Join Eric Just, Senior Vice President of Product Development, as he will discuss: How machine learning is now included into our analytics platform and being built into all our applications. The toolsets we have developed to automate and democratize machine learning tasks both within Health Catalyst clients and to the broader healthcare industry. Processes to gain clinician buy-in, and engage the best machine learning engine in the world. Demonstrations and examples of this life-saving technology. Dorian DiNardo, Vice President, will share how the Health Catalyst® MACRA Measures & Insights product can help you: Integrate hundreds of measures across financial, regulatory, and quality departments. Monitor the behavior, activities, and other changing information needed to influence, manage, or change outcomes. Tactically and strategically identify measures to take on risk in multi-year value-based care contracts.

1. Introducing Two New Products From Health Catalyst
· catalyst.ai
· MACRA Measures & Insights
Thursday, February 16
1-2:30 PM EST
Eric Just, Senior Vice President of Product
Development
Dorian Dinardo, Vice President of Product
Development

2. © 2016 Health Catalyst
Proprietary and Confidential
Agenda
1-1:35 – catalyst.ai
1:35-2:05 – MACRA Measures & Insights
2:05 – Q&A
2

3. © 2016 Health Catalyst
Proprietary and Confidential
A.I.: Artificial Intelligence
3
General Artificial Intelligence
“Narrow” Artificial Intelligence

4. © 2016 Health Catalyst
Proprietary and Confidential
Machine Learning
4
Machine learning is a large reason for the recent
progress in artificial intelligence.
Machine learning explores the study and
construction of algorithms that can learn from
and make predictions on data.
https://en.wikipedia.org/wiki/Machine_learning
Predictive analytics, or making predictions
based on past data, is one of the artificial
intelligence tasks that machine learning can
solve.

5. © 2016 Health Catalyst
Proprietary and Confidential
We believe machine
learning can accelerate
outcomes improvement
and save lives
Why is Machine Learning Important to Us?
5

6. © 2016 Health Catalyst
Proprietary and Confidential
Predictive Analytics in Healthcare
6
• Mortality prediction
The Charlson Index was introduced in
1987 in the Journal of Chronic Disease as
mortality risk score.
• Readmission prediction
The LACE Index was introduced in the
Canadian Medical Association Journal in
2010 to predict early death or unplanned
readmission after discharge.
“Classic” Approaches

7. © 2016 Health Catalyst
Proprietary and Confidential
Shortcomings…
7
Using the LACE index
to predict hospital
readmissions in
congestive heart failure
patients
By Wang et. al, BMC Cardiovascular
Disorders , 2014
Predicting
readmissions: poor
performance of the
LACE index in an older
UK population
By Cotter et al., Age Aging , 2012
CONCLUSION: The LACE Index may not accurately predict unplanned
readmissions within 30 days from hospital discharge in CHF patients. The
LACE high risk index may have utility as a screening tool to predict high risk
ED revisits after hospital discharge.
CONCLUSION: The LACE Index is a poor tool for
predicting 30-day readmission in older UK inpatients.
the absence of a simple predictive model may limit
the benefit of readmission avoidance strategies.

8. © 2016 Health Catalyst
Proprietary and Confidential 8
Machine learning is easy
(or at least easier!)
The problem is…
Organizations are struggling with making machine
learning routine, pervasive, and actionable

9. © 2016 Health Catalyst
Proprietary and Confidential
Pervasive Use of Machine Learning
9
Health Catalyst Data Operating System
Accelerate insight
Install technology
and ignite change
ClinicalAnalytics&
DecisionSupport
OperationsandPerformance
Management
PatientRelationship&CM
FinancialImprovementand
AccountableCare
Machine Learning

10. © 2016 Health Catalyst
Proprietary and Confidential
Health Catalyst’s Two-Part Machine Learning Strategy
10
catalyst.ai
Machine learning
models in Health
Catalyst applications to
drive outcomes
healthcare.ai
Education and open
source software
initiative to accelerate
machine learning in
healthcare nationally

11. © 2016 Health Catalyst
Proprietary and Confidential
What is the World’s Best
Predictive Engine in
Healthcare?

12. © 2016 Health Catalyst
Proprietary and Confidential
12
Why Do We Need
Machine Learning
Models?

