KIE A technology toolbox for Health Care Clinical Decision Support Enablers Rachid Snoussi Senior Specialist Solution Architect, Business Automation 1 Marc Mangus Principal Specialist Solution Architect, Healthcare

2 ▸ Improved quality of care ▸ Better Engagement: Patient / Clinician / Partner(s) ▸ Regulations / Compliance ▸ Increasing Value to Business, Delivery Time to Market, ROI ▸ Efficiency - Reducing Debt: Technical and Operational Healthcare Leadership Drivers Introduction

For Healthcare Leadership Drivers 3 FHIR (by HL7) BPMN, CMMN, DMN (by OMG) PMML (by DMG) CloudEvents (by CNCF) Open source Platforms Open Standards Containers Automation (Infrastructure, Security ...) Intelligent DaaS OpenCDS/BPM+ Health (Decision Management) Open Data Hub (AI/ML) OptaPlanner (Constraint Solving) Initiatives / Practices Agile Innovation IT Modernization Hyperautomation Connected Health / IoT / Edge Clinical Reasoning / CDS Technological Enablers Introduction

Red Hat Healthcare Predictive Prescriptive Methods and Technologies. Types of Artificial Intelligence in use in business today. The challenge is using the right tool for the job and having the tools interoperate. Simulation Testing Production ● Mathematical Models ● Graph Models ● Deep Learning ● DMN / PMML ● Mathematical optimization ● Heuristics ● Constraint Solving ● Workflows: BPMN / CMMN ● Complex Events ● Resource scaling ● Self-healing ● Security remediation ● Data provisioning ● Workload management Data-driven Knowledge-driven

✓ Simplify maintenance and delivery processes ✓ Latest and most commonly used tooling (skill availability, support) ✓ Improved capabilities through community knowledge sharing (rapid innovation) Clinical Reasoning 5 ✓ Drive better care through expert-based rulesets and learning models ✓ Improve information sharing with ecosystem: Patients, Providers, etc. ✓ Build custom capabilities Optimizing costs Business Drivers IT Requirements Hyperautomation for Healthcare The Red Hat difference: Hyperautomation - Open AIML platform enablement - “One Stop Shop”

6 A Reference Architecture

7 7 Clinical Process Automation EXPERT / CLINICAL INFORMATICIST DEV Integration Layer API Management Layer µSrv µSrv µSrv Legacy Environment External Entities (Government, Insurance, Pharmacy, Labs, ...) EHR FHIR Server Process Driven Apps Open Data Hub Predictive Analysis END USERS Single Sign On Authentication & Authorisation SEC On-premise iDaaS CDS / Clinical Reasoning Auditing / Governance / Compliance Security / (HIPAA/GDPR) Compliance Automation IoT & Connected Health Business Process Automation Decision Management Resource Planning & Optimization µSrv µSrv µSrv µSrv µSrv µSrv µSrv µSrv Other Services (RPA,MDM, CRM, DMS ...) Messaging / Streaming Caching Collaboration Metadata Management Storage DEV DATA ENGINEER DATA SCIENTIST OPS

8 8 Clinical Process Automation EXPERT / CLINICAL INFORMATICIST DEV Integration Layer API Management Layer µSrv µSrv µSrv Legacy Environment External Entities (Government, Insurance, Pharmacy, Labs, ...) EHR FHIR Server Process Driven Apps Open Data Hub Predictive Analysis END USERS Single Sign On Authentication & Authorisation SEC On-premise iDaaS CDS / Clinical Reasoning Auditing / Governance / Compliance Security / (HIPAA/GDPR) Compliance Automation IoT & Connected Health Business Process Automation Decision Management Resource Planning & Optimization µSrv µSrv µSrv µSrv µSrv µSrv µSrv µSrv Other Services (RPA,MDM, CRM, DMS ...) Messaging / Streaming Caching Collaboration Metadata Management Storage DEV DATA ENGINEER DATA SCIENTIST OPS #1

