NDR Webinar Series Technical Deep Dive

NDR Webinar SeriesTechnical Deep Dive15/12/2020

House Rules

2

Before we get started, we kindly ask that you follow a few simple rules to ensure the event runs smoothly.

• Please mute your microphone and keep your camera off throughout.

• Share your questions and comments on the Teams chat. Questions will be answered in the live Q&A at the end of the session. Please use the chat responsibly.

• There will be short comfort breaks at approximately 11am and 12noon.

• The session will be recorded and put online after the event. All names will be obfuscated.

Poll: Where are you joining us from today?We want to hear from you!

At various points throughout the event we’ll posing some questions to you. To start us off, please let us know where you are joining from today.

Please go to www.menti.com and enter the code 10 73 53 1 or scan the QR code on your mobile phone to take part.

We’ll be showing the results of the polls during the session.

3

1

WelcomeRob JonesChief Architect (NDR)

Welcome & Introduction

Mark FrayneAssistant Chief Architect (NDR)NDR Technical themes and architecture

10:15

10:05

Eugene O’SullivanAPI Senior Product Specialist

API management

10:45

George OlneyAssistant Chief Infrastructure Architect (NDR)Delivering a Track and Trace system for Wales

Mark FrayneAssistant Chief Architect (NDR)FHIR and interoperability

Mike GreenhalghSenior Product Specialist (BCU NDR)Kafka messaging and event streaming

Dale ParsonsProject Officer, SNOMED CT

National Terminology Service

11:50

11:30

11:10

12:15

Agenda and Speakers

Mark FrayneChair

Live Q&A Panel with speakers

12:35

NDR technical themes and architectureMark Frayne

NDR Assistant Chief Architect

Poll: Approximately how many systems holding patient data are in use in Wales?Please go to www.menti.com and enter the code 10 73 53 1 to take part.

6Technical themes and architecture 2

Poll: Approximately how many systems holding patient data are in use in Wales?

7Technical themes and architecture 2

NDR Programme Themes1. Opening up the Architecture

2. Standards-based Interoperability

3. Federated Welsh Clinical Data Repository

4. Data Analytics

National Data Resource

Opening up the Architecture Standards-based Interoperability Federated Welsh Clinical Data Repository

Data Analytics

8Technical themes and architecture

NDR Programme ThemesThemes 1, 2 ,3 are not sequential tasks, but describe how capability can be built cumulatively, each theme building up our capability to allow us to introduce Open Architecture, adoption of Open Standards and ultimately, the Federated Clinical Data Repository.

The work in theme 4 has in effect already begun, accelerated by the NDR team’s involvement with the COVID data hub, using theexisting ISD Data Warehouses. Work to ingest the remaining priority datasets will continue. Theme 4 includes also includes the creation of the NDR platform which will provide:

• Data pipelines to ingest, transform and merge datasets into the NDR, handling both bulk and real-time feeds of data

• Secure storage for large volume datasets

• Platform for data analysis tools and applications (such as Jupyter notebooks, for Python, R etc.)

• Provides new, or integrates with existing Data Visualisation tools (such as Power BI)

Opening up the Architecture Standards-based Interoperability Federated Welsh Clinical Data Repository

Data Analytics

National Data Resource9Technical themes

and architecture

1. Opening up the Architecture

Real time secure access to the data (01b)

Infrastructure and security operating model (02)

Ontology services (10)


API ManagementAPI management is an absolute necessity for opening up the architecture

Allows endpoints for APIs to be published openly

Provides enterprise strength security against malicious use

Requires both application and user to be authenticated

Provides fine-grained control over access to the data

Access is centrally logged and monitored

Rate capping can be introduced if necessary to protect performance of key systems

Performance telemetry provided via dashboard to ensure all APIs perform as expected


Master Patient IndexMPI

DiagnosticsWRRS

Introduce API Manager

SNOMED, Reference Data

OntoServer

Early opportunities for introduction of API management include:

1. Provide access to existing SOAP Web Services for WRRS, MPI, etc.

2. Access to SNOMED and NHS Wales value set reference data via the OntoServer FHIR API

3. Development of the infrastructure and security operating model for APIs

Access to new APIs via API management will be provided as they are introduced.

API ManagementAPI Management


2. Standards-based Interoperability

Real time access to the data (01b)

APIs and supporting standards (01c)

Patient facing applications (09)

Interoperability (11)


DiagnosticsWRRS

FHIRFACADE

SQL ADAPTER

Connect The FHIR Façade To Back End System The introduction on FHIR facades will allow the existing repositories of clinical data to present a FHIR API.

A facade allows a system to present an API to calling systems that provides the data in a different format i.e. it translates the data of the backend system into a different schema.

In this way, no change is required to be made of the backend system.


