Session
15
PD,
Predictive
Analytics
for
Actuaries:
Predictive
Modeling
of
Health
Insurance
Big
Data
Moderator:
Brian
Matthew
Hartman,
ASA,
Ph.D.
Presenters:
Brian
Matthew
Hartman,
ASA,
Ph.D.
Session 15: Predictive
Analytics for Actuaries:
Predictive Modeling of
Health Insurance Big Data
Brian Hartman, ASA, Ph.D.
Chris Stehno
Big Data
What is it – and what does it mean for the insurance industry?
SOA Annual Meeting
Austin
October 12, 2015
Chris Stehno
Deloitte Consulting | US
Chris Stehno – Director - Advanced Analytics and Modeling
Chris Stehno is a Director at Deloitte Consulting in the
United States, as well as a member of Deloitte’s
Advanced Analytics and Modeling practice.
Chris has applied statistical and machine learning
methods to such diverse business problems as
healthcare utilization, customer and employee
retention, talent management, insurance agent
recruiting, customer segmentation, life and health
insurance pricing and underwriting, medical
malpractice and patient safety, claims management,
preventive healthcare, suicide prevention and fraud
detection. He is well known for the expansion of
traditional health risk analysis through the use of
non-traditional data sources and for developing behavioral
tactics to promote wellness and preventative services.
Chris is a frequent author and conference speaker.
Prior to Deloitte, he was the co-founder and President
of MedAnalytics, the company that pioneered the field
of Lifestyle Based Analytics in the healthcare arena.
Deloitte Consulting LLP
111 S Wacker Dr
Chicago
, IL 60606
Chris Stehno, MBA
Director
Deloitte Consulting | US
Tel: +1 312 206 4024
[email protected]
Member of
Big Data is in the news
From the dawn of civilization until 2003, humankind
generated 5 exabytes of data. Now we produce 5
exabytes every two days, and the pace is
increasing.
Eric Schmidt, Executive Chairman, Google
Every century a new technology – steam power,
electricity, atomic energy or microprocessors – has
swept away the old world with a vision for a new
one. Today, we seem to be entering the age of big
data.
Michael Cohen, Author, Speaker, Broadcaster
We’ll see this as a the time in history when the
world’s information was transformed from an inert,
passive state and put into a unified system that
brings the information alive and lives on forever.
The evolution of data science in insurance
1990s
Credit Scoring - an
early bellwether of the
disruptive power of
“big data” in insurance.
The evolution of data science in insurance
1990s
Credit Scoring - an
early bellwether of the
disruptive power of
“big data” in insurance.
2000s
Predictive modeling
becomes mainstream in
non-life insurance.
Personal insurance:
rating
plan and price optimization
Commercial insurance:
Underwriting, prospecting,
claim adjustment models
The evolution of data science in insurance
1990s
Credit Scoring - an
early bellwether of the
disruptive power of
“big data” in insurance.
2000s
Predictive modeling
becomes mainstream in
non-life insurance.
Personal insurance:
rating
plan and price optimization
Commercial insurance:
Underwriting, prospecting,
claim adjustment models
Today and tomorrow
Health / Life insurance:
Underwriting/risk triage
Application triage models
In-force management models.
Use of analytics to better
understand risks at
individual (not just group)
level
Telematics and self-tracking
devices link insureds to the
Internet of Things [IoT].
New data sources, new
business models …
Big data:
Three definitions of big data
1. Data sets with sizes beyond the capability of
standard IT tools to capture, process, and analyze
in reasonable time frames.
Three definitions of big data
1. Data sets with sizes beyond the capability of
standard IT tools to capture, process, and analyze
in reasonable time frames.
2. Data with high
V
olume,
V
elocity,
V
ariety
• Huge datasets
• … emanating continuously from smart phones, sensors,
cameras, GPS devices, computers, TVs, …
Three definitions of big data
1. Data sets with sizes beyond the capability of
standard IT tools to capture, process, and analyze
in reasonable time frames.
2. Data with high
V
olume,
V
elocity,
V
ariety
• Huge datasets
• … emanating continuously from smart phones, sensors,
cameras, GPS devices, computers, TVs, …
• … involving all manner of numeric, text, photographic data
Emerging trends - social analytics, quantified self, etc.
Digital Analytics
leverages a number
of different tools to collect social
conversations, then uses a
combination of automated and manual
processes to analyze the data.
