Predictive analytics triggering workflows to support carers, prevent falls and reduce loneliness

sitting in wheelchair

Problem to be solved

A report produced by the Institute of Public Care in May 2019 identified the top seven “primary risk factors” that have an impact on older people’s independence and wellbeing:

  • comorbidity
  • dementia with comorbidity
  • carer stress
  • falls
  • loneliness
  • poor self-esteem
  • poor self-perception of health

The presence of these risk factors - both alone and in combination - increases the likelihood that older people will experience crises in their health and care, resulting in poor outcomes and costly interventions such as hospital admission. What can be done to prevent or delay this escalation of need, so that older people can avert crises and stay well and independent at home for longer?

Nottinghamshire County Council - in partnership with community health, primary care and social care providers - concluded that they were not effectively signposting citizens to the interventions on offer which promote wellbeing and independence. Although gaps in people’s healthcare were reasonably well understood and evidenced through accessible and shared data, this shared picture was not in place for social factors beyond health conditions. This prevented effective signposting.

The project

Sharing more information about risk factors

Proactive Interventions aims to deliver direct holistic care interventions to Nottinghamshire residents at an earlier stage than would otherwise happen, triggered by analysis of data shared from multiple sources. By bringing more information about risk factors into a common data warehouse, the project will create an “early warning system” so that relevant staff can take appropriate actions quickly, before emerging issues get more serious.

Building on predictive analytics

The project builds on a predictive analytics proof of concept between 2018 and 2019 - supported with funding from an NHS Digital Social Care Pathfinders Programme grant - which showed it was possible to safely and securely share combined health and social information about older people aged 65+ between partner organisations by feeding data into and accessing a single data warehouse. With technical input from Canon Medical Systems, AI techniques were used to spot health care gaps - e.g. patients with severe COPD not under a community specialist nurse - and predict the risk for individuals of admission to hospital or a care home.

Expanding the range of daily social care data feeds

This project phase extends that work by expanding the range of daily social care data feeds coming into the warehouse - e.g. including more detail from assessments, data from assistive technology, and information from both social prescribing and contracted services. This expanded dataset will support better prediction and mitigation of individuals’ vulnerabilities to the risk factors highlighted above, focusing initially on carer stress, falls and loneliness. 

The predictive model will trigger alerts to health and care staff, with recommended workflows. These recommendations are important because practitioners do not always remember the variety of relevant services on offer. Tailored advice and signposting - e.g. to a carers’ hub - may be sufficient to support people with relatively simple needs.

In more complicated cases, it may be appropriate to offer an individual referral to a specific proactive intervention such as:

  • assistive technology equipment
  • falls prevention exercise classes
  • befriending and social activities offered by the voluntary sector. 

Currently, the project aims to increase referrals to:

  • universal services
  • existing health and care provision
  • community provision.

It’s possible that what is available may not meet what is needed through these additional referrals. The project will keep this under review and report into the regional Integrated Care System, for partners to take strategic commissioning decisions.

By following the progress of older people who take advantage of proactive interventions, the project will be able to analyse the impact of interventions on the primary risk factor(s), and evolve a more sophisticated model for proactive support. To save staff time chasing information and to support their decision making, future plans involve sharing more of the existing data held in discrete systems across health and care - e.g. dementia scoring by health, continence information from social care assessments. This data will be increasingly structured and comparable, through use of standardised questions and tools.

Lessons learned

Sharing data is difficult. Data sharing protocols need to be agreed at senior levels across partner organisations to unblock potential barriers such as risk-averse attitudes or local protectionism. Managers and staff at all levels need to understand the value and importance of collecting high quality data - it informs more targeted support - and of recording it in a way that can be coded and shared consistently.

Dealing with the COVID-19 crisis has helped to focus minds on the importance of data sharing - it saves time and improves the quality of decision-making. The project expanded to facilitate data sharing on homeless people and rough sleepers, together with vulnerable adults aged 18 plus known to primary care. Armed with the data, proactive interventions with a variety of services could be put in place to support this cohort.

To find out more about this project, contact Rasool Gore - Rasool.Gore@nottscc.gov.uk


sitting in wheelchair

Explore Further