DiagnosticsWRRS

DocumentsWCRS

FHIRFACADE

MedicinesWHEPMA

Implement FHIR Facades Against Existing Systems

FHIR facades can be implemented against the existing national stores of clinical data that form the Single Patient Record:

• Welsh Result Reporting System (WRRS)

• Welsh Clinical Document Repository (WCRS)

• Welsh Image Archive Service (WIAS)

It is also possible to create a FHIR against an OpenEHR repository containing Medicines data as proposed by the WHEPMA project.


Create MPI Feed to FHIR Clinical Data Repository via Restful API

A feed from the MPI to FHIR Clinical Data Repository will be created to ensure patient demographics are accurate and kept up to date

• This provides a FHIR enabled search for patient data.

• This also allows the clinical data to be linked with up to date patient demographics, and changes to patient demographics, patient record merges etc. are managed centrally.

The feed will make use of the existing HL7 demographic message feeds which are currently routed by the Fiorano integration engine.

Clinical Data Repository<Patient>

Master Patient IndexMPI

HL7 -> FHIRAdapter

POST <Patient>

PUT <Patient>

Request Patient Composite Record

GET [baseURL}/Patient/patient.identifier=M12344|WPAS


Create Master Data Feeds to FHIR Clinical Data Repository

Master data feeds will be created so that the Clinical Data Repository holds accurate master data.

A feed from the Welsh Reference Data Service will be created to the Clinical Data Repository to allow clinical data to be accurately linked to clinicians, organisations and locations.

• Clinician data will be held centrally, in a standardised format

• Includes information about their specialties and professional registration.

Using the existing HL7 message flows, information from PAS and Emergency Department systems about patient encounters will be recorded in the CDR, and can be linked to clinical data.

• A timeline for the patient’s encounters with the health service can be easily presented.

• By recording pathways and linking them to clinical data it will be possible for a clinician distinguish, say, between diabetes activities and mental health activities.

• Data from other scheduling systems, such as from Radiology, Radiotherapy systems, Primary and Social Care can also be added to this feed by making use of existing feeds or developing new HL7 feeds.

Clinical Data Repository<Patient><Clinician><Organisation><Location><Encounter><EpisodeOfCare>

HL7 -> FHIRAdapter

RadiologyRADIS

ADTsPAS

Other Scheduling systems

Reference Data SQL-FHIR Adapter

TerminologyReference

Data WRDS


Engagement with UK CoreThe HL7 FHIR UK Core Programme set up by NHS Digital seeks to create a UK national standard for use by NHS organisations across the four UK nationals.

The NDR Programme will align with UK Core standards

NHS Wales needs to continue to engage with this initiative to ensure that the UK Core support NHS Wales information standards:The NDR Programme will contribute resource to form the UK Core virtual team along with colleagues from NHS Digital and the other home nationsNDR Programme will seek to create an NHS Wales Interoperability Working Group from the wider NHS Wales organisations to provide support, guidance and decisions to the NDR teamClinical and Terminology support is also requiredGovernance process to be established involving WTSB and WISB to ensure that the standards are fit for use in NHS Wales before they are formally adopted


3. Federated Welsh Clinical Data Repository

Self-service access to data (01a)

Real time access to the data (01b)

APIs and supporting standards (01c)

Federated Welsh Clinical Data Repository (01d)

Patient facing applications (09)

Interoperability (11)

Wider system requirements (14)


Federated Welsh Clinical Data RepositoryThe NDR team have engaged with a number of commercial and NHS organisations to seek to opportunities for learning and knowledge sharing.

The Yorkshire and Humber Care (YHCR) record provides an example of how, using FHIR interoperability, local and regional repositories of clinical data may be linked together to create a ‘system of systems’.

The systems brought together in within the YHCR include both national and local clinical data repositories, as well as GP supplier systems.

In this way patient information from across the region can be brought together into a single comprehensive platform. This approach can be considered a model for the Federated Welsh Clinical Data Repository.

Subject to agreement with the NDR Board, the NDR team are seeking to engage further with the YHCR to create a proof-of-concept application to test this approach in Wales


Secure Self-service Access to DataThe link below shows how the YHCR provides self-service access to data via the Query Builder application. This service provides secure and logged access to data for analysis.

https://www.youtube.com/watch?v=ZTYGx_LQA6M

It also can provide data to automated systems that, for example:

• Inform a clinician that a patient of interest has been admitted

• Automate workflow when a diagnosis is made

• Warn when an at risk child has been seen in ED

Technical papers for the YHCR are available via the link below.

https://yhcr.org/resources/technical-papers/



https://yhcr.org/resources/technical-papers/

YHCR Example Data Extract: Living with and beyond cancer



YHCR Example Data Extract: Living with and beyond cancer



4. Data Analytics

• Community, mental health and social care data (03)

• GP and pharmacy data (04)

• Single Cancer Pathway (05)

• National registries/audits (06)

• Diagnostics data (07)

• Ambulance data (08)

• Value-based Healthcare (12)

• Genomics research data (13)

• Wider system requirements (strategic operational and clinical) (14)


