HOW ARTIFICIAL INTELLIGENCE & DATA ANALYTICS ARE TRANSFORMING - - PowerPoint PPT Presentation

HOW ARTIFICIAL INTELLIGENCE & DATA ANALYTICS ARE TRANSFORMING HEALTH & CARE ONE LONDON LOCAL HEALTH & CARE RECORD

12th March 2019

INTEGRATED CARE & POPULATION HEALTH MANAGEMENT: SHARING DATA

LEADERSHIP – A LEARNING HEALTH SYSTEM FOR LONDON

What Do You Do With An Idea? Kobi Yamanda The story of one brilliant idea and the child who helps to bring it into the world. As the child's confidence grows, so does the idea itself. And then, one day, something amazing happens. This is a story for anyone, at any age, who's ever had an idea that seemed a little too big, too

• dd, too difficult.

….strive to become learning health systems by making clinical data research grade and lowering the cost of data acquisition and knowledge generation

Victor Dzau. Transforming Academic Health Centres for an Uncertain Future (2013)

Every consenting patient’s characteristics and experience is available to learn from Best practice immediately available Improvement is continuous This happens routinely and efficiently This is part of a culture

Charles Friedman. Toward Complete & Sustainable Learning Systems (2014)

ONE LONDON ENTERPRISE LAYERS

• 1. Extend and build upon a single approach to record sharing via federated exchange

mechanisms both within STP footprints and pan London

• 2. Data service for 9M registered people in London in full, plus those outside London

where treatment is delivered via London providers in part.

• 3. Exploit significant opportunities to enable active patient participation by linking them

to their data

• 4. Support for utilities that sit on top of the data service to provide population health

and business intelligence (information service).

• 5. To develop London-wide governance of trusted clinical improvement methodology
• 6. To provide a single data resource for research pan London

Minimum Required Optional

ONE LONDON STRATEGY – NEW MODELS OF CARE

New models of high quality, sustainable and integrated care Unscheduled care | Planned care | Population Health Management Robust digital operations in each organisation Ubiquitous viewing of records across care

• rganisation

Normalised data service for proactive care and population health management Patient access and control De-identified information for system planning & research

1 2 3 4

We need to ensure that each provider of care has in place secure, reliable, user-friendly and inter-operable technology systems (and

• perational processes) to capture the right information at the right

time, making the right thing to do the easy thing to do We need to ensure that providers of care are able to work together as an integrated system of care, with all relevant information about a person being held in a way that it can be viewed so that vital information is available to view (according to staff role) and so that people don’t need to tell their story over and over again We need to ensure that integrated care systems can move from providing mainly reactive care, to a position where it is possible to analyse clinical information across a population in order to spot people with high need, complexity or deterioration, and to offer targeted support to them We need to ensure that people really are empowered to be at the centre of care, developing systems that allow people to access their

• wn information, to add to that record, to register preferences, and to

use new technology to self-manage, to plan their care, and where appropriate to consume care services We need to ensure that integrated care systems have in place the infrastructure and applications to allow information about the whole population to be used to understand patterns of need and service utilisation, to predict demand and flow and to improve quality, as part

• f functioning ICSs and learning health systems

6

ONE LONDON TECHNICAL ARCHITECTURE

ENTERPRISE LEVEL 1 – LONDON PATIENT RECORD

400,000

November 18

ENTERPRISE LEVEL 2 – FRAILTY FLAG (NHS 111 API)

“As a London Ambulance Service (LAS) clinician who provides clinical expertise for NHS 111 … I want to obtain access to the local … information within the Discovery Dataset to determine whether the caller is potentially seriously frail.” Since the API went live on 27th November 2018, the Discovery Data Service has:

➢

handled 317,314 API requests

➢

matched 133,785 patients

➢

flagged 6,195 potentially frail patients as at 6th March 2019

AI?

Accelerating Artificial Intelligence in health and care: results from a state of the nation survey

• By mapping some of the methods employed by survey respondents against Professor Jeremy Wyatt’s complexity scale (see previous section), it can be seen that many of

the current solutions are using ‘lowest complexity’ advanced statistical techniques rather than more complex AI applications.

Autumn 2018 https://www.ahsnnetwork.com/wp-content/uploads/2014/12/AHSN-Network-AI-Report.pdf

Code of conduct for data-driven health and care technology

• respect for persons
• respect for human rights
• participation
• accounting for decisions

September 2018, updated February 2019 https://www.gov.uk/government/publications/code-of-conduct-for-data-driven-health-and-care-technology/initial-code-of-conduct-for-data-driven-health-and-care-technology

RECENT LHCRE-RELEVANT ARTICLES

Stella Creasy | London MP (Walthamstow)

“…it’s not for senior managers in the NHS to tell me what is going on [so I can] tell everybody else. My job is to create a platform on which a constructive conversation can happen about the kind of healthcare we need.” “[Public involvement is] not necessarily a trade-off between empowerment and equality. You get better decisions for the whole community and you hear voices you don’t hear in our current structures.” “if the NHS actively asked [the public] to make balanced decisions designed to “to help each other” rather than simply to meet their own needs it might be pleasantly surprised.”

Links: https://www.hsj.co.uk/comment/the-bedpan-its-difficult-for-people-to-imagine-different-could-be-good/7024439.article https://www.hsj.co.uk/expert-briefings/the-download-selling-patient-data/7024573.article

Ben Heather| The Download Column

“Senior NHS officials [have] insisted LHCREs will not be used to create a national patient “data lake” for researchers [but] NHS Digital told suppliers in October last year that the LHCREs will feed data directly into its central platform.” “…some local LHCRE [have said] it wasn’t what they’d ‘signed up for’” Quoting Dame Fiona Caldicott: “Dialogue with the public about data use has not grown at the same speed as the capacity of technology…Where there is a gap between expectations and reality, anxiety may grow” “Arguments for and against selling (or giving away) NHS patient data are complex and fraught. Is it exploiting confidential patient records for private profits or a public resource that, with the proper safeguards, can be used to save lives?.”

BUILDING PUBLIC TRUST

In all cases though, we must take the public with us … and avoid a complacency about the desire for digital and information sharing

Low levels of awareness, understanding and active support High levels of awareness, understanding and active support

LESSONS

1. Partnership working is hard (agreement first) 2. Expectation runs faster than delivery (at all levels-scope creep) 3. Public trust is critical (deliberative processes) 4. End point maturity is very variable (time to UPnP) 5. Technical Standards must prevail (FHIR API) 6. Some of what we are doing has not been done (invention vs procurement) 7. Align incentives and levers at all levels (hard & soft) 8. National data research approach must be aligned (DiH) 9. Land grab for data! (public vs commercial)

LEADERSHIP IN COLLABORATION (1)

LEADERSHIP IN COLLABORATION (2)

LEADERSHIP IN COLLABORATION (3)