Predictive
Predictive
Analytics
Analytics
for hospital management
for hospital management
Hans Levenbach, Delphus, Inc. and Paul Savage, HCI
Hans Levenbach, Delphus, Inc. and Paul Savage, HCI-
-LLC
LLC
Email:
Email: hlevenbach
[email protected]
@delphus.com
ISF 2010
ISF 2010
–
–
San Diego, CA
San Diego, CA
June 21, 2010
Overview
Introduction
Introduction
Predictive Analytics –
Predictive Analytics
–
something new?
something new?
Approaches and methods –
Approaches and methods
–
model complexity vs
model complexity
vs data volume
data volume
Supporting the Hospital Value Chain
Supporting the Hospital Value Chain
Geography and product lines
Geography and product lines
-
-
Patient care activity
Patient care activity
Multiple competitor and product
Multiple competitor and product
-
-
mix forecasting
mix forecasting
Competing for new hospital locations
Competing for new hospital locations
–
–
Simulations
Simulations
Hospital closings
Hospital closings
-
-
Berger Commission type simulations
Berger Commission type simulations
–
–
Data architecture and reporting
Data architecture and reporting
Multi
Multi
-
-
State & Current Perspective
State & Current Perspective
“
We have to bring the
We have to bring the
We have to bring the
We have to bring the
science of management back
science of management back
science of management back
science of management back
into Healthcare
into Healthcare
into Healthcare
into Healthcare
”
Donald Berwick, MD
Predictive Analytics –
Something New?
Predictive analytics encompasses a variety of
Predictive analytics encompasses a variety of
Predictive analytics encompasses a variety of
Predictive analytics encompasses a variety of
techniques from
techniques from
techniques from
techniques from
statistics
statistics
statistics
statistics
, , , ,
data mining
data mining
data mining
data mining
and
and
and
and
game
game
game
game
theory
theory
theory
theory
that analyze current and historical facts
that analyze current and historical facts
that analyze current and historical facts
that analyze current and historical facts
to make predictions about future events.
to make predictions about future events.
to make predictions about future events.
to make predictions about future events.
Predictive Analytics - Methods
CLUSTERING CLUSTERINGCLUSTERING CLUSTERING FORECASTING FORECASTING FORECASTING FORECASTING MONITORING MONITORING MONITORING MONITORING &&&&ADVISING ADVISINGADVISING ADVISING
SIMULATION SIMULATIONSIMULATION SIMULATION &&&& SCENARIO
SCENARIO SCENARIO
SCENARIO PLANNINGPLANNINGPLANNINGPLANNING DECISION
DECISIONDECISION
DECISION TREETREETREETREE
The use of current and past data, in conjunction with statistical, structural or other analytical
models and methods, to determine the likelihood of certain future events
Predictive methods cover a range from relatively simple classification and forecasting to
more advanced techniques such as dynamic modeling and
simulation
As you move down the spectrum, the complexity of the
approaches and their implementation increase, while
Framework for SaaS Planning
SPARCS
Data
Resource
Mining &
Analytics
PEER
Planner
Dashboard
Competitive
Simulation
Model
Multi-year
Multi-hospital
Information
Product Line
Trends
Spatial Analysis
Market Shares
Budget
Forecasting &
Collaborative
Planning
Interactive
Decision Tree:
All Patients
Primary Tumor Site
Psychiatry
Surgery
Maternity
Surgery Colon: Classification by Stage
C1 Oncology Cardiac New Born Cardiology Pulmonary
> 100+ product line/service combinations
> 220 Hospitals > Payor Class > Region
Descend tree using any available
