Global status report on noncommunicable diseases 2014,” World Health Organization, Tech. Rep., 2014.
/Digital Health Revolution
• ICT are called upon to be a cornerstone of the new health era, playing a crucial role in empowering people to take charge of their own health and wellness, by providing them timely and ubiquitously with personalized information, support and control• Many applications and devices are increasingly available;
however, these systems are not currently meeting the needs of those they serve, and there is a paucity of current offers adding value• The immediate targets of these solutions are healthy lifestyle
services, especially oriented to the fitness domain, which primarily allow to track primitive user routines and provide simple motivational instructions
3
Need of Digital Health and Wellness Frameworks!
/Key Limitations of Existing Digital Health Frameworks
• Most mobile health frameworks are bound to the computational capabilities of the smartphone, require continuous maintenance and updates of end-user applications and normally trap data into their devices • Moreover, multiple systems and applications can be
generate similar health data and outcomes leading to unnecessary redundancy and overcomputation• These systems mostly operate on-demand, thus
determinants of health and wellness states can be also lost if not registered in a continuous manner • Platforms devised to share and integrate health and
wellness data underuse cloud resources, by only utilizing them for storage
4
/Mining Minds in a Nutshell 5
“Collection of innovative services, tools, and techniques, working collaboratively to investigate on human's daily-life routines data generated from heterogeneous
resources, for personalized wellbeing and healthcare support”
/Mining Minds Scope 6
Pers
onal
ized
Hea
lthca
re
Man
agem
ent S
ervi
ces
Personal Big Data
Variety
Velocity
Volume
Evolutionary KnowledgeKnowledge
Feedback
User Adoption and EngagementUI/UX
Education
Goal Objectives Challenges
7/
Smart Cup
Smartphone
Survey Data
Social Networks
Wearable Sensor
Kinect Camera
Personal big data
Volume• 800 thousand personal
data• 5 billion SNS data
Analysis &Processing
Existing Big Data PlatformsProposed Big Data Platform
Multimodal Sensor
Variety
Velocity
Heterogeneous sensory data and structured and unstructured diverse big data processing• Conformed data structure• Data Representation & Mapping
Real time data processing technology which requires timely analysis• Real-Time Data Labeling• Streaming Data Retrieval and Inter-
mediate Data Generation
Privacy
Personalized data protection technology• Service Aware Autonomous
anonymization technology• Oblivious Term Matching• Private Matching
Hong gil dong, KHU180cm, age 25
->Hong**, **Univ170-180cm, age 20-30
Oblivious Term Matching
Hong gil dong, KHUKim chul su, KHU
->86e0109, 638560c691ed13, 152aa3a
Private Matching
Real-Time Sensor Data:1.2, 1.0, 2.2, 3.1
->1.2, 1.0, 2.2, 3.1, “Work”
Real-Time Data Labeling
“Work“, “Seould Gangnam”, “16C”, “165kcal”
-> “Work”, “165kcal”
Streaming Data storing(Storing automatic data selection)
Mining Minds Aims: Personal Big Data
8/
Generate structured knowledge
Knowledge Base
Provide recommendation
service
Existing Knowledge Maintenance SystemsExercise, activity, etc.
Structured static knowledge
Mining Minds Aims: Evolutionary Knowledge
Feedback
Knowledge maintenance engine
Update knowledge Userrequirements
Knowledge Maintenance
Knowledgebase update technique based on user feedback • Expert and automatic knowledge
maintenance• Multi-level maintenance
SelectorAutomatic Algorithm selection using Meta-learning• Meta-features computation• Algo. performance evaluation• Problem meta-features to Algo.
performance Mapping
Classification Algorithms-> J48, SVM, NB, ...
Knowledge Management-> Data Curation,
Information Curation,Service Curation
Personalized dynamic knowledge
Proposed Knowledge Maintenance System
/ 9Existing UI/UX Technology
Create UI/UXRule
UI/UX Knowledge
Gender, age, Using pattern… etc
Structured static knowledge
ProvideUI
Provide Feedback UI
UI/UX Authoring tool
Gender, age, using pattern, feedback, etc
Personalized dynamic knowl-edgeAdaptive UI/UX
Context based personalized and customized UI• Adaptive UI• UX
Survey individual UX
BehaviorMeasurement
User-machine interaction analysis based on UX• Feedback• Behavior Measurement
Trust: App Usage LessInteraction: Less No of Clicks
Reaction: ComplexityFunctionality: Less features
Predictability: Easy NavigationIndividuality: Color Scheme
Induce habituation
Mining Minds Aims: User Adoption and Engagement
Proposed UI/UX Technology
10/Mining Minds Architecture
Delivers timely and accurate personalized cross-domain recommendation based on domain knowledge and users preferences/context
Creates and maintains health and wellness knowledge using expert-driven and data-driven approaches
Provides real-time data acquisition from multimodal data sources and its persistence using big data technologies. Activity and context data are mapped for life-logging and personalized predictions from life-log ontology
Facilitates information to the users in the most intuitive
manner, in a secure environment reflecting their personal needs
and preferences
Converts the data obtained from the user interaction with the real and cyberworld, into abstract concepts or categories, such as physical activities, emotional states, locations and social patterns, which are intelligently combined to determine and track context and behavior
/Mining Minds Scenario
• Personalized Recommendations• Preferences, Activity Level and
Possessions
• MM Platform development• Services based on layered
architecture
• Personalized Big Data Processing• Considers multiple users
• Users Feedback• For knowledge evolution
11
/
Feature exists (fully) Feature exists (partially) Feature does not exist
Mining Minds Core Platform vs Existing Solutions
/Conclusions 24
• Lifestyle diseases linked to unhealthy habits kill millions of people prematurely• Digital health solutions are increasingly available; however, application-specific
systems present important limitations to widely inspect on human’s lifestyles • Mining Minds, a novel digital framework, is presented to seamlessly investigate
on people’s behavior and lifestyles in an holistic manner, through mining human’s daily living data generated through heterogeneous resources• An initial realization of the key architectural components, as well as an
exemplary application that showcases some of the benefits provided by Mining Minds, have also been presented.• Next steps include to complete the implementation of the devised architecture
as well as to evaluate its services on a large scale testbed
Thank you for your
attention. Questions?
25
Dr. Oresti BañosUbiquitous Computing Lab (UCLab)Kyung Hee University (KHU), South
KoreaEmail: [email protected]
Web: http://uclab.khu.ac.kr/oresti