The ability to use data better has benefits for councils, their public sector partners, and citizens. However, as local government strives to make better use of data assets, issues of privacy, confidentiality, cybersecurity, and ethics emerge, particularly around data linkage, analytics, and the use of predictive models.
For data linkage a lack of transparent and shared understanding about how local government data can be lawfully linked for different purposes (in line with GDPR requirements), along with a lack of data quality standards, makes it difficult for the sector as a whole to confidently make progress.
For analytics and predictive modelling (and particularly compared to NHS), there is:
- a lack of agreed standards and guidance (e.g. around pseudonymisation and governance of partnerships);
- insufficient debate around the ethical use of models;
- lack of transparency around the methods being used which limits intellectual challenge or learning from others, and
- insufficient evaluation of predictive capabilities.
Across all of these areas, public engagement and acceptability is key.
We will therefore:
- collaborate across local government to start sharing and aligning standards
- engage with residents about the acceptability of analytics, particularly the use and ethics of predictive modelling in council settings
- publish all methods and frameworks developed to promote transparency and learning.
- Produce benefits case through groundwork in ethical and information governance (IG) and citizen engagement for technical data linkage projects in local authorities (LA):
- Develop an IG framework for data linkage applicable to wide range of LA scenarios
- Draft guidance on incorporation of ethical data standards with practical use case
- Produce 3 exemplar Data Protection Impact Assessments
- Test the views of local residents on LA data linkage to inform ongoing citizen engagement and transparency
- Produce case study including draft guidance to share lessons learnt
- Transfer knowledge to increase confidence and understanding of specialist staff in Islington (LBI) and Surrey (SCC)
- Project Board Meetings: Monthly
- Citizen and Staff Engagement Events: Jan-Feb
- Draft guides, publications, briefings for peer review: Feb
- Final Publications (Mar):
- IG Framework for data linkage
- Guide on implementing ethical standards for predictive analytics, with case study
- Citizen and Staff Engagement Report
- Final report including benefits case.
Management: LBI will manage the project. LBI, SCC will both contribute staff time, expertise and contacts. In recognition that this is a complex area requiring specialist knowledge, we have partnered with:
- Kaleidoscope who have a strong track record of providing specialist IG expertise to the public sector.
- Oxford Insights, a UK-based international public sector consultancy specialising in artificial intelligence and data ethics.
Independent researchers will be commissioned to carry out citizen engagement work and survey of LA staff.
A project board, led by senior LA officers with representatives from digital, analytical, IG teams and partners, will monitor progress through monthly meetings. The board will be responsible for approving all publications.
Measuring achievement – Obj 1-3 will be met by delivery of the key publications. Obj 4 will be measured by pre- and post- surveys of staff groups
LAs need to harness the data they hold to improve efficiencies in service delivery and achieve better outcomes and experiences for citizens. To realise potential cost-savings, LAs need to link data across multiple and siloed data sources and explore the use of modelling to better predict demand for prevention and early intervention. This requires adherence to data protection legislation, consideration of the ethical impacts, as well as an understanding of citizen acceptability and trust. Non-compliance with data protection legislation, including GPDR, which puts LAs at risk being fined by up to £20m. The potential for reputational damage is high, making confident and rapid progress harder as citizens, rightly, lose trust.
For residents, our project will help increase understanding, ownership, and transparency around how their data is being used or could be used. Building upon previous work, we will seek resident’s opinions on the use of data for linkage and predictive analytics to inform the next steps for our digital data programmes. These will contribute to the wider body of evidence, for use by other councils and central government. Early and robust consideration of the potential harms of predictive analytics will help to ensure that there is appropriate consideration of bias within these models so that they are constructed in a way that does not have a detrimental impact on particular groups of people.
For local government, this work will:
- Help to deliver savings (time and money).
- Support IG teams and data specialists with tools and skills to help apply consistent standards across local government in these areas.
- Reduce duplication, particularly where councils are at an earlier stage in their digital data programmes.
- Help local government to more consistently conform to GDPR and avoid the risks associated with using poorly built predictive models.
- Raise the bar for engagement on data and transparency across local government.
Given the time constraints, we will maximise existing strong links with other LAs to proactively test out ideas and share learning.