9 9 Clinical Process Automation EXPERT / CLINICAL INFORMATICIST DEV Integration Layer API Management Layer µSrv µSrv µSrv Legacy Environment External Entities (Government, Insurance, Pharmacy, Labs, ...) EHR FHIR Server Process Driven Apps Open Data Hub Predictive Analysis END USERS Single Sign On Authentication & Authorisation SEC On-premise iDaaS CDS / Clinical Reasoning Auditing / Governance / Compliance Security / (HIPAA/GDPR) Compliance Automation IoT & Connected Health Business Process Automation Decision Management Resource Planning & Optimization µSrv µSrv µSrv µSrv µSrv µSrv µSrv µSrv Other Services (RPA,MDM, CRM, DMS ...) Messaging / Streaming Caching Collaboration Metadata Management Storage DEV DATA ENGINEER DATA SCIENTIST OPS #2

10 10 Clinical Process Automation EXPERT / CLINICAL INFORMATICIST DEV Integration Layer API Management Layer µSrv µSrv µSrv Legacy Environment External Entities (Government, Insurance, Pharmacy, Labs, ...) EHR FHIR Server Process Driven Apps Open Data Hub Predictive Analysis END USERS Single Sign On Authentication & Authorisation SEC On-premise iDaaS CDS / Clinical Reasoning Auditing / Governance / Compliance Security / (HIPAA/GDPR) Compliance Automation IoT & Connected Health Business Process Automation Decision Management Resource Planning & Optimization µSrv µSrv µSrv µSrv µSrv µSrv µSrv µSrv Other Services (RPA,MDM, CRM, DMS ...) Messaging / Streaming Caching Collaboration Metadata Management Storage DEV DATA ENGINEER DATA SCIENTIST OPS #3

11 Enabler #1 Decision Management

12 Enabler #1 Decision Management Business Automation Offering Key capabilities Optaplanner Business Optimizer Process/Case Management Decision Management

13 Enabler #1 Decision Management Decision Management Providing an easily manageable, single source of truth for clinical decisions Codifies clinical policies, treatment/therapy guidelines, and regulations/protocols Enables change to be easy by Clinical Informaticist (CI) / Subject Matter Expert (SME) Automates clinical decision making with the fidelity of an expert.

Enabler #1 Decision Management 14 DRL Based Rules ● Mature approach, largely used & adopted ● Multiple authoring tools ● Predictive decisions, by embedding PMML Models from an Clinical Reasoning ● Can be used for CEP problems ● Used by the OpenCDS community ● Standard based approach ● DMN Modeler/Runtime ● Portability: no vendor locking ● Predictive decisions, by embedding PMML Models from an Clinical Reasoning ● Used by the BPM+Health initiative Two Decision Management Options DMN Based Rules

15 LACE Score Hands-on demo! “LACE score” DMN decision service (The LACE index identifies patients that are at risk for readmission or death within 30 days of discharge.) - Length of Stay - Admitted via Emergency - List of Conditions - Nb of Emergency Visits within the last 6 Months Risk of Readmission Inputs Outputs Kie Server

16 LACE index Scoring Hands-on demo!

Source: RTInsights and Red Hat report. “Building a Clinical Decision Engine Platform with Red Hat Decision Manager,” Feb. 2018. Challenge Find new, innovative ways to support better healthcare with technology. Solution Northwell Health built a decision engine to let clinicians use self-service tools to easily design applications and set business rules. Why Red Hat Northwell Health chose Red Hat® solutions because they cost-effectively support better communication between hospital staff and patients. Results ● Improved the speed of decision making for faster diagnoses ● Enhanced the ability of clinicians and caregivers to design care plans ● Established real-time alerts for conditions and drug interactions At the point of care the physician would not only see the recommendation, but would also see the reasons why the engine is making that recommendation, in clinician- and business-friendly terms. Vipul Kashyap Director, Clinical Information Systems and Enterprise Information Architect, Northwell Health Products and services Red Hat Decision Manager Red Hat Enterprise Linux® Red Hat Process Automation Manager Red Hat JBoss® Enterprise Application Platform

