UK Ministry of Housing, Communities and Local Government (MHCLG) 
Care & Housing predictive model for improved life chances
The cost of providing social care to families has increased by 8.4% in Waltham Forest over the last 5 years and at the same time the pressure on providing quality homes in London as result of the housing crisis is increasing dramatically. The new ten year housing plan target (2019 – 2029) in Waltham Forest is 17000 new homes and in Hackney, it’s13000 aimed at coping with increasing demand. At the same time the cases of social care for families has increased for the over 65s by 24% on average across both boroughs over the last 3 years. The demand for temporary accommodation has also increased by 9% from 2015 creating even more pressures for councils and residents in London.
We have a wealth of information which could improve supporting services and move the Council to become more prescriptive and predictive in our approach. This could ease the burden on councils.
The data to run and support Social Care and Social Housing services are held securely in separate systems and directorates across Councils all over the UK. We store an enormous amount of information in order to work out what we need to provide families for improved outcomes and care plans for vulnerable adults and children in care. Both data sets have common indicators which could provide improved likelihood for better outcomes, however, the data sets are disparate and not correlated in any way we believe we miss opportunities to provide better care, homes and outcomes for residents across London.
The impact on families to care for vulnerable adults and children in the home should affect the outcome of housing decisions by providing better provision. Indicators in housing could provide a likelihood of treatment or packages for people in Social Care, whereas the reverse should be true, indicators in social care systems should provide a likelihood of improved housing needs in the housing service.
The tables below show the demand from SALT for adult social care in Hackney and Waltham Forest
| Number of requests for support received from new clients | |||||||||||
| 18 to 64 | 65 and over | ||||||||||
| 2015-16 | 2016-17 | 2017-18 | 2015-16 | 2016-17 | 2017-18 | ||||||
| Hackney | 2,790 | 4,370 | 4,180 | 2,930 | 3,190 | 3,370 | |||||
| Waltham Forest | 3,140 | 2,860 | 2,235 | 3,240 | 3,645 | 4,310 | |||||
|
|
|||||||||||
| New clients with an episode of short-term care and a known sequel | |||||||||||
| 18 to 64 | 65 and over | ||||||||||
| 2015-16 | 2016-17 | 2017-18 | 2015-16 | 2016-17 | 2017-18 | ||||||
| Hackney | 80 | 75 | 50 | 375 | 375 | 330 | |||||
| Waltham Forest | 110 | 90 | 85 | 870 | 715 | 770 | |||||
| Receiving Long Term Support during the year | |||||||||||
| 18 to 64 | 65 and over | ||||||||||
| 2015-16 | 2016-17 | 2017-18 | 2015-16 | 2016-17 | 2017-18 | ||||||
| Hackney | 1,635 | 1,550 | 1,180 | 1,855 | 1,885 | 1,905 | |||||
| Waltham Forest | 2,025 | 1,785 | 1,785 | 2,065 | 2,120 | 2,050 | |||||
Statistics for Homeless and Temporary accommodation in Hackney and Waltham Forest.
| 2008-09 | 2009-10 | 2010-11 | 2011-12 | 2012-13 | 2013-14 | 2014-15 | 2015-16 | 2016-17 | 2017-18 | ||
| Hackney | Households accepted as homeless, eligible and in priority need | 615 | 646 | 814 | 686 | 698 | 906 | 902 | 1,017 | 803 | 949 |
| Households in temporary accommodation | 1,654 | 1,384 | 1,296 | 1,313 | 1,523 | 1,755 | 2,021 | 2,495 | 2,900 | 2,861 | |
| Waltham Forest | Households accepted as homeless, eligible and in priority need | 419 | 286 | 311 | 600 | 1,045 | 804 | 1,051 | 1,087 | 820 | 586 |
| Households in temporary accommodation | 1,736 | 1,240 | 1,101 | 1,307 | 1,325 | 1,469 | 1,990 | 2,181 | 2,299 | 2,235 |
The council doesn’t have the capability to combine information together to look for opportunities for better life chances for our most vulnerable residents. This proposal looks to address this need and provide a tool with modern data techniques to intervene for families who may need help further down the line.
Waltham Forest has been running a pilot with Amazon Web Services to see how we can use this data to plan our services better. We found:
- Correlation between social housing customers and families being helped with social care. Specific areas where this is more prevalent in the borough.
- It may be possible to enhance our quality assurance by applying machine learning from past case notes.
- Using Cloud technology we can link to legacy systems and analyse data securely, quickly and efficiently.
We will build on this to:
- Create reusable dashboards which show correlations between Housing and Social Care data and can be actively used by operational managers across to re-plan their services in order to intervene earlier in cases which typically required more intervention.
- Create at least one operational use case where machine learning can be used to actively quality assure social care / housing decisions to ensure a fair outcome.
- Develop a reusable predictive analytical model against the data to provide a better mechanism for intervening in families sooner.
- Continue the development of the machine learning algorithm against the case notes to provide an early indicator of outcome for social workers.
- Develop a model to predict homelessness to enable better life plans in housing.
- Create input APIs to enable plugin capability for other councils.
We will partner with the London Borough of Hackney so that this capability will be reusable across all Councils once developed.
We estimate the current cost of continue the pilot and discovery stage to create a series of dynamic real time dashboard, predictive model and machine learning algorithm is the following;
- The cost of the AWS platform for consuming the data.
- Rental cost of the visualisation and analytical tool
- The cost of the utilising third party experts
The benefit to residents in councils across the UK is to provide intervening measures for families much earlier than we can deliver currently. This will reduce the burden on social care, reduce the pressure on providing provision for homelessness and provide models to improve the outcomes in Social Housing. This will mean a reduction in long term cost and an increase in revenue from housing for councils.
At this stage we are focusing on developing the models to improve on quality and form part of the input for Alpha stage to prove the model in the field.
The model will connect multiple systems: i) Adult Social Care and ii) Housing Manage Systems in Waltham Forest and together with London Borough of Hackney we plan to create APIs so that the underlying data model can be reused. The success will be that Hackney can reuse the model with minimal effort and recreate similar results.
This approach will help refine and demonstrate the reusable components.
The outputs for the project include
- A business case describing the overall cost, benefits and savings for the providing predictive models across social care and housing.
- A report describing the model and showing the potential for this approach. This will include the results as well project conclusions.
- A series of real time dynamic dashboards for use in the field on tablets and hybrid devices for Social Workers and Housing Officers.
- A re-useable model ready for use in the Alpha project .
The users for the project are social workers and housing officers who work face to face with residents across the borough. They will critical in defining the dashboards as well as refining the predictive and machine learning algorithm to improve outcomes.
The current pilot phase was developed using AWS credits. We have received no additional funding at this point.