Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
Rapid Response to Hospital
Operations using Data and AI during
COVID-19
Rohan D’Souza
Head of Product, KenSci
Tony Pastori...
Agenda
Leveraging Data & AI investments
How IU health leveraged the cloud and data to build self-
service offerings for CO...
Indiana University Health is
a non-profit Health System,
and the largest healthcare
provider in the state of
Indiana.
PATI...
Our Journey with Data & AI
ZettabytesGigabytes
2000’s 2020’s
Azure Data Platform
Social
Graph IoT
Image
LOB
CRM
ERP
SQL Se...
Our Investments in Data & AI
Self-Service analytics
for COVID-19
Advanced analytics & ML
for improving hospital ops
SQL
Mo...
Leveraging Data & AI Investments for
COVID-19 Response
Our Ecosystem & Process
Azure Data
Lake Storage
STORE
Power BI
VISUALIZE
Azure SQL
Data Warehouse
MODEL & SERVE
Informatic...
Data Layers
REPORTING LAYER
Team Workspaces
Intra-departmental, Securable Reporting Layer
Supporting 27 different IUH Anal...
Empowering Self-service analytics during COVID-19
• Over 250 BI Analysts and Citizen
Data Scientists across the system
• S...
Improving Patient Flow with AI/ML
Understanding Healthcare as a complex system
Interdependent
Cascading effect of bottlenecks
leading to downstream patient
...
DischargeAdmi
t Inpatients
Length of Stay
Risk of Readmission
Risk of Obs. Failure
Discharge Disposition
Inpatient Predict...
Length of Stay
Risk of Readmission
Risk of Obs. Failure
Discharge Disposition
Inpatient Predictions
Enable clinicians to p...
© 2019 KENSCI CONFIDENTIAL
Solution
generating scores
in minutes from
data delivery
85% of encounters predicted
within 1.6...
KenSci Realtime Command Center
Bringing Together Data & AI for COVID-19 Response
Risk Stratification Realtime Hospital CensusSurge Planning
IDENTIFYING P...
Unified Data Analytics Architecture with Azure Databricks and KenS
Data Ingestion Agent
Runtime ML engine
Data Preparation...
Learnings & Takeaways
For other health
systems
For other Data &
AI Teams
Feedback
Your feedback is important to us.
Don’t forget to rate and
review the sessions.
Rapid Response to Hospital Operations using Data and AI during COVID-19
Rapid Response to Hospital Operations using Data and AI during COVID-19
You’ve finished this document.
Download and read it offline.
Upcoming SlideShare
What to Upload to SlideShare
Next
Upcoming SlideShare
What to Upload to SlideShare
Next
Download to read offline and view in fullscreen.

1

Share

Rapid Response to Hospital Operations using Data and AI during COVID-19

Download to read offline

Healthcare leaders today are faced with increasingly complex and unprecedented challenges. With COVID-19 taking the world by storm, the need for an intelligent system of insights that can proactively deliver actionable and real-time knowledge on patient populations is imminent to providing better care.