18 Enabler #1 Decision Management Rules vs Predictive Models Rules (Knowledge-based CDS) Predictive Models (Data-based CDS) Created by people, based on experience, policies, laws, regulations ... Created automatically from analysis of historical data: ML Produce results that are explainable Produce results that are generally not explainable Subject to human error Subject to bias in historical data, legal/ethical issues Lengthy initial development, long-term results Quick deployment, rapid results Can we combine both approaches?

19 Enabler #1 Decision Management Predictive Decision Automation Data Scientists Clinical Informaticist Machine Learning Frameworks Medical Training Data ( EHR, documents, images, … ) Predictive Model Business Rules Knowledge Discovery / Machine Learning Decision Automation DRL Open Data Hub

20 Enabler #1 Decision Management Predictive Decision Automation Data Scientists Machine Learning Frameworks Medical Training Data ( EHR, documents, images, … ) Predictive Model Decision Model Knowledge Discovery / Machine Learning Decision Automation … A standards based approach for predictive decision automation Clinical Informaticist Open Data Hub

21 XRay analysis automated pipeline Hands-on demo!

*The Surviving Sepsis Campaign is a global initiative to reduce sepsis mortality. Source: “HCA Healthcare uses innovative data platform to save lives,” May 2019. Challenge Use predictive data analysis to find new, innovative solutions to healthcare challenges. Solution HCA Healthcare used Red Hat® software to build a platform to collect and analyze clinical data and signal caregivers in real time to improve early sepsis care. Why Red Hat HCA Healthcare wanted to support its care providers with an open source platform that was flexible, scalable, and fast. Results ● Saved thousands of lives by creating a data analysis platform to detect sepsis ● Built a reliable, highly available container environment ● Established a supported yet flexible environment for a learning health system As part of the Surviving Sepsis Campaign,* HCA Healthcare has saved approximately 8,000 lives from sepsis. Products and services Red Hat OpenShift® Container Platform Red Hat Enterprise Linux® Red Hat Ansible® Tower Red Hat Satellite Red Hat Insights

23 Enabler #2 Process Automation

24 Enabler #2 Process Automation Business Automation Offering Key capabilities Optaplanner Business Optimizer Process/Case Management Decision Management

25 Enabler #2 Process Automation Process / Case Management Wide Spectrum of process types Humans are frequently involved to add information and take decisions. the tasks sequence is pre-ordered and prescribed. ● Order fulfillment ● Purchase requests ● Medical workforce support ● Patient follow-up workflow Orchestration of automated activities, involves humans by exception. ● Automatic Laboratory Test Ordering ● Automatic claims processing ● Automatic reordering of medical equipments / furnitures ● Inventory reduction / rebalancing ● Order to cash automation Adaptive and dynamic processes: Knowledge Workers guide the activities flow by selecting among suggested options or creating on the fly ad hoc activities. ● Investigation solution ● Patient onboarding / offboarding ● Employee onboarding / offboarding ● Personalized patient care Straight Through Human Intensive Case Management Flexibility Automation Imperative / prescriptive sequence of activities Declarative unstructured activities

26 Enabler #2 Process Automation BPM+ Health CDS with open standards

27 Enabler #2 Process Automation BPM+ Health Examples

28 Enabler #2 Process Automation “What you model is what you execute ?” Really ? Process modeling 100% executable Process data, Mappings, and Expression language Process tasks logic EXPERT / CLINICAL INFORMATICIST DEV DEV

