Transcript DiabeticLink: An Integrated and Intelligent Cyber

DiabeticLink: An Integrated and Intelligent
Cyber-Enabled Health Social Platform for
Diabetic Patients
Joshua Chuang1, Owen Hsiao2, Pei-Lin Wu2, Jean Chen2, Xiao Liu3,
Haily De La Cruz3, Shu-Hsing Li2, and Hsinchun Chen3
1Caduccus
Intelligence Corporation
2National Taiwan University
3Artificial Intelligence Lab, University of Arizona
1
Outline
•
•
•
•
•
•
Introduction
Literature review and related systems
System functionalities
User studies and lesson learned
System adoption and usage statistics
Future development plan
2
Introduction
• 371 million people are living with diabetes all over the
world (International Diabetes Federation) by 2012.
– Shortage of healthcare resources
– Calls to focus on patient empowerment
• The growth of health social media
– Capability to satisfy users’ information needs and provide
emotional support (e.g., PatientsLikeMe and DailyStrength )
– A new market space for social media patient empowerment
tools in the healthcare delivery
3
Introduction
• A growing need for online social communities and
patient portals
– Allow patients to connect and promote disease selfmanagement
• An integrated cyber-enabled diabetic patient
empowerment portal
– Help diabetes prevention and management
– Alleviate some pressure on the current healthcare system
4
Introduction
• Motivated by these factors, we aim to
– Develop a patient portal, DiabeticLink, targeting both
US and Taiwan audiences
– Enable patient social connectivity
– Provide diabetes management tools for diabetic
patients with advanced data, text and web mining
techniques
5
Literature Review and Related Systems
• To build the foundation of new features for DiabeticLink,
we investigated the existing competitors in the following
areas:
– Diabetes patient portals and online social communities
– Diabetes tracking and progression visualization
– Diabetes risk engines
6
Diabetes Online Patient Portals
DiabeticLink
dLife
ADA
Diabetes.co.uk
Health information
mashup (news, blogs,
recipes, videos,
twitter)
Risk Engine
Yes, but no risk
of
hospitalization
Restaurant Search by
calories
Portal tracking
Community Forums;
Social Connectivity;
Food & Fitness
Guides;
Diabetes Guides
7
Diabetes Tracking and Progression Visualization
• Tracking applications : track diabetic measurements such as
blood glucose, Hemoglobin A1c, carbohydrates and physical
activity.
• The current offerings in the market are divided into products
that are
– Focusing on exercise and weight management (e.g., SparkPeople and
MyFitnessPal)
– Mobile diabetes management systems to track blood glucose, food and
activity levels (e.g., Glucose Buddy and OnTrack)
8
Diabetes Tracking and Progression Visualization
App
DiabeticLink
(web based)
Glucose
Buddy
(Web, iOS,
Android)
Telcare
Diabetes Pal
(Web, iOS,
Android)
GluCoMo
(iOS)
BGluMon
(iOS)
OnTrack
(Android)
dLife
(iOS)
Glucose
Food Medication
Exercise
Hb
A1C
B.P.
Weight

































(insulin only)


9
Diabetes Tracking and Progression Visualization
• In terms of progression visualization, many applications
provide basic charting.
• PatientsLikeMe provides patients with a platform to
enter and view his/her own disease data in a more indepth visualized manner.
– All disease categories on one single site
– Less useful for regular diabetic patients
10
• Diabetes Risk Engines
• The goals of patient empowerment are to
– Advance preventive and personalized care
– Raising their awareness of their risks and conditions
• Diabetes risk engines use predictive modeling techniques with
longitudinal patient data to
– Predict the risk of specific medical events for patients based on their
health indicators
• The most well-known risk engine, UKPDS risk engine, predicts
heart disease and stroke for type 2 diabetes patients based on
clinical trial data from 53,000 patients.
11
Literature Review and Related Systems
• Based on our review of related systems, we identified
the following gaps:
– The design of social features in many portals does not promote
deeper user interaction through linkage to users’ personal
health records.
– Most diabetes tracking apps lack the ability to facilitate users’
learning from their data.
