Python What? The Strategist as Data Geek
-
Upload
stratasan -
Category
Healthcare
-
view
786 -
download
4
Transcript of Python What? The Strategist as Data Geek
The opinions expressed are those of the presenter and do not necessarily state or reflect the views of SHSMD or the AHA. © 2015 Society for Healthcare Strategy & Market Development
Python What? The Strategist as Data GeekOctober 14, 2015
Speakers• Patrick Saale– Manager Strategic Resource
Group, LifePoint Health
• Lee Ann Lambdin – Vice President Strategic Resources, Stratasan
2 Source: Stratasan & LifePoint Health, 2015
Outline• Role of the Analyst• Skills of the Analyst• Tools of the Analyst
– Python What?– Resources
• Data Information Better Decisions• Case Study – LifePoint Hospitals, Physician Referral
3 Source: Stratasan & LifePoint Health, 2015
Survey Results
• Surveyed Stratasan customers for their perspective on the role of the analyst
• 21 responses are summarized in the following slides
• 14 responders were analysts and 7 were users of analysts
• Employed primarily by health systems and hospitals
4 Source: Stratasan & LifePoint Health, 2015
Analyst Role
Turn data into information so
leadership can make better decisions
Use data and analytical tools at the team’s disposal to support a data‐driven
approach to organic growth
Acquire data and turn it into useful information
for customers
Summarize and provide meaningful insight into what data is dictating
Assist in analyzing data for purposes of strategic planning and business development support
Provide market share, competitor, and referral
info
5 Source: Stratasan & LifePoint Health, 2015
Skills of the AnalystAbility to be creative and tell a story with
data
Analytical ThinkingData Interpretation
Summarizing vast amount of information
Computer/Arithmetic SkillsExcel #1
PowerPointMappingAccess
SPSS/Statistics
Critical Thinking/Strategic
ThinkingProblem Solving
Inquisitive
Attention to DetailAccuracy
Ability to Communicate/Present
Time Management
6 Source: Stratasan & LifePoint Health, 2015
Best Verbatim Comments on Skills
7
“Ability to determine what your customer really needs instead of always just doing exactly what they ask you to do”
“Support and influence others in appropriate use of data”
“Master necessary programs to analyze data to tell the story”
“Analytics tool knowledge (Excel, Access, SPSS, etc.) doesn’t really matter which one as long as you know it”
“Ability to understand data trends and use it to tell a story”
“Knowledge of the field’s terminology and data available including sources of data”
Source: Stratasan & LifePoint Health, 2015
Tools of the Analyst: Most Important Data SourcesInternal hospital or system data(financial &
volume)/company results, E.H.R.
State IP, OP, ED, Observation
databases (where available)
Demographics
Mapping software Federal Data (Medicare)
Industry and competitor research
Google searches
Psychographics (Tapestry
Segmentation)
8 Source: Stratasan & LifePoint Health, 2015
Top Questions & Project Requests• What’s my market share?
– Reasons for growth/decline?
• What’s our outmigration?• What are my competitors doing?• What are my opportunities for growth?
– Are there needs in the area not currently being met?
• What is the profitability of service lines?• How many cases are coming from ____?• How many doctors do I need?• Where do the doctors need to be located?• Operational performance?
9 Source: Stratasan & LifePoint Health, 2015
Best Verbatim Comments: Questions
10
“Can we change this?Can we have an update?Can we get it before the deadline?”
“Market share reports to determine current volume, potential added volume, capacity and service needs”
“Market and finance data reports for specific service lines”“Do we need more or less
physicians and where do they need to be located?”
“‐ Create a map with data‐ Summarize the data‐ Trend the data”
Source: Stratasan & LifePoint Health, 2015
Anything else we need to know about Analysts?
11
“They need to always be focused on helping planning and marketing generate ROI. Because they are most often the most analytical thinkers of the group, they need to lead the charge in measuring and planning how we can prove the value of what marketing and planning brings to the table.”
“Need to be creative and think outside the box. Good communication skills and ability to ask questions about what is trying to be accomplished that will influence data support and analysis.”
“They are really smart!”