Data Gathering

Published data

GIG06SRVSQLISD2

WRRS

Analytics Server

Health Board 1

ADMISSIONSDISCHARGESTRANSFERS

MPIDATA

RESULTS

GIG06SRVSQLISD1

ndracqHere data is shredded into relation SQL tables/views

ISD consume & publish the data to Health Boards

POLLING

GIG06SRVSISISD1

poller

ACQUIHOUSE

BI

PRODUCTION ENVIRONMENT

Analytics Server

Health Board 3 etc

Analytics Server

Health Board 2

WCRS

XML Mortality

SQL Server Integration Services Polling

LIMS DR

NWIS Covid Dashboard View

POLLING

POLLING

SQL

Mechanics Of Covid 19 Data Gathering The NDR team has been involved in the COVID-19 data gathering process. This has made use of existing message flows from PAS, together with data from WRRS, WCRS and LIMS.

This diagram shows the process of data ingestion and delivery to the existing NWIS ISD reporting server.

Although this might represent a temporary process that has been expedited using the current emergency arrangements, this activity has provided valuable insight as to how these activities may continue in the longer term.

Work will continue to collect other priority data sets into the existing ISD environments.

Meanwhile, the NDR team will continue to evaluate alternative big data platforms that provide tooling for data ingestion, data analysis (including NLP, Machine Learning), data visualisation etc.


IngestionFabric

Streaming

ETL Extracts

Database data

File data

Consumption Fabric

Dashboards

Applications

Data Science

WRRS

ADMISSIONSDISCHARGESTRANSFERS

MPIDATA

RESULTS

WCRS

XML Mortality

LIMS DR

NWIS Covid Dashboard View

SQL

DiagnosticsWRRS

WCCG Test Data

Test Data Sets (tenanted)

`

PRODUCTION ENVIRONMENT

PseudonomyisedData Sets

Cloudera Internal Sandbox

The Cloudera proof-of-concept environment, hosted within the NHS Wales Data Centre will allow the NDR team to explore how data ingestion, storage and analysis can be achieved in an alternative ‘big-data’ platform.

The choice of platforms, languages and tools is huge, as shown in the following slides.

Work to explore and gain knowledge of these alternative cloud-based and on premises platforms will continue.


https://cloud.google.com/blog/topics/healthcare-life-sciences/getting-to-know-the-google-cloud-healthcare-api-part-1

Google Cloud Healthcare Platform


https://www.youtube.com/watch?v=ixJlsYc3Lwc&list=RDCMUCJ9905MRHxwLZ2jeNQGIWxA&start_radio=1&t=309

Microsoft Azure Synapse Workbench

DiagnosticsWRRS

WCCG Test Data

PseudonomyisedData Sets

Test

Dat

a Se

ts (t

enan

ted)

Other Test Datasets

Amazon Web Services

In a NutshellOpening up the Architecture

Standards-based Interoperability

Federated Welsh Clinical Data Repository

Data Analytics

FHIR RepositoryClinical Data Repository

<Patient><Organisation::Practice>

<AllergyIntolerance><Observation>

<MASTER DATA>

Application/Patient/

patient.identifier=M12344|

WPAS,Familyname=williams,birthDate=1982-

09-04

NDR Warehouse<ecosystem>

FHIR Façade(legacy

management)

FHIR RESTful

API

APIManagem

ent(URL

Scripts)

FHIR

WRRS<DiagnosticReport>

WCRS<Document>

MPI -> FIORANO

(MPI)

POST <Patient>PUT <Patient>

Reference Data Team(Adapter)

POST <Organisation>PUT <Organisation>

<Adapter>HL7 -> FHIRHL7

(WRDS)SQL Server

<AcquisitionServices>

ISD

Other sources(WCCIS, GP Activity …)

<AllergyIntolerance><Observation>

VisualisationBI tools

FHIRAggregator

Local Clinical Data Repository

Local – FHIR API

Strategic Alignment:Enable Federated repositories

Advanced Analytics

UKSeRP/NDR


National Data Resource (NDR)Project Plan v0


and architecture

National Data Resource (NDR)Technical consultation, governance and assurance planKey groups that we will use for consultation, governance and assurance:

• NDR Technical Steering Group (monthly, ongoing for the lifecycle of the NDR) - to steer technical direction and work

• WTSB - feedback and advice on technical standards and approaches, consult minimum quarterly

• Architectural Advisory Group - feedback and advice on technical approaches, consult minimum quarterly

• WIAG - to assure technical developments, consult minimum quarterly, but more frequently as required when developments need assurance (based on NDR Chief Architect(s) advice)

• Heads of Information and National BI/DWH Group - consultation and feedback, will consult a minimum bi-monthly

• CIF/CIO/ADI - consultation and feedback, will consult a minimum quarterly

• NDR Programme Board - considers feedback and recommendations from the technical groups and makes key decisions based on technical groups advice and recommendations, monthly (NDR SRO, Chief Architect(s) and Programme Lead to judge what requires Programme Board approval)


and architecture

Any questions?