– The focus of diabetes tracking apps is logging instead of driving
and facilitating improvements in patient health.
12
Literature Review and Related Systems
Social community
Health information
mashup
DiabeticLink
Tracking and
monitoring
Risk assessment
13
System Functionalities
• The current project involves building the
DiabeticLink patient portal for both US and
Taiwan audiences.
– The functionalities of each are summarized in the
following sections.
14
System Functionalities
• DiabeticLink-US
– Social community platform
– Tracking
– Health information mashup
– Risk engine
15
DiabeticLink-US: Social community
Number of people in
discussion
Topics
Main features: activity feeds, friending, commenting, user profile, user blog, private
messaging, login with Facebook/Twitter, discussion forum
DiabeticLink-US: Tracking
Tracking
options
Logs
Tracking:
• Allows monitoring of all aspects of health
• Trends can be observed easily with advanced visualizations
• Reporting made easy to physicians
17
DiabeticLink-US: Mashup
Sort by
source
Dynamic
tag cloud
with
keywords
• Health information aggregation from a number of credible sources
• Health social media content in the form of videos, tweets, and blogs
• Promoting meaningful use of social media and information available online
18
DiabeticLink: Risk engine
Landing page with basic
descriptions about the
purpose, methods, and
references of the risk engines
Risk engine
• Leveraging UKPDS’ Risk Engine to assess risk of stroke and CHD
• Use DL’s own Risk Engine to determine risk of hospitalization
• Actionable recommendations for risk reduction
19
DiabeticLink-US: Risk engine
Input values that are
significant to hospitalization
and stroke predictions
Optional inputs regarding
diagnoses and drugs
20
System functionality: Risk engine
Show how well the patient
performs in each philological
indicator
The horizontal line in the
middle of the chart represents
the performance of an
“average patient”
21
DiabeticLink: Risk engine
Set personal goals and
perform “what-if”
analysis
Identify risk reduction
opportunities
22
System Functionalities
• DiabeticLink-Taiwan has been publicly
available since July 2013.
• The services DiabeticLink-Taiwan provides are
categorized into three main functions:
– Education
– Social connections
– Online disease management tools
23
DiabeticLink-TW: Education
Education: Provide hundreds of news articles, healthcare and medication related articles,
research reports and healthy diet recipes and provide resources for users to search for
frequently asked questions from diabetes patients and caretakers.
24
DiabeticLink-TW: Social connection
Social connections: provide forum discussion, personal blog, private message, and
friending functions to form a diabetes supportive environment
25
DiabeticLink-TW: Tracking
Tracking: keep on track of nine major health indicators such as blood glucose, diet,
insulin, medication, activity, blood pressure, weight, HbA1C and cholesterol. 26
DiabeticLink-TW: Drug safety
Drug safety: make adverse drug event reports of diabetes medications from US FDA’s
Adverse Event Reporting System (FAERS) database available to users to help them
27
better understand the drugs they are prescribed
User Studies and Lessons Learned
• We conducted user test for DiabeticLink-Taiwan beta
release.
• Feedbacks are collected from both diabetic patients and
health professionals regarding their experience of our
functionalities including:
–
–
–
–
–
Health information
Healthy recipe search
Healthy restaurant search
National Health Insurance (NHI) data query
Social forum and member profile
28
User Studies and Lessons Learned
 The above chart shows the average scores on a five-point rating scale for each section.
 From the user tests, we found that users valued the relevancy and accuracy of the
information on our website.
 This is shown by the high score of information practicability in each section.
29
User Studies and Lessons Learned
In order to understand users’ behavior and reactions toward our Tracking module,
we also conducted a user test.
 The current Tracking module received 5 out of 6 rating.
 The high rating shows that the users are satisfied with the features we included
in our Tracking module.
 Moreover, the usability of the Tracking module, which received 4.9 out of 6
rating, also shows that it had a fairly easy-to-use interface for the users.
 User suggestions included showing a larger font-size for senior users, providing a
mobile app, and integrating with glucometers to reduce user-input errors. 30
System Adoption and Usage Statistics
• In the nine-month period since July 2013, DiabeticLink-Taiwan
attracted more than 33,357 page views with over 6,805 visits,
3,615 unique visitors and 146 registered members.
