The opinions expressed are those of the presenter and do not necessarily state or reflect the views of SHSMD or the AHA. © 2014 Society for Healthcare Strategy & Market Development
Using Big Data &
Predictive Modeling to
BIG DATA IN HEALTH CARE
Size & Growth of Health Care Data
Health care data will reach 35 Zettabytes
by 2020, a 44-fold increase from 2009.
How Big is it Really?
Source: Cisco 1 ZETTABYTE 250 BILLION DVDSBig Data Utilization Gap
“Most Wired” Hospitals
• 61% use predictive modeling and data to improve decision-making across multiple
departments.
• 36% conduct controlled experiments or scenario planning to make better
management decisions and to do forecasting.
All Other Hospitals
• 49% use predictive modeling and data to improve decision-making across multiple
departments.
• 27% conduct controlled experiments or scenario planning to make better
management decisions and to do forecasting.
The Size/Relevance Disconnect
SOLUTION:
Make Big
Data
“Small”
Data Size Does Not Equal Data
Relevance
Better to Use Right Data Than
MAKING BIG DATA SMALL
Tips for Making Big Data Small
• Use only the data that you need to answer the question/solve the problem • Strategically combine in-house data with outside data sources that enhance
the predictive power of your internal data
INTERNAL DATA: Names & Addresses EXTERNAL DATA: Psychographics & Lifestyles Complete Picture of Patients & Ability to Predict New Patients
PATIENT ENGINE
Who/Where/Value
MARKET
PLANNING MARKETING OPTIMIZATION SERVICE LINE OPERATIONS
SCOUT DATA MART
Applications for Marketing &
Network Development
DEMOGRAPHICS Age: 35-50 Income: $75 – 100k Married DEMOGRAPHICS Age: 35-50 Income: $75 – 100k MarriedApplications for Marketing &
Network Development
DEMOGRAPHICS Age: 35-50 Income: $75 – 100k Married DEMOGRAPHICS Age: 35-50 Income: $75 – 100k Married PSYCHOGRAPHICS Tom King 1308 Center St PSYCHOGRAPHICS Ben Shaw 1010 Imperial WayVisits Chiropractor 2x/month Prefers To Deal In Cash Receptive To Direct Mail
Keeps Annual Dermatology Appt. Spend $150+wk At Grocery Store
Prefers To Visit Urgent Care Clinic Visa Credit Card Holder
Gather Health Info Online Takes OTC Asthma Medicine Spend $250/Month On Clothing
CASE STUDY: MERCY
Who is Mercy Health?
• Health system founded by the Sisters of Mercy in 1986
• 5th largest Catholic Health Care System in the U.S. with a presence in Arkansas, Kansas, Missouri, and Oklahoma
• 33 Acute Care Hospitals
• 4 Heart Hospitals
• 2 Children’s Hospitals
• 3 Rehab Hospitals
• 1 Orthopedic Hospital
• 700 Clinic And Outpatient Locations
• 2,000 Integrated Physicians
Why Big Data & Analytics?
Expansion and Optimization
• Understanding who our best patient is
• By region, population density, service line
• Long term planning!
• Where is the demand?
• Where is the supply/competition?
Long Term Planning
•
Created a long-term growth plan for each community
•
Organic and acquisition
•
Applied analytics to markets of all sizes and densities
•
Urban, suburban, rural
Data Applied at the Facility Level
• St. Louis community • Primary Care facility
• Currently an average-performing location
• Opportunity to increase usable square footage with upcoming remodel
FACILITY #5
Utilize Patient Engine for St. Louis Evaluate Current Performance Identify New Opportunity Determine Influence of Existing Network
Apply Big Data
Current Service Lines Potential Service Lines
Family Medicine Cardiology
Internal Medicine OB/GYN
Pediatrics Orthopedics
Current Service Lines Potential Service Lines
Family Medicine Cardiology
Internal Medicine OB/GYN
Pediatrics Orthopedics
Target best patient look-a-likes to
drive traffic
This site is too close to our Heart Hospital
Market Analysis Case Study
• Devastating tornado hit Joplin, MO on May 22,
2011
• 161 deaths
• 8,000 structures destroyed or damaged
• 4,000 businesses destroyed or damaged
• St. Johns Mercy Hospital severely damaged
• Analytics deployed to identify the top 3 sites for a
replacement hospital
• Accounting for households lost
• Accounting for expected rebuild over 5-10
years
• 2nd highest rated site selected for the new
Marketing Application
Once market planning is selecting sites with a high concentration of your best potential patients, use data to help talk to them in their own words.
• Empty-nesting couples enjoying active leisure lives
• Increasingly health conscious • Have regular medical checkups • Take preventative medicine • Gather health information from
websites
Marketing Application
• Maximize results by targeting the right households based on campaign type
• Example: Ladies’ Night Out
– Start with gender and age requirements
– Overlay patient engine developed by studying Women's Services patients from the area
– Utilize model to select the most likely patients and prospects – Suppress patients who have already had the services we are
The opinions expressed are those of the presenter and do not necessarily state or reflect the views of SHSMD or the AHA. © 2014 Society for Healthcare Strategy & Market Development BILL STINNEFORD Buxton 817-332-3681 [email protected] NIKKI VINER Mercy 314-628-3412 [email protected]