34

API managementEugene O’Sullivan - Senior Product Specialist

Boris Napoleonov – Lead Developer

DHEW

36

Introduction

36API ManagementWhat Is API Management? - NGINX

https://www.nginx.com/blog/what-is-api-management

37

APIs

37API Management

• Application Programming Interface

• Defines a “communication contract”• One app provides the interface• Another app uses the interface

• Web APIs send/fetch data over HTTP Service

Client

Interface

API Examples

38API Management

C++

Response (C++ object: PrintDocumentPackageTarget)

HTTP

HTTP

POST /{page-id}/feed?message=Bore da!&access_token=... Host: graph.facebook.com

targetFactory->CreateDocumentPackageTargetForPrintJob(printerName,L"Example print job", ...);

GET /wrds/GetTable/nrds/WelshOrganisation?nrd%3Aorg_code=49998Host: sandbox.api.wales.nhs.uk

Microsoft Office App

WindowsPrint API

FacebookMobile App

FacebookAPI

Healthcare App

NHS WalesReference Data

API

Response (JSON data)

Response (XML data)

Print a Document

Post a message to a Facebook page

Get details of an NHS Wales organisation

Web APIs

• RESTful (FHIR)

• SOAP

• GraphQL

• gRPC

39API Management

{REST:API}

<SOAP/>

NHS Wales APIsExamples of NHS Wales Web APIs, available in production

40API Management

Demographics

Patient details such as:• NHS Number • Date of Birth• Gender• Address

Reference Data

Organisational data andreference, such as:• Hospital lists• Practitioner lists• SNOMED codes

Clinical Documents

Store and retrieve:• Referral letters • Consultation letters• Questionnaires• Consent documents

Test Results

Request tests and retrieve results for:• Lab tests on blood,

urine, saliva, etc. • Radiology scans

National Services

User Applications and Information Management Systems

Messaging Components

National Patient Record Systems

PharmacyPatient Record Portal

Master Patient IndexTest ResultsClinical

Documents

Radiology Cancer Child HealthLaboratory

Address Lookup Organisational Reference Data

HospitalPatient Admin

Documents & Referrals

GP Record

Test Requests

NHS Wales APIs

41API Management

demographics Lab

GPGP

Enterprise Service Bus

sendconsultationdocument

look up and synchronisedata

fetchpatient record

send and receive updates topatient demographic details

requesttests

WCP

WCRS WRRS WGPR

via WDS

WRRSWCCG

WAMS WRDSMPI

42

APIsStandards based interoperabilityhttps://www.hl7.org/fhir/organization.html

https://www.hl7.org/fhir/organization.html

Governance of APIs

• Unauthorized access

• Misuse of information

• Logging of access and usage

• Many considerations and concerns

• How can we answer these concerns

43API Management

Why API management is important

44API Managementholistic-api-management.png (1100×719) (theapipractice.com)

http://www.theapipractice.com/images/holistic-api-management.png

Tools and Options

• Enterprise Level• Azure• Apigee

• Gartner Report

• Cloud/Hybrid/On-prem

45API Management

46

API Management

46API Management

Developer Portal

API Gateway

Admin Portal

Documentation / API Catalogue / Standards/ Accounts / Access Analytics

Accepts API Call / Routing / Verifies API Key /Enforces Policies / Transformations / Analytics

Define and Import APIs / Create Products / Set up andadminister Gateway / Analyse usage / Manage users

47

API Management

47API Management

Application Developers

MY HEALTHAPP

Request

NHS Wales

API Management

LayerProxy

Proxy

security, authentication and access control rules

traffic managementand rate limiting rules

monitoring and dashboards

Response

What can API Management do?

48API Management

Authentication

Security

Auditing & Logging

Caching

Traffic Management

Analytics

49

API Lifecycle

• Publish

• Secure

• Authentication

• Mocking

• Versioning

API Management 49

50

Products

• Group APIs

• User Groups

• Subscriptions

API Management 50

51

Policies

• Limit call rate by subscription• Usage Quota• IP Filtering• Controlling• Transformation

API Management 51

52

Monitoring

• Metrics

• Alerts

• Auditing

• Insights

API Management

53

Benefits

Open Architecture and standards

Enterprise security against malicious

use

Standard authentication

approach

Access/Usage centrally logged and monitored

Control at subscription level for API integration

API Management Vision

54

Connected Experiences

55API Management

API Management Layer

AppDevelopers

Sandbox API QA Testing API Production API

User ID ProvidersNHS Staff ID (NADEX)

Patient ID (coming soon)