– The average visit time was over 4 minutes with 4.9 pages viewed per visit.
– The Tracking module accounted for 35% of website traffic, followed by
health information (17%), user test (7%), forum (7%) and others.
• The fact that our Tracking module attracted 35% of the website
traffic shows that the features provided by our Tracking modules
show promise for meeting the needs of diabetes patients.
31
System Adoption and Usage Statistics
• From our website analytical tool, we also found that
–
–
–
–
60% of our users were between the ages of 18 and 34;
33.5% were between 25 and 34;
27.5% were between 18 and 24;
and 40% were over 35 years of age.
• This shows that we are reaching primarily a younger
population.
– Promote our website and to reach a greater number of
potential users.
32
Future Development Plans
• In July 2013, DiabeticLink-Taiwan released its beta version and became
the first diabetes social media platform in Taiwan.
– After the first year of website operations, we now have a better
understanding of the needs of the diabetes patients and other user groups.
• In the future, DiabeticLink-Taiwan will continue
– Collecting and providing users with relevant and accurate information
from credible sources.
– Collecting feedback from medical personnel including physicians,
registered nurses, and educational personnel within the endocrinology
departments of various hospitals in Taiwan.
• Due to the day-to-day contact with actual diabetic patients, these
personnel have the potential to provide new insights and valuable
feedback to the DiabeticLink website and the ongoing development of
our online tools.
33
Future Development Plans
• With production experience and lessons learned from
DiabeticLink-Taiwan, DiabeticLink-U.S. is under active
development for a new version to be publicly released in
August 2014.
• Future goals for DiabeticLink
-
Automatic data capture- complete integration with 3rd party
medical devices
-
Event planner (community and individual)
-
Package insert, FDA’s Adverse Event Report and social media
adverse event discussions
-
Integration of risk analysis results with individual user profiles
-
Link with local support group activities and events
34
References
•
•
•
•
•
•
•
•
•
•
•
International Diabetes Federation Annual Report:
http://www.idf.org/sites/default/files/IDF_Annual_Report_2012-EN-web.pdf
Greaves, F., Ramirez-Cano, D., Millett, C., Darzi, A., & Donaldson, L. (2013). Harnessing the cloud of
patient experience: using social media to detect poor quality healthcare. BMJ quality & safety, 22(3),
251-255.
Chen H., Compton S., & Hsiao O. (2013). DiabeticLink: A Health Big Data System for Patient
Empowerment and Personalized Healthcare. In Conference Proceedings of the International Conference
for Smart Health (ICSH). Beijing, China, August 3-4.
Stevens, R. J., Kothari, V., Adler, A. I., and Stratton, I. M. 2001. The UKPDS Risk Engine: a model for the
risk of coronary heart disease in Type II diabetes (UKPDS 56). Clinical Science 101(6) 671–679.
Kothari, V., Stevens, R. J., Adler, A. I., Stratton, I. M., Manley, S. E., Neil, H. A., and Holman, R. R. 2002.
UKPDS 60: risk of stroke in type 2 diabetes estimated by the UK Prospective Diabetes Study risk engine.
Stroke; a journal of cerebral circulation 33(7) 1776–1781.
Companion App by mySugr, https://mysugr.com/companion/
Liu, X., & Chen, H. (2013). AZDrugMiner: an information extraction system for mining patient-reported
adverse drug events in online patient forums. In Smart Health (pp. 134-150). Springer Berlin Heidelberg.
Lin, Y.-K., Chen, H., Brown, R. A., Li, S.-H., and Yang, H.-J. 2014. Time-to-Event Predictive Modeling for
Chronic Conditions using Electronic Health Records. IEEE Intelligent Systems (Forthcoming)
Patient@home, http://www.en.patientathome.dk/
Technology Quarterly. The quantified self, counting every moment, technology and health: measuring
your everyday activities can help improve your quality of life, according to aficionados of “self-tracking”.
The Economist. 2012. http://www.economist.com/node/21548493
Withings API, http://www.withings.com/api
35