Embedding Predictive Analytics in All Aspects of a Health Plan Ian - - PowerPoint PPT Presentation

Embedding Predictive Analytics in All Aspects of a Health Plan

Ian Blunt Health Datapalooza 3/27/2019

HIGHMARK.COM

Overview of Highmark Health

2

• Aiming for transformational change in

healthcare delivery and financing

• We will be focusing on the Health Plan

today

• One of nation's 10 largest health

insurance organizations

• Geographic Service Area includes

Pennsylvania, West Virginia and Delaware

• Over 5 million active members
• Customer Engagement & Insights is a

~200 person organization focused on advanced analytics, data science and member engagement

• Highmark is one of the largest integrated

healthcare, delivery and financing networks in the nation

• We are a national healthcare brand, we are

a leading edge technology platform, we are a healthcare provider.

• Employs more than 40k people nationwide

and serves over 50 million Americans in all 50 states

• Our Mission: To create a remarkable health

experience, freeing people to be their best.

• Our Vision: A world where everyone

embraces health

Our journey into advanced analytics

3

Enrollment and claims, read-only Wide range of datasets, growing API ecosystem, push to frontline systems Full interoperability with partners’ systems

Data and interoperability Analytics Tools and talent Evidence driven culture

2016 2019 Beyond

Retrospective, descriptive, manual Predictive, near-time, automated Seamless prescriptive analytics Mainly SAS, many staff in same role 10+ years SAS plus full open source stack (R, Python, H2O, Keras etc). Staff rotation and skills matrix Scouting for new technologies, deep AI integration, leading talent culture “Gimme ‘the data’ ”, varying definitions caused swirl and distrust Partnership for insight, standard data definitions, increasing self-serve, robust evaluation Insights embedded within business, innovation center, data “in the DNA”

Embedding predictive analytics in all aspects of a health plan

4

Customer Engagement & Insights

Highmark proactively identifies when a member would benefit from a clinical intervention

Providers

Highmark meets the member where they are in their journey using personalized messaging Highmark continuously engages the member (via multi-channel) to achieve desired outcome Clients receive tailored recommendations to improve quality and lower cost for their employees based on their risk profile Clients understand the value Highmark’s programs create for them through robust evaluation Predictive analytics embedded in the underwriting process Machine learning provides customized recommendations around lower efficiency care episodes Share predictive analytics with strategic partners to enable joined-up intervention with members Provider staff is well trained on the Highmark analytic suite and can easily generate their own actionable insights

Maximizing available data produces more impactful analytics

5

• Common to predict likelihood of direct-pay member

retention – Highmark was using basic age and tenure regression model in early 2010s

• We wanted to gain the maximum insight from our retention

model, by predictive power and influential factors

• Dramatically increased the scope of data we trained the

model on by chasing processes and feature engineering

• Our predictive power now matches leading published

retention work in other industries, yielded new business insights, shaping retention strategy

• Data sources unlocked can be used for other modeling

projects Modeling retention in our Medicare Advantage direct pay members

Demographic Claim activity Clinical

• utreach

engagement Network disruption Enrollment Access Social determinants Household membership Provider affinity Product change Health program engagement Complaints Customer service calls Marketing responses Stars metrics Morbidity and frailty Base 2015 model Added to 2017 model Added to 2019 model

Comprehensive data assembly and creative feature engineering are a powerful asset in driving value from predictive analytics

Inserting analytics into operational processes drives value

6 Using advanced analytics to route proven health advocacy in near-time

• Started a dedicated health advocacy team to guarantee effective and

efficient care for member with most complex needs (analytics included from start, w/ senior support)

• Needed to predict which cases are most likely to become complex as IP

requests arrive, can’t delay workflow

• Developed a extreme need predictive model, trained on everything we know

about a member’s care history

• Allied to business rules which respond swiftly to specific pieces of

information, all validated by outcomes measurement

• Leverages sophisticated off-platform model to directly drive the routing in-

system – required cultural change

• Members with complex care needs now receiving dedicated support from a

multi-disciplinary team from day one through to post-acute case management in the community from our integrated care center of excellence, leading to dramatic reductions in readmissions and length of stay

Incoming

authorization

request Extreme need model Trigger rules UM care manager ICT

Old process New process

Embedding predictive analytics in the operational workflow requires strategic alignment, senior support, cultural change, and pragmatism

Scale insight processes with machine learning

7

• Analyzing of the drivers cost variation within similar

episodes produces actionable insight to drive efficiency, but using clinical expertise is resource intensive

• Combine clinical expertise with machine learning

techniques to expose causes of cost variation rapidly and at scale

• Automated feature engineering 000s of metrics for each

episode

• Machine learning algorithms to test every possible

combination to “learn” which ones best separate the episodes into less or more efficient

• The characteristics which led to them being classified

gives questions to ask SMEs and clinicians

• Actionable insights are turned into recommendations to

impact workflow at the appropriate level (medical policy, health system, or specific facility/providers) Finding new, ambitious use cases for predictive analytics can scale up expert processes to impact value, if done sensitively Automatically detecting drivers of cost variation within treatment episodes

No catheter procedure ? Single physician ? Less than 3 caths?

A B X X

More efficient Less efficient Mixed

N Avg O/E Std Dev N Avg O/E Std Dev N Avg O/E Std Dev N Avg O/E Std Dev

Using predictive analytics to understand our own effectiveness

8

• Essential to know value from clinical programs,

studies can be slow, expensive

• Retrospective matched controls method -

quick, cost effective, uses existing data, and practical in real-world settings

• Match “control group” members that did not get

the intervention to the intervention group on a member-by-member basis using predictive algorithms - represents what would have happened without the intervention

• Compare any set of outcome metrics among

the two groups using a simple statistical test

• Informs scale up/down of programs, improves

their design, influences routing

0% 20% 40% 60% 80% Prevalence Potential control Matched control Intervention

Quick, cost effective evaluation of clinical programs Predictive analytics techniques deployed to measure value form clinical programs, builds culture of evidence-based decision making and feeds into what other predictive analytics are used to trigger

Conclusion

Just a few examples of our analytics transformation, lots more we could discuss, and even more work still to do While predictive models are easy to develop in isolation or as one-offs, building them into the workflow to be routinely used to actually impact the way we care for our members requires a different mindset to overcome technical and cultural barriers What's next? Continue to explore new use cases where machine learning can add value, while exploring and developing ever more advanced analytic approaches (such as deep learning) 9

Relentless seeking

• ut of data

sources held by or accessible to the

• rganization

Sustained senior support for analytics being a core part of wider strategy Achieve adoption and value by injecting predictive analytics direct into the workflow, even if that means challenge and trade-offs Taking machine learning into new areas of the business with ambitious use cases Build confidence through human integration, and setting an expectation of evidence-based decision making