Webinar. Reducing Readmissions with BI and Analytics. 23 March 2018 Copyright 2016 AAJ Technologies All rights reserved.

Transcription

1 Webinar Reducing Readmissions with BI and Analytics Copyright Reducing 2016 Readmissions AAJ Technologies with BI and All rights Analytics reserved.

2 Hospital Readmissions Michele Russell, CEO, Russell Consulting Group Copyright Reducing 2016 Readmissions AAJ Technologies with BI and All rights Analytics reserved.

3 Hospital Readmissions Overview Under Hospital Readmissions Reduction Program (HRRP), CMS withholds up to 3 percent of regular reimbursements for hospitals if they have a higher-than-expected number of readmissions within 30 days of discharge for seven conditions: Chronic lung disease Coronary artery bypass graft surgery Heart attacks Acute myocardial infarction Hip and knee replacements Pneumonia Heart failure 3

4 Hospital Readmissions Overview About 80 percent of the 3,241 hospitals evaluated by the Centers for Medicare and Medicaid Services (CMS) this year will face penalties Medicare under the Hospital Readmissions Reduction Program (HRRP) will reduce reimbursement for 2,573 hospitals for fiscal year (FY) 2018 An analysis of the data also showed CMS under HRRP will withhold $564 million in payments over the next year Source: Kaiser Health News 4

5 HIPAA Privacy & Security / Risk Mitigation In addition to creating a culture that focuses on the security and privacy of Protected Health Information (PHI), our technology plays a significant role in preventing data breaches. Tracking and audit trails Physical security of the data Limited user access to data Role-based security Protection of sensitive subsets of PHI Ongoing control of user access regardless of the hosting environment 5

6 Where does / will your data live? The three major types of cloud storage used in enterprise deployments are: Public Private Hybrid Cloud 6

7 A Care Transition System (CTS) to Reduce Hospital Readmissions Ed Kirchmier, VP Global Delivery, AAJ Technologies Copyright Reducing 2016 Readmissions AAJ Technologies with BI and All rights Analytics reserved.

8 Care Transition Problem Pre- Discharge Hospital Visit EMR/ADT (Hospital) Discharge Home ALF SNF Mental Health Discharge Instructions PCP/Specialist Follow-up Rx Scripts Nutrition Guidelines Wound Care PT Orders Insufficient Education and Lack of Coordination Leads to Readmission Patient Practice Mgmt Sys (PCP/Specialists) Rx Dispensing (Pharmacy) Order Management (Home/Community Providers) 8

9 Care Transition Platform EMR/ADT (Hospital) Goal: Reduce Re-admissions Pre-Discharge Care Coordination & Coaching Discharge Post- Discharge Follow-Up HOSPITAL VISIT HOME/FACILITY VISIT VISIT(S)/CALLS Case Visits/Assessment Care Plan Appointments Orders Workflows Reminders Notifications Alerts PCP/Specialist Follow-up Medication Management Nutrition Management Home & Community Services Red Flags / Signs & Symptoms Personal Health Record Practice Mgmt Sys (PCP/Specialists) Rx Dispensing (Pharmacy) Order Management (Home/Community Providers) 9

10 Dashboards to Forecast Healthcare Outcomes Kevin Oppenheimer, Principal/Owner KGO Consulting Group Copyright Reducing 2016 Readmissions AAJ Technologies with BI and All rights Analytics reserved.

11 Dashboard 30 Day CMS Readmissions Report 11

12 Data Analytics to Predict When a Patient Will Readmit Andrew Satz, Co-Founder, Data Scientist and Futurist, Metrix Labs Copyright Reducing 2016 Readmissions AAJ Technologies with BI and All rights Analytics reserved.

13 Causation versus Correlation A patient is hospitalized for pneumonia with a history of chronic COPD. The patient is rehospitalized within thirty days. Only 7.4% of patients readmitted had pneumonia. 92.6% of those readmissions were caused by comorbidities rather than the pneumonia While a prior pneumonia case is highly correlated to readmission, it s not necessarily the cause 13

14 Data Dimensionality Sex Weight Ethnic City BP Smoker COPD Pneumonia Heart Disease GI F 142 H M 130/80 Y M 178 A F 170/90 N 5 1 F 203 C M 130/90 N M 187 A P 170/90 Y 3 2 F 162 A P 170/90 Y F 120 H M 80/50 N 4 2 M 263 A F 80/50 Y M 207 C P 130/80 N

15 Data Dimensionality 15

16 Reducing Data Dimensionality: Feature Importance If we want to identify these additional factors for readmission, we need to evaluate feature importance. This lets us find causal and correlated features associated with the outcome. Then we can build an algorithm off of those features, which can be used to classify patients and predict readmission As a result, we can identify those patients most likely to be readmitted and why. This would enable caregivers to deliver focused interventions to vulnerable patients 16

17 Using A.I. to Avoid Readmissions The effort: A health system aggregated and integrated their clinical, financial, administrative, patient experience, and other relevant data Important features were algorithmically selected based upon their statistical importance. Features included: Medicare severity diagnosis related group (MS-DRG), use of tobacco, residential zip code, financial risk, healthcare utilization, age, marital status, admission source, medication, provider, and more More than two dozen machine learning models were built to predict readmission 17

18 Using A.I. to Avoid Readmissions The Results: The team s hypothesis was that patients on a larger number of medications would be at the greatest risk for readmission But the data revealed that patients with no medications were at the greatest risk The data showed similar patterns regarding age. While the team assumed older patients are at the highest risk, the data showed that its younger patients were at greatest risk 18

19 Questions & Answers 19

20 Additional Information: Jim McKeen Director of Healthcare Solutions