29 XRay’s manual assessment process Hands-on demo!

30 Source: Red Hat case study. “TMG Health accelerates to real-time data data access for clients,” March 2016. Challenge Provide healthcare data to customers in real time. Solution TMG Health migrated legacy systems to a new application platform built with Red Hat® solutions. Why Red Hat Red Hat provided a better price for the solution than competitors, and offered expert consulting during the project. Results ● Transitioned from batch to real-time data processing to let clients access up-to-date data at any time ● Gained potential to increase market reach by 50% ● Improved return on investment by lowering development costs by 60% Red Hat gives us a price point we couldn’t get from any other leading vendor. And we felt we could grow with Red Hat. Nicolaas Botes Executive Director of Enterprise Architecture, TMG Health Products and services Red Hat JBoss® Enterprise Application Platform Red Hat JBoss Developer Studio Red Hat Process Automation Manager (formerly Red Hat JBoss BPM Suite) Red Hat Fuse Red Hat Enterprise Linux® Red Hat Consulting

31 Enabler #3 Resource planning & Optim.

32 Enabler #3 Resource planning & Optim. Business Automation Offering Key capabilities Optaplanner Business Optimizer Process/Case Management Decision Management

33 Enabler #3Resource planning & Optim. Optaplanner Business Optimizer Do more business with less resources ● ● Optimize Goals With Limited Resources Under Constraints Reducing Cost of Care Improving Patient Health / Care Quality Doctors, Nurses ... Ambulances, Beds, Equipments ... Time vs vs vs Working Hours Skills / Affinity Logistic Conflicts Budget

34 Medical staff rostering in COVID-19 fight Demo!

35 Vaccination appointment scheduling optimization Demo!

36 Hospital Bed Planning Demo!

Automated shift scheduling software. Created by a physician on the frontlines EMEA Solution Equina is an automated shift scheduling application that helps you create and manage employee schedules. With an easy-to-use interface, Equina is designed to think like the shift worker and shift scheduler. Why Red Hat Equina was built to provide the best user interface for doctors and nurses to manage their shift schedules. To take Equina to the next level, we run OptaPlanner (an AI constraint solver), on OpenShift Online to create optimized schedules for healthcare workers. We are a new player in the market with the goal of providing a simple, affordable approach to automated shift scheduling in healthcare -- we look forward to a long relationship with OptaPlanner as we scale. Results - Equina helps you schedule doctors and nurses fast with a smart, easy to use app. - Equina schedule 5,858 physician shift hours in 360 seconds. Products and partner resources Red Hat Decision Manager (Optaplanner) Red Hat OpenShift Online https://www.equinascheduling.com/news-equina-meets-optaplanner equina

Axians’s myStaff is in use for shift rostering for anesthesiologists in some of the Belgian hospitals: Saint-Jean UZ Brussel, OLV Aalst, … Solution myStaff is a planning tool tailored to physician teams. Drawing up fair schedules according to the schedules, conditions and expectations of the service or department on the one hand, and the individual work preferences within the care team on the other: it is an almost impossible task for the human brain to solve this puzzle within an acceptable time. In addition, any change to the schedule means that all conditions, expectations and individual preferences must be re-evaluated. To make this process automatic, flawless and fast, Axians uses a powerful optimization engine. Why Red Hat Axians is a Red Hat partner and they used Optaplanner, our optimization engine, included in Red Hat Decision Manager to build myStaff. Results Planner benefits - Fair schedules according to work regime - Personal and flexible approach - Errors eliminated as much as possible - Focus more on core tasks Advantages doctor team - Optimal care provision - Time-saving and cost-efficient - Personal and flexible approach - Satisfied healthcare workers Products and partner resources Red Hat Decision Manager (Optaplanner) myStaff https://healthcare.axians.be/solutions/mystaff/oncall/

linkedin.com/company/red-hat youtube.com/user/RedHatVideos facebook.com/redhatinc twitter.com/RedHat 39 Red Hat is the world’s leading provider of enterprise open source software solutions. Award-winning support, training, and consulting services make Red Hat a trusted adviser to the Fortune 500. Thank you