13. © 2016 Health Catalyst
Proprietary and Confidential
Discussing Predictive
Models With Clinicians
Clinicians will adopt predictive
analytics… insofar as they
understand it
catalyst.ai includes performance
reports for every model we bring
13
Powered by
catalyst.ai

14. © 2016 Health Catalyst
Proprietary and Confidential
COPD Readmissions
Powered by
catalyst.ai

15. © 2016 Health Catalyst
Proprietary and Confidential
Practice Management Explorer
Powered by
catalyst.ai

16. © 2016 Health Catalyst
Proprietary and Confidential
16
Propensity to Pay
 Magnitude of Problem
 $50+ BILLION lost annually to bad debt
 What We Predicted
 Likelihood of making a payment on an outstanding debt: 85% positive predictive rate
 Important Variables
 Leveraging ACTUAL patient payment history + demographic data (most just use credit history and demographic)
 Highest impact levers are: payment history (payments made and # of times sent to collections), age (older more likely to pay), and
balance size (almost nobody will pay a $6,000 bill).
 Expected Interventions
 Outreach for people that are likely to pay but are close to collections. Do we have the right address and does patient know they have a
bill?
 Quickly give charity care when needed: For individuals that have a low likelihood of paying, a high balance and have been on Medicaid
or charity care in the past
 Expected Results
 Improved collection rates and lower cost to collect for health systems
 Less patients being sent to bad debt inadvertently = better patient satisfaction
 Seamless path for patients that need charity care
Powered by
catalyst.ai

17. © 2016 Health Catalyst
Proprietary and Confidential
Roadmap: Preventing Chronic Disease
1717
Managing
chronically
ill patients
Prevent need for
inpatient care
Monitoring and
managing
patients at high
risk for
developing
disease
Prevent progression
to disease state
Managing
inpatient
populations

18. © 2016 Health Catalyst
Proprietary and Confidential
c
18
Roadmap: Fully Optimized Closed Loop Architecture
Data Acquisition closer to real-time – leveraging API’s (Smart on FHIR)
Data Science algorithms leveraging machine learning
and real-time data pushes data back to the workflow
engine for integrated display
Workflow engine context passed to
analytics engine via API’s to
customize analytics to workflow.
NLP and closer to real-time data acquisition allows precise customization of cohort by context.
Customizable widgets present end-users
with most important analytics relevant to
this patient at this time.
‘Check list’ Surveillance.
Display of relevant information not part of
the workflow engine can be
accommodated.
Allows immediate action.
Closer to real-time data acquisition and
context passing allows analytics to be
hosted directly in workflow and allow
immediate action.

19. © 2016 Health Catalyst
Proprietary and Confidential
Health Catalyst’s Two-Part Machine Learning Strategy
19
catalyst.ai
Machine learning
models in Health
Catalyst applications to
drive outcomes
healthcare.ai
Education and open
source software
initiative to accelerate
machine learning in
healthcare nationally

20. © 2016 Health Catalyst
Proprietary and Confidential
We believe:
• Machine learning can greatly increase the pace of improved
outcomes in healthcare nationally
• The rate of adoption of this technology is too slow. Barriers cited
include not having the right technology or people.
• We can increase the adoption through
• Education
• Collaboration
• Better tools
Why healthcare.ai?
20

21. Machine Education
and Community

22. 22
Thursday, February 23, 2017 – 3 PM EST

24. Machine Learning
Software

25. © 2016 Health Catalyst
Proprietary and Confidential25
Algorithm 1 Algorithm 2 Algorithm 3
Model &
Accuracy
Report
Model &
Accuracy
Report
Model &
Accuracy
Report
Model &
Accuracy
Report
Model &
Accuracy
Report
Model &
Accuracy
Report
Model &
Accuracy
Report
Model &
Accuracy
Report
Features (i.e. age, comorbidities, polypharmacy)
Result:
• Handful of best (most
predictive) features
• Best algorithm that
computes the relationships
between input features to
generate prediction
• Performance report
summarizing best ‘model’
Algorithms (i.e. Lasso, Random Forest, k-means)
Definition: Simply put, a feature is an input to a machine learning model
Definition: Algorithms are complex mathematical processes that
discover the relationship between features (input) and the
outcome being predicted.

26. © 2016 Health Catalyst
Proprietary and Confidential
Typical ‘Current State’ for Predictive Analytics
26
Data
Source
Predictive
Model ?
Gnarly
SQL Query
Data
Manipulation
Tools/
Algorithms
SAS | Weka |
R | Python
Deploy
Even organizations that have good data
scientists often struggle to operationalize
machine learning.

27. © 2016 Health Catalyst
Proprietary and Confidential
healthcare.ai Open Source Software
27
Our open-source
machine learning
software product
Automates key tasks
in developing
models, or
customizing existing
models using local
data
Makes deployment
in an analytics
environment easy
and ‘production
quality’

28. © 2016 Health Catalyst
Proprietary and Confidential
Scaling People
Data Architects
 Great domain knowledge
 Often looking for opportunities to advance
career/skills
With the right tools…
 Data architects make great feature engineers
 Data architects can easily get started in predictive
analytics.
28
With healthcare.ai, you have the people to do data
science right now.