“We're awesome ;)”
“good analysts want to spend more time thinking about how to help solve problems by drawing conclusions from data, and less time on mundane task work.”
Source: Stratasan & LifePoint Health, 2015
Hiring: What to Look for in an Analyst
• Critical thinking skills• Holistic decision‐making• Use of data to inform decision‐making• Knowledge of how to leverage people who know Python and big data
• Understanding that no one person can do it all• Specific skills for specific roles
12 Source: Stratasan & LifePoint Health, 2015
Best Use of Analysts’ Skills
13
14
DIKW Pyramid: The ConceptDIKW Pyramid: The Concept
WisdomWisdom
KnowledgeKnowledge
InformationInformation
DataData
Source: Stratasan & LifePoint Health, 2015
Executives
Drivers
Implementers
15 Source: Stratasan & LifePoint Health, 2015
Executives
Drivers
Implementers
What to do?
How to do it?
Delivery
16 Source: Stratasan & LifePoint Health, 2015
ExecutivesWhat to do?
DriversHow to do it?
ImplementersDelivery
17 Source: Stratasan & LifePoint Health, 2015
ExecutivesWhat to do?
DriversHow to do it?
ImplementersDelivery
18 Source: Stratasan & LifePoint Health, 2015
DriversHow to do it?
ExecutivesWhat to do?
ImplementersDelivery
Translation
Translation
Source: Stratasan & LifePoint Health, 201519
Bridging Worlds: Generate Data-Driven Insight
Attributes, Skills and Tools
? ?
?
!!
?
?
!
20 Source: Bridging Worlds, SHSMD, 2014 p.57; used with permission
!!
!
!
!
!
What’s important for you?
21 Source: Bridging Worlds, SHSMD, 2014 p.57; used with permission
22
The Analyst
Source: Stratasan & LifePoint Health, 2015
How Big is Big?
23
Big Data
Medium Data
Small Data
A lot more problems with medium and small data, and opportunities in the data you deal with every day
Source: Stratasan & LifePoint Health, 2015
Look at the Data
• What is in this data?
• What question am I trying to (can I) answer with this data?
• How do I leverage the data to answer the question?
24 Source: Stratasan & LifePoint Health, 2015
Glossary: Analyst as Data Geek
• Handout –Definitions–Uses
25 Source: Stratasan & LifePoint Health, 2015
Look at the Data• R
– R Studio is a free software environment for statistical computing and graphics. It compiles and runs on a wide variety of UNIX platforms (foundation operating systems are built on), Windows and MacOS.
• SPSS – IBM SPSS Statistics is an integrated family of products that addresses the entire
analytical process, from planning to data collection to analysis, reporting and deployment. Used for describing large data sets, for example 3 years of patient data.
• SAS– Another brand of statistical software
• Python– is a programming language that has powerful libraries for data analysis.
It also allows you to automate steps of processing or analyzing data.
26 Source: Stratasan & LifePoint Health, 2015
Actual Python code: script that is loading ICD10 codes into a database from CSV files so we can run queries and joins
27 Source: Stratasan & LifePoint Health, 2015
Process the Data
• How do I make the data useful? • What are we going to do to it?
– Rollups, aggregation, curation, cross‐walking – Machine learning (fancy statistics)
• Where are we going to do it?– Your laptop– Cloud computing– Hadoop
28 Source: Stratasan & LifePoint Health, 2015
Nobody Understands the Cloud
29 Source: Stratasan & LifePoint Health, 2015
• Cloud computing– Cloud computing allows you to use computers you don’t own to operate
programs you use. Gmail, anything from Google. You can purchase access to warehouses of computers via cloud providers like Amazon, Google, and Microsoft. This allows you to run tools like Hadoop to process large amounts of data.
Process the Data: Cloud
30 Source: Stratasan & LifePoint Health, 2015
Intel has launched Collaborative Cancer Cloud, a new service to enable providers and researchers to securely share genomic, imaging and clinical data among participating organizations across the globe.
By 2020, the goal is to have physicians be able to give a patient a diagnosis and generate a specific treatment plan within 24 hours. Over time, the platform will be modified to support other types of research and treatment.
