Box 13 provides a summary of Chapter 8 of this report.
Summary of key points
- Incorporating AI-based technology into existing models of social care provision requires alterations to existing funding models and care pathways, as well as concerted training to increase the digital literacy of the workforce.
- New and emerging technology-enabled care solutions require a robust digital infrastructure, lacking for many of those who most draw on care and support.
- Short-term service pressures and a sense of crisis management are not conducive to the kind of culture and approach that might be needed in order to reap the potential longer-term benefits promised by AI-enabled technology.
Beyond the specific issues related to the reliability and stability of the digital technology described in Chapter 7, sites experienced a series of dilemmas as to how to make the technology work in practice.
Funding, training and resources to incorporate new workstreams
There is an assumption that AI-based technology may, in the long term, reduce the need for human analysis (i.e. the technology would identify key patterns in people’s routines and raise an alert if a significant deviation was detected, without a member of staff needing to monitor the data in detail and make this judgement). However, the reality for a number of sites was that IndependencePlus generated a large amount of data which needed significant staff time to monitor and interpret. Current funding models mean that care providers are paid by the hour for the time spent providing in-person care, not to monitor additional data such as those produced by the IndependencePlus sensors. This also raises questions about who is responsible for monitoring the data from the sensors, and some sites seemed uncertain around exactly who was responsible for what in this regard. This was particularly the case when the technology evolved to include more personal health data, leaving some care staff feeling that they may receive medical data which they did not feel qualified to interpret or comfortable receiving:
Our lead domiciliary care provider, they’re not geared up to looking at health data and making health judgements based on that, so quite rightly they were saying ‘we’ve got this thing that says heart rate spike, what does that mean, do we have to contact a GP, what’s going on?’ So there was a lot of confusion around that, and we actually stopped using the system through COVID because of that.
The whole point around these sensors and these monitors is it’s just continuously feeding on data that a client produces, so the challenge to us in social care is, well, is some of this health-related data? In which case, do we have a real remit to be capturing and processing that data?
Because so much of it is health metrics or what would indicate medical problems or something that needs medical attention and support, I think that it needs to be the team that knows the most about that information, it needs to be people who can interpret what normal heart rate data needs to look like and what normal sleep patterns might look like… so it needs people who know what they’re looking at, know how to interpret it and are skilled and already knowledgeable about how to take that medical data and turn it into actions. This is when we need to call the GP. This is when we need to call an ambulance. This is when we need to change these meds. This is when we need to – you know, it needs to be the people who will make those medical decisions who are interpreting those data.
Incorporating AI-based technology into service delivery also requires that workstreams are reconfigured to accommodate sustained data collection. However, the current model of care management in many local authorities requires a social worker to assess need, determine eligibility for care and support services and design an initial care package, with a separate reviewing team (or even the provider itself) carrying out what can be a fairly minimal review at infrequent intervals. This series of relatively short-term contacts with care practitioners means that there may be few, if any, professionals in the person’s life who might be well placed to understand a baseline of someone’s usual routines and respond to sudden variations. Of course, there is no reason in principle why care agencies contracted to provide care could not monitor AI data, if this formed part of an agreed care package, though this needs a fundamental shift in the skill set and expectations for some members of the workforce:
We’re changing the way [we provide] care and that needs some different mindsets and skills or additional to what the care staff have. So the care staff generally have very – they want to be supportive and help people – they’re that kind of character generally – some more than others are comfortable around technology.
Interviewees from regulatory organisations noted that social care, in comparison to health care, has fewer professional regulatory bodies. This may create a risk that training on the use of these AI technologies is not standardised. Fewer professional regulatory bodies might also make it more difficult to collate an evidence base on the efficacy (and cost-effectiveness) of a new technology used in a social care setting.
Digital infrastructure
People who draw on care and support do not always have appropriate digital connectivity to enable or maximise the use of these technologies, especially in rural areas. Organisations providing social care also need to have appropriate and robust IT systems in place whereas, in reality, some reported the lack of Wi-Fi or the absence of a work computer to view the dashboard:
In areas that had poor signal or Wi-Fi connections, you could be thinking that everything’s going alright and then they might be in need of help and you wouldn’t sort of know because the Wi-Fi’s dipped or something like that, you know, so they were the doubts going through my mind at the time.
We didn’t know that, you know, that there would be places that staff don’t know their e-mail addresses and have worked there for 20 years and there’s no staff computer. So we’re telling them ‘you just log into this database with your e-mail address and you can see all this data from this person’ and they’re like ‘we don’t have a computer here’. That is a problem!
This creates challenges as the effective use of AI-based technology requires large amounts of good-quality data to be both collected and shared with the relevant individuals.
Cultural shift towards preventative care
However, perhaps the biggest barrier to successful implementation was to do with a broader cultural shift from one that rewarded short-term gains towards one that recognised the longer-term benefits of a more preventative approach (which might take years before any investment is recouped):
Social care doesn’t have a predictive, preventive culture. It will be great if someday we do, but right now all of the incentives and all of the monitoring and the way sites are graded for quality is not based on preventing issues, it’s based on handling the ones in front of you. So there’s a lot of work, kind of, at a more systems level on how social care is monitored and how quality is assessed.
This is about getting big data and going ‘well actually in 10 years’ time Mr Smith is likely to develop these conditions because people with these sorts of characteristics develop these conditions’. Now I completely and 100% support that as we move towards a place-based system, but councils and CCGs are just not there – and that’s a real challenge for them. They can’t fund stuff that’s going to save money in 10 years’ time. They need to save money tomorrow, and that’s a real challenge for actually doing anything that’s genuinely going to prevent increase in need over time.