“
Quantified Self
” applications such
as Fitbit, Apple Watch and Smart
Phone Apps allow customers to
monitor and share lifestyle/health
data
The development of Lifestyle Based Analytics (LBA)
Deloitte Consulting’s Proprietary Disease
State Algorithms
Using only third-party data we have built
algorithms to provide insights into individuals
afflicted with 20 plus lifestyle diseases (e.g.
diabetes, female cancer, tobacco related cancer,
cardiovascular, depression, etc.) which impact
morbidity. In addition we have used over 1 million
paramedical exam results to identify individuals
who are at extreme risk or have a condition that
has not been otherwise detected or diagnosed.
3
rd
Party Marketing Data Types
Disease State
1
Algorithms
Survey Data
– Self-reported information collected over the last 18 months
– Contains many lifestyle elements
Observed Data:
– Basic individual and household demographics
•
Age, sex, number and ages of children, marital status
•
Occupation categories, education level
– Financial information
•
Income level, net worth, savings and investments
•
Home value, mortgage value
– Lifestyle data
•
Activity — running, golf, tennis, biking, hiking, soccer, tri-athlete
•
Inactivity — TV, mail-order, computers, video games, casino gambling
•
Diet, weight-loss, exercise, cooking, gardening, health foods, pets
Small Area Characteristics:
– Matched to carrier route modeled data
– Reports average data for that route
– Approximately two city blocks
1 – Deloitte Consulting proprietary method
Third party marketing datasets are used to develop health-related algorithms. These datasets include over 1,000 fields of
data and the match rate with a client’s policyholders is typically around 95% based only on the individual’s name and address.
Lifestyle predictive analytics allow us to better understand
individual/population health risks
Beth
Tom
Sarah
• Female age 45
• Employed
• No significant claims
• Male age 46
• Employed
• Knee surgery
• Female age 46
• Unemployed
• No significant claims
T
ra
d
it
io
n
a
l d
a
ta
T
ra
di
ti
on
a
l
da
ta
• Renter / Owner
• Commutes 45 miles
• Bankruptcy indicator
• Diet/weight loss
purchases
• Fast food purchaser
• Self help books
• High TV consumption
• Manager level position
• Owns home
• Has lived in hometown
all his life
• Married with two children
• “Suburban Striver”
Psychographic Cluster
• Avid golfer
• New to town
• Reading: foreign
travel-related magazines
• Good credit
• Healthy food choices
• Little to no TV
consumption
• Running and yoga
L
if
est
y
le
-b
ased
d
at
a set
Cost index: 1.3
Diabetes Prob: 2.5
Average Cost
Expectation
Cost Index: 0.75
Diabetes Prob: 0.30
Risk
an
aly
sis
EHR – A current federal mandate
• American Recovery and Reinvestment Act of 2009
•
("stimulus package”)
• Established timeline for future incentives for health care providers
to offer patient health records in electronic format
• Healthcare providers which demonstrate “meaningful use” of
EHR receive increased levels of Medicaid and Medicare
reimbursements
2009
New Legislation
2011 – 2014
Positive Rewards /
Incentives for
“Meaningful Use”
of EHR
2015 +
Penalties
for Lack of
“Meaningful Use”
of EHR
• Healthcare providers which fail to demonstrate “meaningful use”
of EHR receive reduced levels of Medicaid and Medicare
Future disruptors – electronic medical/health records
The next big disrupter in
insurance market place
113883.10.20.3" /><templateId root="2.16.840.1.113883.10.20.1" /><templateId root="2.16.840.1.113883.3.88.11.32.1" /><templateId root="1.3.6.1.4.1.19376.1.5.3.1.1.5" /><templateId root="1.3.6.1.4.1.19376.1.5.3.1.1.2" /><templateId root="1.3.6.1.4.1.19376.1.5.3.1.1.1" /><templateId root="1.2.840.114350.1.72.1.77896772" /><id assigningAuthorityName="EPC" root="1.2.840.114350.1.13.126.2.7.8.688883.4663085" /><code code="34133-9" codeSystem="2.16.840.1.113883.6.1" codeSystemName="LOINC" dispxxxxxxName="Summarization of Episode Note" /><title>Patient Health Summary</title><effectiveTime value="20150507131421-0700" /><confidentialityCode code="N" codeSystem="2.16.840.1.113883.5.25" /><languageCode code="en-US" /><setId assigningAuthorityName="EPC" extension="a5af8356-f4f5-11e4-91ee-f802765cdbf3:121222" root="1.2.840.114350.1.13.126.2.7.1.1" /><versionNumber value="1" /><recordTarget><patientRole><id root="1.2.840.114350.1.13.126.2.7.3.688884.100" extension="K263709407" /><addr