Monitoring, Logging and Audit solutions

Routing&

Caching

Authorization&

Security

Adapters&

Facades

Developer Portal

New Service Request Process

API Documentation

Change Agreement Board

Demographics Assurance Process

App Users

Self Service Access Request

API Development& Support

Service Management

Learn to use APIs. Request access to APIsUse apps that connect to APIs

Health App

56

API Management Journey

56API Management

Production API Management

Build a plan and vision for production APImanagement in Wales to support OpenArchitecture and Integration

Sandbox APIsIn order to build proof of concept demosand understand integration requirementsand standards we expose Sandbox andmock APIs

API CatalogueDHEW have started the process ofcataloguing NHS Wales APIs

01

02

03

57

NHS Wales Developer Portal

• Backed by API Management

• Proof of concept in setting up

• A step along the journey

• developer.nhs.wales

API Management 57

58

Production API Management

• Address the complexities of bringing in APIM• New APIs• Onboarding

• Can we speed up access• Migration of service• How it will impact current systems

• Service Management and API Management

• Investigate requirements and change to support NDR vision

API Management 58

59

Enabling NDR

• API Management Supports• Opening up the architecture

• Improve the way data is used.

• Common interfaces and technical standards

• Drive Standards based interoperability

Poll: What information and APIs would you like to see in a Developer Portal?Please go to www.menti.com and enter the code 10 73 53 1 to take part.

60API Management3

Poll: What information and APIs would you like to see in a Developer Portal?

61API Management3

Demo

62

Any questions?

63

Comfort breakWe’ll be restarting at 11:10

FHIR and InteroperabilityMark Frayne

NDR Assistant Chief Architect

Poll: 1. What is your knowledge of FHIR?

2. How important is interoperability to meeting the challenges of the NDR?

Please go to www.menti.com and enter the code 10 73 53 1 to take part.

66FHIR and interoperability 4,5

Poll: What is your knowledge of FHIR?


Poll: How important is interoperability to meeting the challenges of the NDR?


Why HL7 FHIR?

• HL7 defines interoperability standards

• In 2011, the Board of HL7 noted:• Interoperability requirements are increasing• Need for real-time (API) access to data• Vast increase in the amount, type and sources of data

• e.g. devices, genomics• Data is increasingly used for analytics, population health• Implementer expectations

Existing standards were lacking and a fresh look was needed…

69FHIR and interoperability


What is HL7 FHIR

71

• Fast Healthcare Interoperability Resources (FHIR)

• Consistent, simple to use content model (resources)• Controlled extensibility

• Supports all paradigms of exchange

• APIs, Document, Messages and Services

• Designed with implementers in mind

• Detailed online, hyperlinked specification

• Freely available specification, tooling libraries and servers

• Massive supporting community

http://hl7.org/fhir/ FHIR and interoperability

http://hl7.org/fhir/

A detailed online, hyperlinked specification

72https://hl7.org/fhir/R4/patient.html FHIR and interoperability

https://hl7.org/fhir/R4/patient.html

Resources – the Content Model

73

FHIR resources can be though of as ‘information Lego’ that can be joined together to build a model to represent clinical information and activity

FHIR and interoperability

FHIR Supports 4 Paradigms of ExchangeRegardless of paradigm, the content model remains the same


Example: First Consultation


Linking the data together

Each FHIR resource is linked to a unique, reliable URL. To build our models of clinical information and activity, resources can be linked together using this URL.

Links can be made between resources that are available of different systems, allowing us to bring the data together.


FHIR Profiles

77

The FHIR content model has been developed using the 80/20 rule:

• Design for the 80%, not 100%Only include data elements in the artefact if 80% of all implementers of that artefact will use the data element

Allow easy extension for the remaining 20% of elements

Extensibility is built into FHIR

FHIR Profiles provide rules for extending or, modifying FHIR resources, and can be formally published.

The UK Core is developing a set of FHIR Profiles for use across the UK to ensure that implementers are able to adopt a common set of rules for extending FHIR. The profiles are published as part of the UK Core Implementation Guide.


Extension adding Religious Affiliation, coded with SNOMED

Example UK Core Patient

78

Coded using the Wales data standard to record the ethnicity of a person, as specified by the

person, as per the Office of National Statistics (ONS) 2011 Census Categories

Extension to the Identifier data type

to include NHS Number Trace

Status

Meta info indicates this resource implements the UK Core Patient Profile Extension adding an Ethnic

Category field to the Patient resource


Example: Code System for Ethnicity

79Project title

The example CodeSystem shown is published as part of the draft Wales FHIR Standards – which are derived from the UK Core.

This code system defines codes that describe ethnicity of a person, as specified by the person, as per the Office of National Statistics (ONS) 2011 Census Categories.

https://simplifier.net/fhirstandardswales-test/datastandardswales-ethnicity

https://simplifier.net/fhirstandardswales-test/datastandardswales-ethnicity

Terminology

80

Code System: Defines a set of concepts with a

coherent meaning

CodeDisplay

Definition

Element Definition: Type and Value Set

reference

Binds

Value Set: A selection of a set of codes for use in a particular

context

Selects


3rd Party ToolsGoogle Healthcare PlatformCode libraries

ClinFHIR – Clinical Scenario Modelling Microsoft Azure API for FHIR

81Project titleFHIR and

interoperability

Massive Supporting Community


Any questions?

