The DASH project explores the potential impact of integrating new and emerging data sources on emergency response and wider policy. DASH has been underway since April 2017 and is due to complete with dissemination of findings in March 2018.
A broad range of participants from academia, business and ambulance services were invited to participate in an all-day project workshop at King’s on 12 September 2017. The workshop invited discussion on DASH priority data sources in structured format of presentations and conversations over three sessions as follows:
- Health & social care data.
- Environmental data (including weather and air quality).
- Mobility & transport data.
About 40 participants attended the workshop through the day. Thanks to all who attended for their lively contributions. Special thanks to those who kindly made presentations on their work (see agenda below), and to the London Ambulance Service (LAS) for their continuing collaboration.
This brief report sets out to summarise some of the observations made on the day and to give an impression of the discussions.
“[the] mission, really, is to care for the people of London, making sure they get the right help and all of the help that they need”
Session 1 – Health & social care data
Much of the discussion in this session revolved around the idea of better routine linking between ambulance and other health and care datasets, to improve outcomes generally by subjecting interventions to more rigorous evidence standards.
“[paramedics] take patients into hospital and have no idea what happened them. There is no routine link between the ambulance service and other NHS organisations… As data becomes more and more collected electronically, it provides an opportunity … to explore whether this data can be linked.”
“for London, 5,000 calls a day and 3,000 incidents gives a snapshot of London’s health day to day.”
Participants were interested in the power of data to illuminate the social roots of demand for emergency ambulance services, and in using it to develop new conversations between the worlds of health and social care.
“understanding demand differently … and what institutional arrangements might London Ambulance consider or changes to institutional arrangements to interact with the changing health and social care provision landscape [in which it] operates.”
“I think [LAS is] the only London-wide trust and [it is] engaging with – I don’t know how many … different things for different areas. If you are a paramedic and you’ve hopped your way across London, you won’t know what local initiatives are available in an out-of-area catchment that you [don’t know] as well. There’s variability that way as well.”
Although there was enthusiasm for the potential power of health & social care data linkages to support ambulance dispatch improvements, various comments highlighted challenges and risks to the enterprise (in terms of information governance as well as economic and technical, for example, and in terms of resources) and there was some discussion of what might be done to address them.
“It was not an easy process. On average it took us about a year and 13 weeks from the moment of senior approval to the time in which we had a linked dataset…. We found that every IG manager, interpreted the rules slightly differently.”
“It’s not something you can dip in and out of for five minutes here and there; it needs focused attention to get anything meaningful out of it. So [it] need[s] someone doing it almost full time.”
“one of the key things that’s accompanying, maybe unfortunately, the rise of data is that people realise/companies are realising the value of data. As soon as they realise it has value they don’t want to give it away anymore and so this idea of what’s shareable, who owns what and what should be standardised becomes a much more contentious topic.”
“There doesn’t seem to be anyone coordinating, either providing systematic case studies that illustrate how you can make that jump from convincing analysis into transformational action or indeed providing evidence that something has…. What actually has changed on the ground, aside from maybe shifting services from one area to the other?”
Session 2 – Environmental data (including weather and air quality)
“hot and cold temperatures and air pollution have a significant negative impact on ambulance performance”
There is a specific body of evidence which shows the negative impact of weather on ambulance performance, both due to increased demand and to obstructions to response. The discussion here dealt primarily with the challenges to making better use of weather forecasts, contingent on the questions of what forecast data is available and what ambulance services can do to adjust dispatch systems to respond to them effectively.
“we have a whole sweep of environmental data of one sort or another that’s being generated; the kind of Met Office traffic and petabytes of data about all sorts of things…. [There is] the relationship between weather phenomenon and demand for health services, but equally there is also a relationship between weather data of one sort or another and, probably, the ability of the health service to supply… the HR and other sorts of things.”
“There’s a question about whether the forecast product, which it’s shown that the DOH budget is buying, is actually really tied into stuff that’s meaningful for [ambulances, in terms of] when that product … kicks in and actually whether it’s even useful”
Although there is currently less evidence on air quality impacts, the general impacts of poor air quality on health are now well-known. The discussion focused on recent advances in local area air quality modelling and issues of delayed cumulative impact.
“there’s a whole suite of environmental data out there, and just as it’s possible to link into various social care settings, there’s also the environmental data and the potential for linking that to improve intelligence and response is … substantial.”
Session 3 – Mobility & transport data
Mobility data is of interest to researchers and practitioners concerned to understand demand for ambulances, because of the idea that greater numbers of people in an area will tend to exhibit more demand for ambulances. The discussion considered data from mobile telephony networks and transport systems as well as new statistical demographic techniques for understanding mobility around cities. Some comments highlighted the role of a person’s behaviour as well as their location in conditioning ambulance demand.
“If we know what the population of a region is, we can reallocate. It’s effectively a population reallocation algorithm, which is the best we can do for a specific time.”
“the public don’t [necessarily] know when to call 999 or what steps are [properly] involved”
DASH work to date has indicated significant potential to harness transport data, and especially data from intelligent transport systems (ITS), to improve the capacity of ambulance services to respond quickly to medical emergencies in congested cities. The participants tended to agree that this represents a concrete opportunity for ambulance dispatch, and the discussion centred on the risk that ambulance service staff will not be comfortable with decisions taken exclusively by machines within ITS.
“It’s to adjust the lights before the ambulance gets there. So when the ambulance receives the call and there is an emergency [service] response, as soon as the ambulance is moving, the GPS will signal … and then change the lights… effectively clearing the … route, so that the ambulance can go through safely and, hopefully, more quickly.”
“it is important to help the human to understand what AI system is saying … I think the idea is that actually we want to, not necessarily [make it] easier for [staff], but free them up so they can do that more complex decision making”
“TfL has a data sharing agreement with the police with the traffic data. Will there be some interest in doing something similar with the LAS?”
DASH Principal Investigator Dr. Elizabeth Sklar concluded the workshop by highlighting issues of standardisation, ownership, practical implementation and the extent of data required to generate actionable insights. She also addressed the question of ambulance services’ effective capacity to adjust decision-making practices in the light of insights from data.
This workshop successfully validated DASH investigations into priority data sources and generated a wide range of observations, reflections and suggestions for ongoing research. It also advanced a shared interest in and understanding of DASH research questions amongst participants, including various continuing conversations between participants on actions to put some workshop ideas into effect.
Participant feedback indicate that the workshop had been useful, with respondents tending to highlight mobility and transport data as meriting further research for ambulance dispatch purposes.
DASH is funded by the Economic & Social Research Council (ESRC) grant ref. ES/P011160/1 (April 2017 – March 2018)