differentiating attributes; natural, derived or inferred; separately or in combination
Intelligent Dashboard Environment
Manage strategic opportunities
Monitor competitive environment
Enhance physician relations
Dashboard
History, Adjustments and
Forecasts
Report Writing and Additional
Report Writing and Additional
Capabilities
Capabilities
Physician Activity Reports
Physician Activity Reports
Market Level Assessment
Market Level Assessment
Simulation Modeling
Simulation Modeling
Product Level Demography
Product Level Demography
Benchmarking: Quality, Safety &
Benchmarking: Quality, Safety &
Performance
Physician Level Reporting
Attending Physician Activity
Surgeons Activity
Measures of Loyalty Ranking
Physician
&
Group Practice
Analysis
Total Group Practice Inpatient Activity
( 8 0 % p re d i c t i o n i n t e r v a l s ) 1 0 0 1 2 0 1 4 0 1 6 0 1 8 0 2 0 0 2 2 0 2 4 0 2 6 0 2 8 0 3 0 0 J a n- 0 4 Ma y - 0 5 Oc t - 0 6 Fe b - 0 8 J u l - 0 9 No v - 1 0Low er P-L (10%)
Upper P-L (90%)
History
History T-C
Forecast T-C
A v g . 2 16 / Mo 19 % S e a s o n a l R a n g ePercent of Group Practice by Hospital
8 0 % p re d ic t io n int e rv a l s 0 % 1 0 % 2 0 % 3 0 % 4 0 % 5 0 % 6 0 % 7 0 % 8 0 % 9 0 % 1 0 0 % J a n- 0 4 Ma y - 0 5 Oc t - 0 6 Fe b- 0 8 J u l - 0 9 No v - 1 0G-S_History
G-S_History T-C
G-S_Forecast T-C
N-I_History
N-I_History T-C
N-I_Forecast T-C
P rim a ry Ho s p ita l S ha re o f P ra c tic e
S e c o n da ry H o s pita l S h a re o f P ra c tic e
17 %
8 3 %
New Physician Growth
( 8 0 % p re d ic t io n i nt e rv a ls ) 0 2 0 4 0 6 0 8 0 1 0 0 1 2 0 J a n- 0 4 Ma y - 0 5 Oc t - 0 6 Fe b - 0 8 J u l - 0 9 No v - 1 0Lower P-C (10%)
Upper P-C (90%)
History
History L-C
Forecast L-C
Transition to Mature In-Patient Activity
( 8 0 % pr e di c t i o n i nt e r v a l s ) 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 J a n - 0 4 M a y - 0 5 O c t - 0 6 F e b - 0 8 J u l - 0 9 N o v - 1 0Forecast T-C
History
History T-C
Senior Partner In-Patient Activity
( 8 0 % p re d ic t io n int e rv a l s ) 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 J a n- 0 4 Ma y - 0 5 Oc t - 0 6 Fe b - 0 8 J ul - 0 9 No v - 1 0Low er P-C (10%)
Upper P-C (90%)
History
History T-C
Forecast T-C
Avg 26/ MoMature Partner (Stable) Activity
(80% prediction intervals)
0 10 2 0 3 0 4 0 5 0 6 0 7 0 8 0 J a n - 0 4 Ma y - 0 5 Oc t - 0 6 F e b - 0 8 J u l - 0 9 N o v - 10Low er P-L (10%)
Upper P-L (90%)
History
History T-C
Forecast T-C
Avg 50/Mo
Hospital
&
Product Line
Analysis
Community Hospital
Forecast Patient Activity
1400
1500
1600
1700
1800
1900
2000
P
a
ti
e
n
ts
/M
o
.
22,000 Pts/Yr
20,000
Pts/Yr
Simulation
Modeling :
Competing For New
Hospital Locations
Extending Visibility Into The Enterprise
Executive User Functional User Power User Highly Aggregated More DetailComplete Raw Data •Graphical Display, Dashboards, Interactive
•Aggregated Data, Model Execution
•Limited Drill Down
•Standard & Ad hoc Reporting
-Parameter-Driven by Users at Run-Time
-Sorting, Selection, Filtering, Drill-Down
•Utilizing Standard Functions and Models
•Direct Access to Detailed Raw Data
-“Just give me the data in Excel”
•Model Development & Deployment
• Traceability • Consistency
SaaS
Implementation: System & Data Architecture
Hospital Data Operations Data Financial Data External Data Data Warehouse Multi-Dimensional Data Store Data Files Data SetsIntegration Cleansing Data Quality Formatting Aggregation
Predictive Models Report Management Test Management Model/Version Management DATA PRESENTATION/ CONSUMPITON LAYER APPLICATION/ MODEL MANAGEMENT LAYER DATA STORAGE LAYER DATA INTEGRATION LAYER DATA SOURCE LAYER