83

National Terminology ServiceDale Parsons

Project Officer (NHS Wales Informatics Service)

Poll: What is your knowledge of SNOMED CT?


85National Terminology Service 6

Poll: What is your knowledge of SNOMED CT?


Our project’s objective it to provide the tools to…

87

Use SNOMED CT in the Digital Health and Care Record for Wales.


What is SNOMED CT?

The Systemised Nomenclature of Medicine Clinical Terms: SNOMED CT is a comprehensive clinical terminology

The structure is a hierarchy of concepts and relationships

Support for multiple levels of granularity

88National Terminology Service

How is SNOMED CT Used?

SNOMED CT itself is only a part of the solution for effective electronic health records (EHR)

A terminology on its own 'does nothing‘

To realise the benefits of using a clinical terminology, it must be implemented and used as part of an application

RESTful API’s


What is the Role of SNOMED CT?SNOMED CT is designed to be used in clinical applications

EHR’s are usually designed with both a:

• health record data store and

• a terminology data store (Ontoserver)

The terminology data store loads data from a release file.

Patient records are then entered by the user using the terms presented


Common Misconceptions of SNOMED CT

91

• Terminologies are replacing Classifications

• Clinical Coders using ICD10/OPCS will be out of a job

• You have to use the whole of SNOMED CT

• You cannot keep narrative notes anymore

• You have to use a browser/picker to enter data in to the EHR

• There is a Reference Set for everything


Help us with a demonstrationA clinical term is about to be given, and we need your help.

Please go to www.menti.com and enter the code 10 73 53 1 or scan the QR code on your mobile phone to take part.

All you need to do is enter the term!

We’ll come back to this later on.


Ontoserver“A terminology server with fast, incremental search, and context-specific result orderingOntoserver puts SNOMED CT, LOINC, and any FHIR-based Code System at your fingertips.”

93

Key Features

• Ontoserver uses FHIR-based RESTful API

• It supports multiple clinical terminologies

• It has a ranking algorithm

• It is distributed using Docker


How to Interact with Ontoserver?Ontoserver has implemented parts of the FHIR Specification, these can be accessed programmatically using FHIR RESTful API’s for SCRUD interactions.

Ontoserver supports the following FHIR Resource types:

Ontoserver supports the following FHIR Operation types:

94

Code System Code SystemCode SystemCode System Value Set Concept Map

Code System$lookup Code System$subsumes Code System$validate-code Code System$expand Code System$translate Code System$closure


FHIR Resources and their Supported Operations

95

Code System Code SystemCode SystemCode System Value Set Concept Map

Code System$lookup

Code System$subsumes

Code System$validate-code

Code System$expand Code System$translate

Code System$closureCode System$validate-code


SNOMED CT Expression Constraint Language (ECL)

“The Expression Constraint Language is a formal syntax for representing SNOMED CT expression constraints.”

96

BASIC^ 991411000000109 | Emergency care diagnosis simple reference set (foundation metadata concept)

HIERARCHY>> 230690007 |Cerebrovascular accident (disorder)|<< 230690007 |Cerebrovascular accident (disorder)|>! 230690007 |Cerebrovascular accident (disorder)|<! 230690007 |Cerebrovascular accident (disorder)|

COMBINED^ 991411000000109 | Emergency care diagnosis simple reference set (foundation metadata concept) | AND <<230690007 |Cerebrovascular accident (disorder)|

ATTRIBUTE362965005 |Disorder of body system (disorder)| : 363698007 |Finding site (attribute)| = 12738006 |Brain structure (body structure)|


Implementation of Ontoserver, SNOMED CT and ECLMyeloma Outpatient Continuation Sheet (OCS)

The first iteration of the Myeloma form uses SNOMED CT content for:

• Diagnosis and

• Past Medical History

Using the SNOMED CT ECL we were able to constrain the API

97

ecl/<404684003 |Clinical finding (finding)| OR <243796009 |Situation with explicit context (situation)|&count=100&filter=token


Myeloma OCS UI


SNOMED CT and Secondary Analysis

How many patients have been diagnosed with a condition of the lung in an

outpatient setting?


SNOMED CT and Secondary Analysis ExampleBackground

Using synthetic FHIR patient data

Focused on two tables:

1. Patient and

2. Conditions*

*8,376 records, with 1,153 unique patients

How many patients have been diagnosed with a condition of the lung in an outpatient setting?


The Power of SNOMED CT?