LBI is part of GLA’s work on the London Office for Technology and Innovation, the pan-London GDPR programme, and with the NHS, the One London LHCRE. SCC is part of a Partnership with East Sussex County Council and Brighton & Hove City Council, sharing resources and knowledge across supporting services. It is also part of the Thames Valley and Surrey LHCRE.
Kaleidoscope works nationally across health and LAs at both local and national policy levels and will connect with policymakers and existing programmes for digital and data transformation. Oxford Insights works with governments at all levels on data ethics issues, including a project with GDS on ICT commissioning. They will share findings with their network and will incorporate any feedback.
If eventually successful, as an alpha project we would increase the number of councils formally involved.
The Board will publish the following outputs by March 2019.
Citizen and staff engagement report: This will include an evidence review of existing research into citizens’ views on data linkage and predictive modelling. As well as, findings from primary research with local citizens and staff on acceptability, perceived risks and benefits, and views on the right to object. Results will inform further development of privacy notices, policies on the right to object, and future communications strategies.
IG framework for data linkage with three DPIA exemplars: We will provide a specific framework on approaches to data linkage within LAs, building on existing work done elsewhere. It will address pseudonymisation, proportionality and necessity, internal governance, and compliance with DP laws. To illustrate the framework, we will produce 3 example DPIAs on complex issues of local and national importance: transitions from children’s to adult services, the social determinants of health, and a single view of residents and families.
Guide on ethical standards for predictive analytics, with case study: Building on the GDS Data Ethics framework and existing evidence, we will develop a draft framework for LAs, including questions that should be answered at different stages of the process and by different groups (eg SIRO vs data scientists) and the internal barriers to achieving this currently, as well as consideration of the biases in council datasets. We will apply the framework to a case study on the needs of vulnerable children.
Overall project report, including benefits: To include quantification of the local and national benefits of this work; summaries of outputs and learning; or a proposal to continue to alpha phase with our recommendations or a recommendation for the discontinuation of the project.
We will ensure that each output is ready to share online by anonymising our findings and creating easy to read outputs which use simple non-technical language.
The key users of this work are:
- Council staff, particularly those working in analytics, digital and information governance.
Proactive engagement with citizens about the use and acceptability of data linkage and predictive modelling is a major element of our work. We will appoint an independent organisation to conduct primary research with residents and/or local community groups with the objectives of increasing their knowledge and building trust, getting their views on acceptability, as well as on potential benefits and impacts.
The research organisation will also engage local authority colleagues through an anonymous online survey which will ask about their awareness and views on data linkage and predictive analytics, with the objective of understanding baseline knowledge and issues among specific groups.
Engagement of officers working in information governance (including the SIROs and Caldicott Guardians), digital and analytics will be key. They will be proactively engaged through workshops/meetings with the objective of understanding what they know, what would be useful for them, and how they think things would work best in practice. Through this work we hope to increase expertise and confidence across council IG leads on new and emerging uses of data, and for digital and analytics leads to be more confident about governance processes and understand citizens’ views.
We will also include colleagues who work in the areas for the case studies so that they can provide subject matter expertise to help ensure that our work is relevant to their service. For transitions and needs of vulnerable children this will include adults social workers and children’s services. For the single view of the resident and the social determinants of health, it will require a broader range expertise from across different parts of the councils.
We feel that we have built a strong partnership across two local authorities and with industry partners who have well-established track records in their specialist areas. Collectively, we undertake significant citizen engagement as part of our routine work, so are confident that we will find a research organisation for citizen and colleague engagement fairly easily.
Given that the GDS has already developed the data ethics framework and DCMS are launching a Centre of Data Ethics and Innovation, we want to build upon it in this project for the benefit of LAs, and it would be useful for us if the Local Digital Collaboration Unit could introduce us. It would also be helpful to be linked in to others working in central government on these types of issues, and particularly in relation to any of the case studies/specific areas we have outlined in the bid (eg needs of vulnerable children, transitions).
Finally, we would be grateful for support with dissemination via MHCLG communications.
We confirm that we have not been funded for this project in the past and have not applied for any other funding.
This is a new partnership and a new project idea.