Turning Big Data Into Better Care

Transcription

1 Turning Big Data Into Better Care Dickson Advanced Analytics DA 2

2 Who is CHS and What is DA 2? 2

3 Who is CHS? Hospitals 42 Employees 62K Care Centers 900+ Physicians 3K Licensed Beds 7,800 Nurses 14K 3

4 What is DA 2? A centralized advanced analytics capability with more than a 140+ team dedicated to: Consolidate Disparate Sources of Patient Data Assist in Identifying Actionable Insights Predict Healthcare Needs Create Transformative Solutions Based on CHS Data Evaluate Improvement in Patient Outcomes Partner to Promote Community Health 4

5 DA 2 : Operating Model Strategic Services & Operations ACT Implement and operationalize solutions through changes in business processes Information Services ADVISE Business Operations & Strategy (DA 2 ) Define courses of action, optimize decision making that gives the best outcomes PREDICT Utilize past observations, trends and patterns to predict future observations ACQUIRE Collect data from internal and external sources Prepare and refine raw data DESCRIBE AND DISCOVER Enhance and enrich data Cluster and classify data into meaningful groupings Data Services (DA 2 ) Population Health & Analytic Services (DA 2 ) 5

6 Key Uses of Analytics Care Management Avoidable Care Reduction Cost Analytics Experience of Care Per Capita Cost Ambulatory Quality Strategic Services Triple Aim Population Health Outcomes Research 6

7 Business Impacts of Data Science 7

8 Data: How big is BIG? More than a petabyte of data across the System housed in the Electronic Data Warehouse. This is the size of 20 million 4-drawer filing cabinets filled with text! Over a 3-year period (August 2012 July 2015), this includes millions of: Unique Patients Encounters Billing and Problem List Diagnosis Procedures Pharmacy Orders Payor, PCP, Lab Results, Risk Scores, Demographic and External Data Issue: How does data this size get turned into actionable opportunities? 8

9 The Role of Segmentation and Panorama 9

10 Big Data: Management & Analysis CHS Billing Systems CHS EMR System IDX STAR Cerner EDW Panorama Statistical Algorithm Segments Behavior & Consumer Data Experian Address- Based Geospatial Data ESRI I/S Data Services Analytic Services 10

11 Our Results: Compare with Others High Risk 5% High Risk 5% Advanced Cancer Complex Chronic High Risk Rising-Risk 20% Aging, Rising Risk Mental Health Rising Risk Prescriptive, Occasional, Chronic 95% Low-Risk 75% Pregnancy & Delivery Newborns & Toddlers Sparse Information, Acute & Well Low Risk Sg2 The Advisory Board Company CHS 11

12 Use Cases 12

13 Uses for Segmentation and Panorama Segmentation Higher-level view of how our patients can be grouped clinically Segments defined by clinical risk: billed charges and utilization Helps prioritize and streamline programs from a data perspective Panorama Connect billing and clinical data Provide deeper understanding of disease states Study utilization patterns among patients Ability to quickly drill into patient data to reach actionable populations 13

14 Deeper Look Into A Segment Aging, Rising Risk 57 is the avg. age of this segment, 49% between 45 & 67 Highest Pct. with a PCPattributable visit in last 18 months (70%) Below Avg. number of diseases for their age 59% are married (highest among all segments) 62% is the avg. 10-year survival probability Ordinary number of procedures and prescriptions 12% have been diagnosed with at least one cancer 7% of segment is among the top 5% in 3-year billed charges 48% have a mental/behavioral condition $23K is the avg. 3- year billed charges 45% are Commercial 24% of the total billed amount despite being only 17% of the total population Results shown above are hypothetical and are presented for illustrative purposes only. 14

15 Use Case 1: Care Management # of Patients Complex Chronic Segment Sample Process 100,000 and Living 95,000 and Avoidable Utilization 1 in Last 12 Months 25,000 and Active Primary or Specialty Care 2 24,000 and Residence in Core Market 23,000 and > 4 Body System Conditions/Diseases 20,000 and 8 or more Therapeutic Classes 15,000 and High Spend 10,000 and Designated CM Practices 5, Avoidable Utilization consists of Avoidable ED visits (NYU Algorithm) and Avoidable Inpatient Hospitalization (PQI / PDI as defined by AHRQ). 2. Active Primary or Specialty Care patients having a PCP / Specialty (as defined by PCP attribution logic) in the last 18 months. Results shown above are hypothetical and are presented for illustrative purposes only. 15

16 Use Case 2: Avoidable Care Using Tableau with Panorama data provides the capability of interactive discovery with strategic leaders These views are essential to helping practitioners separate or align perceptions with reality across the System Data-driven approach reduces biases and enables fact-based decisions and proper evaluation of the impact of those decisions 16

17 Use Case 3: Extended Hours PCP Locations Utilizing Panorama data and analyses, showcased an example of how past trends could be used to advise on a strategic initiative The approach aligned data to support consumer preferences for easy access and convenience The analysis located opportunities that might have the biggest impact for our patients and for the system to recapture unmet and deflected demand 17

18 Questions? 18

19 Appendix 19

20 Predictive Models and Tools Propensity to Pay supporting Revenue Cycle efforts Time-to-Event modeling: Type II Diabetes Likelihood of Readmission Targeted Communication to Reduce Avoidable Care Utilization Patterns of Patients with Chronic Diseases Timing of Palliative Care Consultations 20

21 Advanced Analytics: Segmentation Modeling 1. Data Sampling 3. Cluster Analysis 2. Factor Analysis 4. Random Forest 21