101Countdown before “free text” entry closes: 3 slides National Terminology Service

SNOMED CT and Secondary Analysis (cont.)E x t r a c t t h e c l i n i c a l c o n c e p t s u s i n g a n A P I :

< 4 0 4 6 8 4 0 0 3 : 3 6 3 6 9 8 0 0 7 = < < 3 9 6 0 7 0 0 8

Re t u r n s : 1 , 0 7 9 c o n c e p t s

102

Descendants Self & Descendants

Clinical Finding Finding Site Lung

Countdown before “free text” entry closes: 2 slides National Terminology Service

SNOMED CT and Secondary Analysis (cont.) Clinical conditions related to the lung identified

Query “Conditions” table

returns: 54* records…..

* However, there are instances where patients have one or more of the conditions being searched:

• 25 patients had 1 condition,

• 13 patients had 2 conditions and,

• 1 patient had 3 conditions

103Countdown before “free text” entry closes: 1 slide National Terminology Service

Free Text Entry Review

At the start of this presentation you were asked to enter “Penicillin”, let’s review

those entries now.

104Countdown before “free text” entry closes: closed National Terminology Service 7

Free Text Entry Review


Penicillin – Live Data

How can Ontoserver, SNOMED CT and SNOMED CT’s ECL be used to reduce this?


Application and Ontoserver – Dynamic LookupThe return (per character) breakdown looks like:

3 chars: Pen, 2 entries (10%) were removed, leaving 19

4 chars: Peni, 8 entries (42%) were removed, leaving 11

5 chars Penic, 2 entries (18%) were removed, leaving 9

7 chars Penicil, 3 entries (33%) were removed, leaving 6

8 chars Penicill, 1 entry (17%) was removed, leaving 5

9 chars Penicilli, 3 entries (60%) were removed, leaving 2

Once “Penicillin” was entered one was eliminated (50%)

107

Pecicillin Pinicillin Pencilin

Pencillian Pencillin Pencillin

Penecilin Penecillin Penecillin

Penercilin Peniillin Peninilin

Peniciilin Peniclin Penicllin

Penicilin Penicillen Penicilln

Penicilln Penicillian Penicillin


Application and Ontoserver – Dynamic Lookup (cont.)If however the user interaction was to enter “Pen” or “Peni”…..

108

The underlying API has been constrained

i.e. removes:

pence (qualifier value)|

Pencil (physical object)|

Pension (qualifier value)|

Déjà pensé (finding)|

etc.


SummarySNOMED CT is the designated clinical terminology for use in clinical systems in Wales that will improve:

• individual direct patient care,

• disease management,

• research,

• service delivery,

• policy development

and thereby drive improvements


Imag

e co

urte

sy o

f Pe

te T

urnb

ull

Any questions?

111

Kafka messaging and event streamingMichael Greenhalgh

Senior Product Specialist (BCU NDR)

What is Apache KafkaApache Kafka is an event streaming platform originally developed by LinkedIn and open sourced in 2011.

It provides a horizontally scalable, high throughput, low latency distributed infrastructure for streaming events and messages.

Used extensively across the private sector including The New York Times, Cisco, Goldman Sachs, Oracle, Netflix and of course LinkedIn.

Messages can consist of any content although by encoding them using Apache Avro or JSON a schema registry can be provide to validate the format of messages.

A plugin (Kafka Connect) provides a flexible framework with connectors for most databases available today including those used in big data solutions such as Apache Hadoop.

113Kafka messaging and event streaming

Advantages of KafkaApache Kafka provides a distributed and fault tolerant infrastructure with messages replicated across a cluster. In the event of a broker failing others in the cluster will take over.

Low latency - able to handle large numbers of messages (thousands per second).

Durability – Message replication can be configured across the cluster.

Follows a publish/subscribe model, multiple consumers can consume from a single topic.

It is possible to add brokers to clusters without the need to stop the cluster and the load can be automatically redistributed.


Operating System decisionsKafka can run on windows but is not officially supported – a test server in this configuration proved to be unstable.

The majority of NHS Wales staff are more familiar with Windows and have little/no experience of Linux.

A couple of online sessions were organised with walkthroughs on installing a base Linux (Centos 8) server then installing Kafka and it’s dependencies on this server.

The support model going forward is yet to be determined.


Poll: How comfortable are you working with Linux?a) Never touched it Windows all the way!

b) Have used it a little in the past.

c) Quite comfortable, use on a regular basis

d) Expert


116Kafka messaging and event streaming 8

Poll: How comfortable are you working with Linux?

117Kafka messaging and event streaming 8

Security certificate fun!As messages are going to cross health board boundaries it is necessary for the traffic to be encrypted.

The Kafka documentation indicates that the extended use attributes (client/server/both) within the security certificates are ignored – this turned out not to be the case!

Certificates were generated with these attributes however the normal signing process within NHS Wales stripped the attributes from the certificate.

We are currently using self-signed certificates until this issue can be resolved.

Each of the plugins needs to be configured individually to use SSL/TLS and the documentation on this can sometimes be a bit sparse.


119

What will we be using it for?Incoming ambulance feed, enabling ED departments to see what vehicles are inbound and their estimated arrival time.