use="HP"><streetAddressLine>XXXXXXXXXXX</streetAddressLine><city>XXXXXXXXXXX</city><state>CA</state><postalCode>91311-5538</postalCode><country>USA</country></addr><telecom use="HP" value="tel:+1-NNNXXXX-7743" /><telecom use="WP" value="tel:+1-NNNXXXX-2544" /><telecom use="MC" value="tel:+1-NNNXXXX-7743" /><telecom value="tel:+1-000-000-0000" /><telecom value="mailto://[email protected]" /><patient><name
use="L"><given>First</given><family>XXXXXX</family></name><administrativeGenderCode code="F" codeSystem="2.16.840.1.113883.5.1" codeSystemName="AdministrativeGenderCode" displayName="Female" /><birthTime value="19770101" /><maritalStatusCode code="M" codeSystem="2.16.840.1.113883.5.2" codeSystemName="MaritalStatusCode" displayName="Married" /><religiousAffiliationCode code="970" codeSystem="2.16.840.1.113883.5.1076" codeSystemName="ReligiousAffiliation" /><raceCode code="2106-3" codeSystem="2.16.840.1.113883.6.238" codeSystemName="CDC Race and Ethnicity" displayName="White" /><ethnicGroupCode code="2186-5" codeSystem="2.16.840.1.113883.6.238" codeSystemName="CDC Race and Ethnicity" displayName="American/United States" /><languageCommunication><languageCode code="eng" /><preferenceInd value="true"
/></languageCommunication><languageCommunication><languageCode code="eng" /><modeCode code="EWR" codeSystem="2.16.840.1.113883.5.60" displayName="Expressed Written" /></languageCommunication></patient><providerOrganization><id root="1.2.840.114350.1.13.126.2.7.2.688879" extension="6100" /><name>Kaiser Permanente Southern California</name><telecom use="WP" value="tel:+1-858-614-3333" /><addr
nullFlavor="NA"><city>Pasadena</city><state>CA</state><postalCode>91103</postalCode></addr></providerOrganization></patientRole></recordTarget><author><time
value="20150507131421" /><assignedAuthor><id nullFlavor="NA" /><addr nullFlavor="NA"><streetAddressLine nullFlavor="UNK" /><city nullFlavor="UNK" /></addr><telecom nullFlavor="NA" /><assignedPerson><name nullFlavor="NA" /></assignedPerson><representedOrganization><id root="1.2.840.114350.1.13.126.2.7.2.688879" extension="6100" /><name>Kaiser Permanente Southern California</name><telecom use="WP" value="tel:+1-858-614-3333" /><addr
nullFlavor="NA"><city>Pasadena</city><state>CA</state><postalCode>91103</postalCode></addr></representedOrganization></assignedAuthor></author><author><time value="20150507131421" /><assignedAuthor><id root="1.2.840.114350.1.1" extension="8.1" /><addr nullFlavor="NA"><streetAddressLine nullFlavor="UNK" /><city nullFlavor="UNK" /></addr><telecom nullFlavor="NA" /><assignedAuthoringDevice><manufacturerModelName>Epic - Version 8.1</manufacturerModelName><softwareName>Epic - Version 8.1</softwareName></assignedAuthoringDevice><representedOrganization><id root="1.2.840.114350.1.13.126.2.7.2.688879" extension="6100" /><name>Kaiser Permanente Southern California</name><telecom use="WP" value="tel:+1-858-614-3333" /><addr
nullFlavor="NA"><city>Pasadena</city><state>CA</state><postalCode>91103</postalCode></addr></representedOrganization></assignedAuthor></author><custodian><assignedCustodian> <representedCustodianOrganization><id root="1.2.840.114350.1.13.126.2.7.2.688879" extension="6100" /><name>Kaiser Permanente Southern California</name><telecom use="WP" value="tel:+1-858-614-3333" /><addr
nullFlavor="NA"><city>Pasadena</city><state>CA</state><postalCode>91103</postalCode></addr></representedCustodianOrganization></assignedCustodian></custodian><participant typeCode="IND"><templateId root="2.16.840.1.113883.3.88.11.32.3" /><templateId root="2.16.840.1.113883.3.88.11.83.3" /><templateId root="1.3.6.1.4.1.19376.1.5.3.1.2.4" /><time value="20130207" /><associatedEntity classCode="ECON"><id nullFlavor="UNK" /><code
nullFlavor="OTH"><originalText>Husband</originalText></code><addr><country>USA</country></addr><telecom use="MC" value="tel:+1-NNN-421-7000" /><associatedPerson><name use="P"><given>GIANN</given></name></associatedPerson></associatedEntity></participant><documentationOf typeCode="DOC"><serviceEvent classCode="PCPR"><effectiveTime><low value="19770101" /><high value="20150507" /></effectiveTime><performer typeCode="PRF"><templateId root="2.16.840.1.113883.3.88.11.32.4" /><templateId root="2.16.840
