Scottish Risk Prediction Frontiers Meeting SUMMARY NOTES FROM 6 SEPTEMBER 2016 SEPTEMBER 6 FROM NOTES SUMMARY

Background & Purpose

Risk prediction tools, used as part of clinical decision making, have been shown to improve the quality of care, prevent adverse events, and decrease costs. They have also been shown to help patients make informed choices regarding treatment. With the formation of Integrated Joint Boards in Scotland, the rich individual level datasets that are available and the established expertise in modelling techniques and linkage of health and care information, there is an opportunity to develop a national consortium of researchers in the use of predictive analytics to improve how we design and deliver person centred integrated care and support.

The aim of the meeting was to bring together lead researchers and practitioners in this field across Scotland to share their current interests and to explore mutual priorities and opportunities.

Meeting Structure & Content

The Risk Prediction event was hosted by Farr Institute, Scotland, on the 6th September 2016 in Edinburgh. It was co-chaired by Hester Ward (NHS National Services Scotland) and Anne Hendry (Scottish Government).

The event was well attended with representation from across the NHS, social care,

academia, Scottish Government and private sector. This report highlights the key points and actions from the event. Scottish Risk Prediction Frontiers Meeting

Presentations

Professor Andrew Morris, Chief Scientist, Scottish Government Health Directorates and Director of the Usher Institute of Population Health Sciences

SUMMARY NOTES FROM 6 SEPTEMBER 2016 SEPTEMBER 6 FROM NOTES SUMMARY and Informatics welcomed delegates and encouraged active participation in the meeting. There were a number of interesting presentations from across the public sector and private industry covering a broad range of different projects and ideas for the future.

The opening presentation from Ahmed Mahmoud Above: Ahmed Mahmoud (NHS National Services Scotland) described the national Risk Stratification tool SPARRA, which is used across Scotland to identify people who are at risk of emergency admission or readmission to hospital. Ahmed invited the attendees to inform and advise on the next stage of development of the SPARRA tool to improve predictive accuracy and to widen the data sets used to drive the algorithm.

Pete Whitehouse from Scottish Government shared Above: Pete Whitehouse an overview on a Delayed Discharges project. There was a wide ranging discussion regarding data sharing and best practice and the need to build momentum in this area.

This was followed by a joint presentation from Adrian Smailes (Napier University) and Brian Brown (CM2000) on how information from patient wearable devices is being developed to identify

and improve support for frail older people at home. Above: Brian Brown & Adrian Smailes The use of subjective data and free text was highlighted as being a key step in improving the data analytics and outcomes available, as was

better identification of patients suitable for this type

of home monitoring. Scottish Risk Prediction Frontiers Meeting

Presentations (cont.)

Prof Corri Black (University of Aberdeen) shared a detailed and thought provoking presentation on ‘data latency’- the time taken from data collection SUMMARY NOTES FROM 6 SEPTEMBER 2016 SEPTEMBER 6 FROM NOTES SUMMARY to taking the appropriate action. The importance of clinical engagement and involvement in the development process was clearly highlighted.

The keynote presentation was from Professor Josep Roca, Professor of Medicine at the University of Barcelona and lead for the ICT programme for chronic disease management at Hospital Clinic Above: Carri Black Barceolna. He presented on risk stratification and the multiple projects underway in Catalonia to drive improvements in predictive analytics. Professor Roca outlined the ambitious NextC@re programme for collaborative clinical management of multimorbidity grounded in prevention and anticipatory care, shifting the focus of analytical activity from population based to individual clinical decision making. He reflected the recent publication

comparing predictive risk modelling in five European Above: Josep Roca regions, including Catalonia and Scotland, and expressed interest in further joint work given many similarities in the delivery systems.

The final presentation was from Dr Kessar Khaliq and Lyall Cameron (NHS Ayrshire and Arran), who outlined the approach taken to develop a tool based on primary care data to identify patients who have complex care and support needs and would benefit from Anticipatory Care Planning or early intervention to manage their conditions. They noted Above: Lyall Cameron & Kessar Khaliq that the work was at an early stage and they invited the group to support further discussion and

development of this tool.

Scottish Risk Prediction Frontiers Meeting SUMMARY NOTES FROM 6 SEPTEMBER 2016 SEPTEMBER 6 FROM NOTES SUMMARY

Group Discussion

The final part of the event was a small groups’ discussion session looking at a gap analysis of current work, collaboration opportunities and horizon scanning for funding opportunities. The series of interesting and inspiring presentations sparked lively discussion and there was much enthusiasm for collaborative working to move ideas and projects forward. In particular, participants considered how to stay connected and to collaborate as a consortium to ensure Scotland is research ready ahead of future funding calls, for example Horizon 2020.

A number of actions and ideas for further collaboration were noted (see below).