Process the Data: Machine Learning & Predictive Analytics
31 Source: Stratasan & LifePoint Health, 2015
“For the past 10 years, we have been working on that area,” Ebadollahi said. “We have very advanced machine learning, pattern recognition, on imaging and video in general, most especially in medical imaging. Now, this intent to acquire Merge will bring a conduit to attach those technologies coming out of our research.”
Analyzing & Presenting the Data
• How to make the data tell a story?– Excel–PowerPoint–GIS– Tableau– JavaScript–D3
32 Source: Stratasan & LifePoint Health, 2015
Analyzing & Presenting: Excel
• Pivot Tables• Macros• Cell Links• V‐Lookup or Index Match• Format Painting• Custom Sorts
33 Source: Stratasan & LifePoint Health, 2015
Analyzing & Presenting: PowerPoint
• Custom color palate and template with logo• Graphs, graphs, graphs• Add maps and photos• Tell a story
34 Source: Stratasan & LifePoint Health, 2015
Analyzing & Presenting: Tableau
• Business analytics software• Business dashboards• Big data analysis• Data discovery• Social media analytics
“We are looking to move our market share reporting to Tableau within the year, as the level of detail we’re being asked to report on has grown beyond Excel’s capacities.… It’ll increase automation and decrease errors on our part.”‐Stratasan customer
35 Source: Stratasan & LifePoint Health, 2015
Analyzing & Presenting: JavaScript
• JavaScript ‐ This programming language is all about presentation layer (charts, graphics, and user interaction). It is the glue that holds Internet together. Every modern browser runs Javascript.
• D3 ‐ D3.js is a powerful JavaScript library for producing dynamic, interactive data visualizations in web browsers.
36 Source: Stratasan & LifePoint Health, 2015
Analyzing & Presenting: GIS
• GIS – A Geographic Information System enables you to envision the geographic aspects of a body of data. This lets us visualize, question, analyze, and interpret data to understand relationships, patterns, and trends. (Esri) Used primarily in government, conservation, zoning and construction. – Esri ArcGIS
• Very granular demographic data – example patient origin by block group, demographics by block group
37 Source: Stratasan & LifePoint Health, 2015
38
C A R D I O L O G Y P RO J E C T E D B L O C K G RO U P V O L U M E
C A R D I O L O G YB L O C K G RO U P S C O R E C A R D
Block Groups outlined in green are considered the best targetsBlock Groups grayed out do not have the desired tapestries
Source: Stratasan & LifePoint Health, 2015
Brentwood Emergency Patient Origin by Block Group
39
B R E N T W O O D M E D I C A L C E N T E RE M E R G E N C Y PAT I E N T O R I G I N B Y B L O C K G RO U P
Source: Stratasan & LifePoint Health, 2015
Analyzing & Presenting: Tapestry Segmentation
• ESRI Tapestry data – Tapestry segmentation provides an accurate, detailed description of America's neighborhoods—U.S. residential areas are divided into 67 distinctive segments based on their socioeconomic and demographic composition—then further classifies the segments into LifeMode and Urbanization Groups. Tapestry Segmentation is used to target your population with specific messages that are meaningful to the specific population.
40 Source: Stratasan & LifePoint Health, 2015
Northern Block Groups are zoomed in the next map41
D OM I N A N T TA P E S T R Y S E GM E N TAT I O NB LO C K G ROU P
• The Tapestry Segmentation and LifeMode (Psychographic Profile) for each Block Group is represented by a Number & Letter combination
• This ID helps guide your marketing execution plan
D O M I N A N T TA P E S T RY S E G M E N TAT I O NB L O C K G R O U P
Source: Stratasan & LifePoint Health, 2015
42 Source: Stratasan & LifePoint Health, 2015; esri.com
43 Source: Stratasan & LifePoint Health, 2015; esri.com
44 Source: Stratasan & LifePoint Health, 2015; esri.com
45 Source: Stratasan & LifePoint Health, 2015; esri.com
Resources to Learn More• Coursera
– https://www.coursera.org/
• Python website– https://www.python.org/about/
• YouTube– https://www.youtube.com/watch?v=puS8Tu3JPnU– https://www.youtube.com/watch?v=GZpwGt0hzKs
•
46 Source: Stratasan & LifePoint Health, 2015
Case Study: Physician Referrals
47
Case Study: Physician Referrals• Situation: A data set that had not been
previously utilized by the organization was introduced
• Outcome: The organization reacted to optimize the use of the information through an entirely new set of processes and a change in organizational structure
• Next comes the “How”…
48 Source: Stratasan & LifePoint Health, 2015
Data Science• Step 1 ‐ Look at the data:
– Source NPI Numbers– Destination NPI Numbers– Shared Patients
• Step 2 – Process the Data (Connecting the Dots):– Data will give information about the relationships
between physicians.– Enough organizational savvy to know who could use the
information (Physician Sales Team) – engage them on the discovery phase.