31. © 2016 Health Catalyst
Proprietary and Confidential
Health Catalyst Data Science Core Team
Levi Thatcher, PhD, Director
Mike Mastanduno, PhD, Data Scientist
Taylor Miller, PharmD, Data Scientist
Taylor Larsen, MA, Data Science Engineer
ChangSu Lee, PhD, Data Scientist (Adjunct)
Scores of additional data scientists throughout the organization!

32. © 2016 Health Catalyst
Proprietary and Confidential
Key Take Aways
32
Catalyst is building
machine learning models
into every Health Catalyst
application to drive
outcomes.
This is catalyst.ai.
We know technology is not
enough to improve
outcomes. We understand
the human factor – the
context in which the
machine learning insight
needs to be delivered, and
the right time and modality
to deliver that insight.
This is Health Catalyst
Catalyst is stimulating the
adoption of machine
learning in healthcare
nationally by creating an
open source repository
for machine learning tools
and expertise.
This is healthcare.ai.

33. MACRA Measures & Insights
February 15, 2017
Dorian DiNardo
Vice President - Operations and
Performance Management Product
Development

34. © 2016 Health Catalyst
Proprietary and Confidential
Payer and Regulatory
Measures
MACRA Measures & Insights

35. © 2016 Health Catalyst
Proprietary and Confidential35
Measures:
Where is the pressure coming from?
Competition
Government Regulations
DSRIP
MACRA Meaningful Use
NHSN
Registries
Risk Contracts
Payers
Joint Commission
Marketing
…the list goes on
Reduced Payments Surveys
Increasing Costs
Increasing Audits
Workload Management
ImproveQuality
ImproveCare
Improve Safety

36. © 2016 Health Catalyst
Proprietary and Confidential
Reporting Burden
• Physicians and their
staff spend between 6
and 12 hours per week
processing and
reporting quality
metrics to the
government1
• $15.4 billion spent
annually1
• Burden expected to
significantly increase2
36
1) Casalino et al. Health Aff March 2016 vol. 35 no. 3 401-406
2) Health Catalyst/Peer 60 Survey

37. © 2016 Health Catalyst
Proprietary and Confidential37
Problem: Aligning Financial and Clinical
• Measures are only increasing in number, data sources,
financial implications
• With this product: Capture the various quality incentives and
financial terms across payers.
• Many challenges to managing measures across departments
• With this product: Identify areas of overlap and determine relative
financial importance to inform initiative selection.
• We provide a ‘quick and dirty’ assessment of what to go at risk for
under the new CMS legislation
– Remains in Qlik/Excel for small and mid-size clients

38. © 2016 Health Catalyst
Proprietary and Confidential38
• For the MIPS track, payment
adjustments begin in 2019 and range
from a -4% penalty to a 12% bonus
• This range grows within first few years
• An estimated 712,000 clinicians will be
impacted in the 2017 performance
year
• CMS calculates 83-90% of eligible
clinicians will be part of the
MIPS track
Implications of MACRA

39. © 2016 Health Catalyst
Proprietary and Confidential39
• 2017 performance will dictate the first payment adjustments
• Yet, only 35% of respondents to a recent Health Catalyst survey
said “we have a strategy and are well on our way to being ready”
• A key decision health systems need to make is which measures to
go at risk on under the Quality Payment Program
• Performance on these measures will be worth 50% of the total score in the
initial year
Are we ready???

40. © 2016 Health Catalyst
Proprietary and Confidential40
What is MACRA Measures & Insights?
MACRA Measures & Insights is the product that will
pinpoint the measures you should take risk on.
1. Help you clearly identify the measures you should go at risk on.
2. Integrate and align your organization on measures.
3. Measure Surveillance.
Surveillance is the monitoring of the behavior, activities, or other changing information,
usually of people for the purpose of influencing, managing, directing, or protecting them.

41. © 2016 Health Catalyst
Proprietary and Confidential41
MACRA Measures & Insights Demo

42. © 2016 Health Catalyst
Proprietary and Confidential42
Features coming soon:
Quadrants

43. © 2016 Health Catalyst
Proprietary and Confidential43
MACRA Measures & Insights Timeline and Availability
Jan ‘17
Feb ‘17 Mar ‘17 Apr ‘17
May ‘17
Jun ‘17 Jul ‘17 Aug ‘17
Sep ‘17
Oct ‘17 Nov ‘17 Dec ‘17
Beta Client Focus
General Availability including
Excel friendly model
Integration of
other measures
into Framework
Integrate into web
applications such as MBL
and API’s into EMR’s
MACRA
enhancements
for additional
quadrants, rule
changes, etc.

44. © 2016 Health Catalyst
Proprietary and Confidential
Q&A
44