In closing the event, Professor Morris emphasised the opportunities for collaboration and funding, and highlighted the small grants available to SME’s for the development of risk stratification tools and products. Professor Morris thanked all of the presenters for their interesting and thought

provoking presentations, and the organisers for making the event happen.

The suggestion of a follow up meeting hosted by The University of Aberdeen will be considered and communicated in due course. Meantime, if you are interested in being involved in taking forward any of the above actions (outlined below) please contact Dr Hester Ward [email protected] Scottish Risk Prediction Frontiers Meeting

Summary of Actions

1. Identifying Complexity in Primary Care and Predicting need for Anticipatory Care Planning (ACP) We need to move from reacting to ‘events’ (illness, falls etc) to predicting

SUMMARY NOTES FROM 6 SEPTEMBER 2016 SEPTEMBER 6 FROM NOTES SUMMARY deterioration and intervening early through anticipatory care planning. The new GP contract in Scotland will result in new models of interdisciplinary team working and anticipatory care in order to identify people with complex needs, who use high levels of health and social care resources.

Action: work with demonstrator practices in interested partnerships as a ‘living lab’ to develop models, building on SPARRA, using primary care and social care data against high health and social care utilisation and adverse health and wellbeing outcomes.

2. Enabling Use of Citizen Data in Prediction Tools Discussion of the role of SMEs as holders of care and technology citizen data and the potential to include social care data to develop more holistic predictive analytics.

Action: work together to develop appropriate methods and governance for inclusion of citizen data from self management applications and platforms, telecare, and home health monitoring within local health and care datasets.

Action: SMEs to consider applying to CSO small grant scheme.

3. Improve User Interface and Use of Risk Prediction Datasets Clinical engagement will be enabled if the tools are intuitive and easily accessible within daily routine care. Important to align this with ehealth developments in primary care and systems.

Action: to explore potential to use new platforms and interfaces to present RP data for use in primary care in light of new ehealth

developments.

Scottish Risk Prediction Frontiers Meeting

Summary of Actions (Cont.)

4. Develop New Modelling Techniques Scotland has expertise in machine learning and rigorous analysis of the existing linked data may offer new insights – particularly in the field of

SUMMARY NOTES FROM 6 SEPTEMBER 2016 SEPTEMBER 6 FROM NOTES SUMMARY multimorbidity and complexity chronicity as proxy for individuals who may benefit from proactive case management and earlier intervention.

Action: to consider multimorbidity datadive using cancer as entry point and using the extensive cancer and national datasets.

5. Delayed Discharges The limited progress in this national project to date could be overcome by taking this work forward from a local perspective.

Action: to work with a single health board that has a high volume of discharges and good health and social care linked datasets to test to develop this work further.

6. Future Collaboration with Catalonia The Scottish Government has an established MoU with the Basque Country and has a draft MoU with Catalonia. The Scottish Centre for Telehealth and Telecare is the main point of contact.

Action: to link with SCTT to contribute to the Scotland – Catalonia

collaboration and to explore forthcoming H2020 calls.

Scottish Risk Prediction Frontiers Meeting

Attendees

Felix Agakov Pharmatics Limited [email protected] Marion Bennie University of Strath Clyde [email protected]

SUMMARY NOTES FROM 6 SEPTEMBER 2016 SEPTEMBER 6 FROM NOTES SUMMARY Corri Black University of Aberdeen [email protected] Brian Brown CM200 [email protected] Jonathan Cameron NHS National Services Scotland [email protected] Lyall Cameron NHS A&A, Primary Care [email protected] Jessica Chen-Burger Heriot Watt University [email protected] Chaloner Chute DHI Scotland [email protected] Nathan Devereux Healthcare Improvement Scotland [email protected] Peter Donnan University of Dundee [email protected] John Frank University of Edinburgh [email protected] Rachel Hardie NHS Lothian [email protected] Donna Henderson NHS [email protected] Anne Hendry NHS Lanarkshire [email protected] Leonie Hunter NHS Lothian [email protected] Susan Kelso Scottish Government [email protected] Kessar Khaliq NHS A&A, Primary Care [email protected] Peter Knight IoRN [email protected] Paul Leak Scottish Government [email protected] Ahmed Mahmoud ISD [email protected] Jim Marple NHS Lothian [email protected] Michael McCabe NSS [email protected] Colin McCowan University of Glasgow [email protected] Andrew Morris University of Edinburgh [email protected] Fiona Muirhead NSS [email protected] Paul O'Donnell CM2000 [email protected] Samuel Oduro NSS [email protected] Andrea Pavlickova NHS [email protected] Jill Pell University of Glasgow [email protected] Rachel Porteous NHS [email protected]

Josep Roca Hospital Clínic de Barcelona [email protected]

Adrian Smales Napier University [email protected] Robby Steel NHS Lothian [email protected] Rishard Stoddart CM2001 [email protected] David Titterington NSS [email protected] Hester Ward ISD, NHS National Services Scotland [email protected] Peter Whitehouse Scottish Government [email protected]