– Identify other sources of information that will help to add context
49 Source: Stratasan & LifePoint Health, 2015
Data Science
• Step 3 ‐ Present the Data:– Nicely formatted Excel table– Map– Infographic
50 Source: Stratasan & LifePoint Health, 2015
Medicare Physician Referral Databases
• https://questions.cms.gov/faq.php?faqId=7977• Files are 1‐7 gigabytes• 30 day referrals 2.4 gigabytes• Smallish data; too big for Excel
51 Source: Stratasan & LifePoint Health, 2015
Physician Network Intelligence
Provider 1
The patient volume when two providers bill Medicare for the same patient within
30 days of each other.Provider 2
52 Source: Stratasan & LifePoint Health, 2015; Medicare Referral Database (2014)
TOTA L V I S I T S B Y Z I P C O D E G R E AT E R H E A LT H S Y S T E M M A R K E T S H A R E
Source(s): Stratasan (2014); Esri (2014); Medicare Referral Database (2014)53
2014 Medicare Physician to Physician Network Summary -Primary Care Doctors to Any Orthopods
James Kessler William MillerLawrence
SupikMartin Senicki
Ryan Slechta
William Handley
James Kessler (Angel)
William Miller (Haywood)
Employed Total 27% 12% 15% 34% 88% 7% 5%Anthony Esterwood 0% 0% 23% 78% 100% 0% 0%
Beth Bailey 46% 25% 0% 30% 100% 0% 0%
Elizabeth Dixon 44% 36% 20% 0% 100% 0% 0%
Ewa Susfal 0% 0% 17% 60% 77% 23% 0%
Lee Ann Manthorne 58% 42% 0% 0% 100% 0% 0%
Randall Provost 60% 25% 15% 0% 100% 0% 0%
Steven Queen 58% 18% 0% 0% 76% 0% 24%
Todd Davis 16% 0% 18% 45% 79% 9% 11%
Private Total 27% 42% 17% 4% 90% 0% 10%Matthew Mahar 29% 35% 8% 0% 73% 0% 27%
Ofelia Balta 45% 24% 21% 10% 100% 0% 0%
Roy Gallinger 34% 27% 16% 0% 77% 0% 23%
Thomas Wolf 27% 44% 19% 11% 100% 0% 0%
Grand Total 27% 23% 15% 24% 89% 4% 7%
PCP Status/Name
Employed OrthopodsEmployed
Total
Sources: 1. CMS Physician Referral Patterns 2013 - 2014 30 day interval: https://questions.cms.gov/faq.php?faqId=79772. NPI Monthly File; http://nppes.viva-it.com/NPI_Files.html3. Physician Compare Downloadable Database; https://data.medicare.gov/data/physician-compare
Targeted Sales Visit initiated with Dr. Susfal to understand reason for leakage
54 Source: Stratasan & LifePoint Health, 2015
Results
• LifePoint has reduced referrals to Dr. Kessler from Dr. Susfal 25%.
55 Source: Stratasan & LifePoint Health, 2015
Conclusions
56 Source: Stratasan & LifePoint Health, 2015
Conclusions• Use data to tell a story and make better decisions• Learn new things all the time• Search for answers• Focus on the important; don’t major in the minors
• Acknowledge you can’t know everything, but recognize what you don’t know
57 Source: Stratasan & LifePoint Health, 2015
The opinions expressed are those of the presenter and do not necessarily state or reflect the views of SHSMD or the AHA. © 2015 Society for Healthcare Strategy & Market Development