Rapid Response to Hospital Operations using Data and AI during COVID-19

  1. 1. Rapid Response to Hospital Operations using Data and AI during COVID-19 Rohan D’Souza Head of Product, KenSci Tony Pastorino VP, IS, Indiana University Health
  2. 2. Agenda Leveraging Data & AI investments How IU health leveraged the cloud and data to build self- service offerings for COVID-19 response Improving Patient Flow with ML Operationalizing Machine Learning at IU Health to improve hospital operations KenSci Realtime Command Center Learning and experience with deployment of the Realtime COVID-19 Command Center at multiple large health systems
  3. 3. Indiana University Health is a non-profit Health System, and the largest healthcare provider in the state of Indiana. PATIENT ADMISSIONS AVAILABLE BEDS TEAM MEMBERS INDIANA RESIDENTS SERVED ~120,000 ~3,000 30,000+ 1M+
  4. 4. Our Journey with Data & AI ZettabytesGigabytes 2000’s 2020’s Azure Data Platform Social Graph IoT Image LOB CRM ERP SQL Server LOB ERM ERP 2010’s Analytics Platform System APS Terabytes Petabytes
  5. 5. Our Investments in Data & AI Self-Service analytics for COVID-19 Advanced analytics & ML for improving hospital ops SQL Modern data warehousing “We require self-service dashboards to uncover insights and take action” “We’re trying to predict when our patients churn LOS ” “We want to integrate all our data—including Big Data— with our data warehouse” 1 / 2/ 3/
  6. 6. Leveraging Data & AI Investments for COVID-19 Response
  7. 7. Our Ecosystem & Process Azure Data Lake Storage STORE Power BI VISUALIZE Azure SQL Data Warehouse MODEL & SERVE Informatica Cloud Azure Databricks INGEST & PREP
  8. 8. Data Layers REPORTING LAYER Team Workspaces Intra-departmental, Securable Reporting Layer Supporting 27 different IUH Analytics Teams INTEGRATED LAYER Harmonized Layer Rebuilt every 24 hours 3+ Years currently SOURCEMART LAYER 31 Sources 1403 Tables Loaded every 24 hours 10+ Years of Raw Data 2 Types Secured By Datatype Secured by Business Owner CUBES Built from Integrated Layer Rebuilt every 24 hours 3+ Years currently ML ANALYTICS * Built from Integrated Layer Deidentified
  9. 9. Empowering Self-service analytics during COVID-19 • Over 250 BI Analysts and Citizen Data Scientists across the system • Strong Self Service Motivation • Centralized and Decentralized Analytics • Data Gathering vs. Data Insight / Analysis • Valuable time gathering, massaging, reconciling, and validating data • More time for data insights EMPOWER INSIGHT DECISION MAKING
  10. 10. Improving Patient Flow with AI/ML
  11. 11. Understanding Healthcare as a complex system Interdependent Cascading effect of bottlenecks leading to downstream patient leakage/loss & crowding Volatile Difficult to forecast variables across patient volumes, acuity, census, and staffing requirements Noisy Finding meaning and opportunities within the vast data ecosystem is a massive undertaking ED Elective Surgery Admit Transfer/ Direct Admit Inpatient OR Outpatients Discharge ARRIVALS INTERHOSPITAL TRANSPORT CAPACITY LENGTH OF STAY STAFFING READMISSION
  12. 12. DischargeAdmi t Inpatients Length of Stay Risk of Readmission Risk of Obs. Failure Discharge Disposition Inpatient Predictions Enable clinicians to prioritize the neediest patients first, and identify opportunity areas for systematic improvement. Optimize availability of inpatient beds Improve readmission rates & penalties Correct patient status assignment Outcomes
  13. 13. Length of Stay Risk of Readmission Risk of Obs. Failure Discharge Disposition Inpatient Predictions Enable clinicians to prioritize the neediest patients first, and identify opportunity areas for systematic improvement. Optimize availability of inpatient beds Improve readmission rates & penalties Correct patient status assignment Outcomes
  14. 14. © 2019 KENSCI CONFIDENTIAL Solution generating scores in minutes from data delivery 85% of encounters predicted within 1.65 days of actual LOS at the time of admission – establishing one of the best documented AI solution in a production setting Summary: Solution Performance
  15. 15. KenSci Realtime Command Center
  16. 16. Bringing Together Data & AI for COVID-19 Response Risk Stratification Realtime Hospital CensusSurge Planning IDENTIFYING PATIENTS AT HIGH-RISK PLANNING FOR PATIENT VOLUMES TRACKING PATIENT CAPACITY KENSCI SMART ON FHIR BASED DATA & AI ACCELERATOR Data Ingestion Pipeline Real-time data feed & InsightsRapid Secure Deployment Built for the Future
  17. 17. Unified Data Analytics Architecture with Azure Databricks and KenS Data Ingestion Agent Runtime ML engine Data Preparation Model Development Model Production ML OpsPipeline Monitoring Key VaultActive Directory RelationalBlob Data Ingress Integration EMR Claims IOMT Streaming Data ? Azure Cloud Services Model Experimentation Model Explainability Single Sign-on Role-based Security Visualization & Analytics Variation Analysis Feature Bank KenSci Analytics Portal
  18. 18. Learnings & Takeaways For other health systems For other Data & AI Teams
  19. 19. Feedback Your feedback is important to us. Don’t forget to rate and review the sessions.
  • DarianHarrison

    Jul. 10, 2020

Healthcare leaders today are faced with increasingly complex and unprecedented challenges. With COVID-19 taking the world by storm, the need for an intelligent system of insights that can proactively deliver actionable and real-time knowledge on patient populations is imminent to providing better care.

Views

Total views

3,610

On Slideshare

0

From embeds

0

Number of embeds

26

Actions

Downloads

71

Shares

0

Comments

0

Likes

1

×