Data science and the use of big data in healthcare delivery could revolutionize the field by decreasing costs and vastly improving efficiency and outcomes. There is an abundance of healthcare data in Canada, but it is mostly siloed and difficult to access due to privacy and security challenges. This session will offer insights into best practices for healthcare analytics programs, as well as use cases that demonstrate the potential benefits that can be realized through this work.
1. HEALTHCARE DATA ANALYTICS
Unlocking The Power Of Healthcare Data TOGETHER
Lisa Lix, PhD, P.Stat.
September 30, 2015
Cyber Summit Generation D: Data Scientists of Tomorrow
2. Outline
Why together is better
• Where I work and do my research
• Examples of healthcare analytics in action
• What lies ahead
3.
4. George and Fay Yee Centre for
Healthcare Innovation (CHI)
• CHI is a partnership between the University of Manitoba
and the Winnipeg Regional Health Authority
• CHI brings together leaders and practitioners from many
academic disciplines and areas of practice
• CHI aims to:
– improve patient outcomes,
– enhance patient experiences, and
– improve access to care
for Manitobans
6. CHI’s Data Science Platform
Our Activities:
Research
Collaboration
Training
Clinical
Research Data
Group
Biostatistics Group
Bioinformatics
and
Computational
Biology Group
Our Vision: To create and integrate diverse types of patient data
and develop and apply the best analytic methods to provide new
insights about patient outcomes, experiences, and care
7.
8. Healthcare Analytics in Action: Provincial
Manitoba Centre for Health Policy Data Quality
Framework
9. Manitoba Centre for Health Policy
Research Data Repository
Population- Based
Health Registry
Social
Housing
Education
Healthy
Child
Manitoba
Immunizatio
n
Medical
Services
Lab
Nursing
Home
Clinical
Provider
Vital
Statistics
Emergency
Dept.
Health Links
Home Care
Pharmaceutical
s
Hospital
Family
Services
Income
Assistance
Census
Data
• Family First
• Healthy Baby
• Intensive
Care Unit
• Fetal Alcohol
Spectrum
Disorder
• Pediatric
Diabetes
11. Healthcare Analytics in Action: National
• The Canadian Chronic Disease
Surveillance System (CCDSS)
• The Canadian Network of Observational
Drug Effect Studies (CNODES)
12. Healthcare falls primarily under
the authority of the provinces
and territories
The provincial and territorial
healthcare systems differ in
structure and operation
This results in a patchwork of
systems and data resources
The Canadian Healthcare System
13. The Canadian Chronic Disease
Surveillance System (CCDSS)
Established by the Public Health Agency of Canada
(PHAC) as a collaborative initiative amongst the federal,
provincial, and territorial governments
Uses health administrative data to estimate chronic
disease prevalence/incidence and the related burden on
the healthcare system
Adopts a distributed surveillance system model that
respects the data custodial responsibilities of the
provinces and territories
Provides a standardized pan-Canadian approach to
chronic disease surveillance
14. Multimorbidity: An Example
Note: The 95% Confidence Intervals shows an estimated range of values which is likely to include the
statistic 19 times out of 20. Data Source: Public Health Agency of Canada: using CCDSS data files
contributed by the provinces and territories as of August 2015
15. Note: The 95% Confidence Intervals shows an estimated range of values which is likely to include the
statistic 19 times out of 20
18. Canadian Network of Observational Drug
Effect Studies (CNODES)
• Network of over 60 Canadian pharmacoepidemiologists,
biostatisticians, clinicians, clinical pharmacologists,
pharmacists, IT professionals, data analysts, and
students using linked administrative data in 7 provinces
plus UK and US data
• Timely responses to queries from Canadian public
stakeholders about drug safety and effectiveness
20. CNODES Database Model
• Data partners maintain physical control of
their data
• Local content experts maintain a close
relationship with the data
• Eliminates the need to create, secure,
maintain and manage access to a central,
complex data warehouse
• Gives a pan-Canadian meta-analysis
“answer” that dramatically increases sample
size for rare events, RAPID RESPONSE
21. CNODES Project
Isotretinoin Use
Amongst Women of
Reproductive Age
and the Risk of
Pregnancy and
Adverse Pregnancy
Outcomes
• Population based
• Multi-province
participation
• Pregnancy/Outcome
s in isotretinoin users
• US Comparisons
23. What Lies Ahead?
• Data Linkage
– Images
– Streaming data from wearable devices
– Electronic medical records
• Analyses
– Biases
– Rare events
• Data Visualization
• Formalized Training
24. Training in Healthcare Analytics
• Strategic:
– Focussed on performance
– Strategic thinking and communication skills
– Less essential to have skills in the technical, nitty-gritty details of
setting up database systems and defining or selecting algorithms
• Operational:
– Training in programming, statistics, mathematics
– Skills in implementing systems to probe and interpret data
25. Essential Skills
• Constructing data queries
• Manipulating data into different formats or
structures
• Modeling & analysis
• Telling the story of the data