RealVNC mid-banner general
Advertisement: Wild Knight Vodka
Advertisement EY mid banner
Barr Ellison Solicitors – commercial property
ARM Innovation Hub
RealVNC mid banner careers
Advertisement: EBCam mid banner
Advertisement: Cambridge Network
Mid banner advertisement: BDO
Advertisement: TTP
Advertisement: Innova Systems
Advertisement: Mogrify
Advertisement: RSM
Advertisement: Simpsons Creative
7 April, 2020 - 15:44 By Tony Quested

Back to the future with global demand for Cambridge ‘FluPhone’ app

A Cambridge invention designed to fight Swine Flu in 2009 is in global demand as a potential weapon in the fight against COVID-19.

The sought after innovation is a smartphone app originated in Cambridge University’s Department of Computer Science and Technology to map epidemics.

Co-inventor of the ‘FluPhone’ app –  Professor Jon Crowcroft, the Marconi Professor of Communications Systems at Cambridge – has now had approaches from healthcare professionals and academics in Australia, Canada and countries in eastern and central Europe, wanting to know more about the technology.  

He has also been contacted about it by NHSX, the arm of the NHS tasked with seeing how the latest technologies can be employed to improve health care.

The research app was originally developed by Jon and his colleague Eiko Yoneki, a senior researcher in the department, in the wake of the 2009 global Swine Flu (H1N1) outbreak. 

The aim was to collect smartphone location data and use it to measure social encounters between people and so model and predict how illnesses like Swine Flu were passed between individuals.  

This idea had since evolved from Jon’s and Eiko’s work in mapping human contacts to measure the ways people met in the real world. As part of that, they had built an app employing the short-range radio capacity of mobile phones (which is what Bluetooth headsets use) to detect the presence of other people coming within close proximity of the phone.

John Edmunds, a noted epidemiologist at the London School of Hygiene and Tropical Medicine, heard them talk about their work. His research interests are in designing effective and cost-effective control programmes against infectious diseases and he said it would be very useful to harness the app for measuring epidemics.

They set up a pilot study and ran it for three months in Cambridge in 2010. Several hundred volunteers were recruited and asked to download an app onto their phones that would collect information on their social encounters and anonymously record how often they met other people. “It was a small study,” says Jon, “but it did effectively prove the concept.” 

This is a second coming. In his latest blog post for the Turing Institute, where he is also a Researcher at Large, Jon argues that the COVID-19 pandemic “is an urgent opportunity for scientists, the NHS and other health services around the world to use smartphones and their owners to gather valuable data on an unfolding pandemic. 

“This data,” he adds, “could not only provide crucial intelligence about how this pandemic is evolving, but also help us organise against the onslaught and develop effective, evidence-based strategies for coping with future outbreaks.”

The original FluPhone project aimed at doing both these things. Its goals were to enable epidemic modelling with precision and trace contacts.  
“If used now, it could help us find the parameters of the epidemic and allow us to predict its growth,” Jon says. 

“With the FluPhone application, people would register with their age, gender and contact info, and then we could measure all the encounters they have with other people who have the app. We could then use that data to map the fraction of the population that is infected and how many of them in turn infect other people.” 

He says the app would be particularly beneficial in identifying people who come into contact with an infected person, become infected themselves and subsequently spread the virus to others without knowing that they have done so because they are asymptomatic. 

“Using this app, we could discover them,” he says, “when without it, it’s very difficult to do so.”

Crucially the app could also map the impact of the disease among children. Back in 2010, there were initial concerns – from a university research ethics committee – about recruiting children for the pilot study.

But the levels of user privacy built into the system subsequently reassured them and the researchers were allowed to invite children over the age of 12 to take part. This is key because “children are very important in epidemics,” Jon says. “In influenza outbreaks, for example, it is very commonly kids who spread the infection between households.”

Though they don’t appear to be susceptible to COVID-19, it may well be that in fact they are. “There is already data on COVID-19,” Jon says, “suggesting that as many as half of children who come into contact with an infected person become infected – and infectious – themselves, but without showing any symptoms.”

If we are going to accurately predict the spread of viruses like COVID-19, we need to have more accurate transmission metrics, he argues, than a standard multiplier that suggests that one infected person will, on average, infect two others. 

“So it really matters for us to know if the teacher is going to catch the virus from the children they teach, or whether the children are going to catch it from the teacher and then take it home where others in the household will be infected. Our app would really help with this.”

Another potential use of the app would be in providing data to help judge how effective interventions like social distancing, quarantine and lockdowns might be in preventing the spread of the disease. 

Some Asian countries have been using technological data gathering to track the disease and try to halt its spread. In South Korea, authorities were granted access to residents’ mobile phone records to round up and test the contacts of those displaying symptoms, and quarantine those found to be positive. 

While this was quite invasive of residents’ privacy, it was also an example of an intervention that has been remarkably successful in stopping the spread of the virus.

Singaporean authorities have also been using an app – strikingly similar to the FluPhone variety – to monitor cases of COVID-19. 

“And they were able to do so without being overly intrusive,” Jon says. “When people developed symptoms there, they were asked to self-report – in other words, to send in the records from their app so that the health authorities could track and inform their contacts. They in turn were offered testing to discover if they had the disease.”

Now there is a similar app in the UK, which has been developed by King’s College London. It enables people to report symptoms of COVID-19 so that the progression of the disease can be tracked in real time.

Jon says: “The need for such an app is not over, it is only just beginning. After the peak, which will be in the next three to four weeks, we’ll still need ways to track the illness because the majority of the population will not be immune. 

“So when future cases turn up, we must be able to track their contacts and quarantine those who test positive, otherwise the disease will break out all over again. Apps like FluPhone will have importance for some time to come.”

• Image courtesy – UCL

Newsletter Subscription

Stay informed of the latest news and features