Referrals for cancer treatment passed from Cardiff and Vale University Health Board to Velindre Cancer Centre.

Real-time feeds from medical monitoring devices.


120

Next Steps• Plan some more online training sessions covering more

advanced administration topics such as security.

• Some health boards do not allow servers internet access, this means software packages/updates need to be manually downloaded and installed, investigate the possibility of building an internal NHS repository.

• Assist other health boards in building their infrastructure.

• Put a robust support mechanism in place.


Any questions?

121

Comfort breakWe’ll be restarting at 12:15

Delivering a Track and Trace System for WalesGeorge Olney

Assistant Chief Architect (Infrastructure)

Poll: How many days and how many people do you think it took to deliver the Track and Trace system in Wales?


124Track and Trace System for Wales 9

Poll: How many days and how many people do you think it took to deliver the Track and Trace system in Wales?

125Track and Trace System for Wales 9

Players

126

Technical Delivery Team – Gareth Williams, George Olney

Programme Management – Alison Maguire, Gemma Sullivan

Public Health Wales – Christopher Johnson, Amrita Jesurasa

Training – Neil Kitching

Testing – Mark Hardwidge, Julia Banks, Sharla Passariello

Reporting – Gareth John, Lauren Watts Keane

Business Analyst – Dan Lewis

Support - Rob Murray, Bethan Turner, Paul Lewis, Grzegorz Bargiel

Commercials – Matthew Perrott

SUPPORTED BY:

POWERED BY:

Track and Trace System for Wales

Contact Track and Trace Concepts

127

Index Case

Index Case

Contacts traced

Contacts traced

Contacts Tracked for 14 days


128

Off The Shelf Track and Trace SolutionsMay 2020

We found that they fell into one of three categories:

(1) Add-ons to an existing system and repackaged

(2) Conceptual solutions

(3) Not suitable for Contact Tracing


129

Realisation

129Track and Trace System for Wales

Background of Changing Guidance


131

Build Our Own Using A CRMEco System


132

NWIS Tight Technical Control Over Requirements

Define and Distill Requirements From / For

To Build A National Solution


133

22 Local Authorities +7 Health BoardsPublic Health Wales

3000+ users

Accessed using their own devices

Login using their Council Accounts in their own Azure Active Directory Tenancies

Other than AAD, completely independent of any local infrastructure.

Self contained, works with Edge Chromium or Chrome browsers.

Context Of Customer


Contact Tracing System Defined

134

TriggersCreate Case

Patient Web Form

Copy Of CRM Database for

Dashboards and Ad-hoc Reporting

And Assign To Correct Local Authority Contact Centre Queue

All Wales Positive/ Negative Test Results

On PremiseIntegration Server

• Contact Advisor• Text• E-Mail• Automated calls

Daily Checks By:

CRM


Wales Contact Tracing System Minimum Viable Product for 8th June 2020 Delivery

135

TriggersCreate Case

• Contact Advisor• Text

Daily Checks By:

All Wales Positive/ Negative Test Results

On PremiseIntegration Server

And Assign To Correct Local Authority Contact Centre Queue


Why use Microsoft?

136

• NHS Wales heavily invested in Microsoft technologies and excellent relationship

• Internal product knowledge

• B2B Authentication

• Dynamics CRM

• Cloud based solution

• Power BI Reporting

• Azure SQL

• Office 365

• Solution will touch and integrate with NHS Wales eco system


The challenge –Deliver Live CRM for 8th June 2020

137

Delivery TimelineContact

DVLook at Ireland

Dynamics Work

Define and build Wales PoC

Demo Wales PoCSign Contract

for Software and Services

Work with MCSBuild solution

Provide training

Live Across Wales


Test Trace Protect CRM User View

138


Test Trace Protect Example Dashboard 1/2

139


Test Trace Protect Example Dashboard 2/2

140

p /


Power BI Dashboard and PBI Gateway

141

PublicEndpoint

NWIS On Premise Network

Power BI Gateway

Firewall

Private Endpoint

On Premise DBs

Test Trace Protect Dashboard

InternetDynamics Data


Reporting Database Local Authority Access

142

Internet

Users

NHS Wales Conditional Access

PublicEndpoint

Dynamics Data


143

Patient Daily Check Methods

143

y

Manual Call

Text Message

Automated Call (IVR) in progress

E-mail


Wales Delivery Team

144

NWIS TTP Team (11)

Supported by:

22 Local Authorities +7 Health BoardsPublic Health Wales Welsh GovernmentNWIS Resource as required


Web forms ready for service before Christmas


146Monitoring Dashboard Track and Trace System for Wales

Tweets from Wales Government

147

Track and Trace

System for Wales

Wales Press Articles

148

Track and Trace

System for Wales

UK Press Articles

149

Track and Trace

System for Wales

Any questions?

150

Live Q&A PanelChaired by

Rob Jones – NDR Chief Architect