Hannah Fry is a Trustee of the Science Museum Group, broadcaster, and an associate professor in the mathematics of cities at University College London. Her edited answers are in italic to distinguish them from my commentary.
what got you interested in pandemics?
We knew that it was not a matter of if a devastating pandemic would ever happen, but when. We knew, in the absence of correct planning and counter-measures, the likelihood of it spreading and killing millions of people around the world was very real.
As discussed in an earlier blog, the pandemic threat has been clear for some time, and is on the UK National Risk Register. In 2018 the World Health Organisation even characterised one hypothetical pandemic candidate, Disease X, as ‘a novel coronavirus other than SARS or MERS‘.
how do you simulate a pandemic?
With an expert team, I conducted a citizen science experiment for a BBC Four documentary, Contagion! We wanted to find out exactly how a real pandemic might unfold using an app built for mobile phones to ‘infect’ users with a virtual virus, which could track movements to within two metres.
The premise was simple: we’d infect with a ‘virtual virus’ the good people of Haslemere, a lovely town off the A3 in Surrey, and wait and see what happened. Hundreds of people across the town downloaded our app so they could ‘catch’ the virus if there was an infected person nearby.
With me as the first to be infected – ‘patient zero’, the term used to describe the person identified as the first carrier of a contagion – I wandered around Haslemere.
I went to a yoga class, did a little shopping, had lunch in a cafe, and ended my day with a pint in the local pub. I acted as if I had no symptoms and, of course, we now know that with COVID-19 you can shed virus before symptoms are apparent.
Using the app, we could track where participants went and who they interacted with over the course of a few days, allowing us to view how quickly a potentially deadly virus spreads and who the most dangerous carriers were.
Her team selected Haslemere for good reason: a pandemic would most likely originate from overseas and Haslemere is no more than an hour from Britain’s two biggest airports, less than an hour’s drive from the ports of Portsmouth and Southampton, and has regular commuter services to London.
As it turned out, the first person to contract COVID-19 in the UK was a patient at Haslemere Health Centre.
what did your app reveal about the pandemic?
It started slowly, slowly, slowly, then hit you like a truck – that’s exponential growth. I infected nine people in eight hours of mooching. In turn, they spread the virtual virus to 69 more. After three days, more than 500 people were infected across town.
There are always ‘super-spreaders’, people who – through dint of their job, or lifestyle, or perhaps even genetic makeup – would be more dangerous during a pandemic than the average person.
In Haslemere, our biggest super-spreader was a lady who worked in a hardware shop, but it could anybody who has close contact every day with high numbers of people.
what has your discipline – mathematics – got to do with pandemics?
There’s a mathematical angle to almost anything whatsoever, for instance modelling social networks for epidemiology, the science dealing with the spread and control of diseases. Mathematical models are the only tools we have for peering into the future. They’re unavoidably flawed, and shouldn’t ever be taken too literally – but without them you’re feeling your way in the dark.
The use of mathematics to understand disease dates back centuries. In 1766, the Swiss mathematician and physicist Daniel Bernoulli, working on a mathematical analysis of the value of smallpox inoculation, remarked: ‘in a matter which so closely concerns the well-being of the human race, no decision shall be made without all knowledge which a little analysis and calculation can provide’.
When it comes to social networks, the evocative suggestion that we’re all interlinked through mutual acquaintances – to people living in the COVID-19 epicentre in Wuhan in China – is often credited to the Hungarian author and playwright Frigyes Karinthy. Although perhaps the best-known study was by American social psychologist Stanley Milgram in the 1960s, which suggested that there tended to be six people on average linking one person with any another— six degrees of separation.
The British epidemiologist Roy Anderson and Australian biologist Bob May, the recently-deceased former President of the Royal Society, did pioneering work on the spread of epidemics through social networks in the late 1970s and helped to make ‘mathematical epidemiology’ a field of biology.
Mathematical models are indeed powerful predictive tools, for instance, to show the role of worldwide air travel in the spread of COVID-19.
Even so, there are fundamental limits to what can be done with models on digital computers (not least, as pointed out in an earlier blog, ‘garbage in, garbage out.)
Our network of contacts, of course, is critical to how the virus spreads between us – something we set out to see play out in our experiment in Haslemere.
how did you use mathematics to extend your haslemere work?
We persuaded people around the country to download a second app, using GPS to locate people down to a square kilometre, so that Julia Gog and her team at the University of Cambridge could model how that virus might spread, not just showing where people are but how many others they come into contact with, which gave much richer insights than earlier efforts.
Our simulation of the events following patient zero predicted that 43 million people in the UK could be infected in a pandemic, and with up to 886,000 of those infected expected to be fatalities.
is this public participation experiment unusual?
Yes, in terms of the size of the study, with 40,000 participants, and the impact of the data. The big impact was that we had that data waiting in the wings, as it were, for when a real pandemic hit to show how people would behave.
No, in the sense that public participation in science goes back three centuries, and is as old as the scientific method. More recently, the BBC did many mass participation experiments with The Daily Telegraph that started in the mid-90s (Megalab and Live Lab).
what does your research tell us about covid-19?
Using the BBC Pandemic data gathered by Hannah Fry and colleagues on more than 40,000 participants a simulation of the impact of a range of different testing, isolation, tracing and physical distancing scenarios was carried out with Adam Kucharski, John Edmunds and colleagues at the London School of Hygiene and Tropical Medicine.
They concluded that if the Government could persuade 75 per cent of smartphone users to download an app to trace contacts, which is being tested on NHS and council workers on the Isle of Wight, it would still be necessary to have social distancing measures to stop the pandemic returning in a second wave when lockdown is lifted.
Even if a massive three quarters of us use the app, lifting the lockdown without any social distancing measures (such as limiting gatherings, would mean the virus’s ‘reproduction number’ could reach 1.4. (The number has to be below one for the epidemic to subside). In other words, the app is not the whole solution, just one tool among many.
The NHS app was developed after scientists concluded digital contact tracing was necessary to manage a pandemic. Rather than use GPS to locate people, the NHS app relies on Bluetooth to determine when your phone is near another phone running the app. It then uses notifications to alert you if you have been near someone with symptoms of the virus and also collects data about how the user is feeling.
The NHSX app is quite like the Singapore and Australian apps, in that it is set up to provide central understanding of which phones have come into contact with which others. In other countries, a decentralised version is used where no one knows all the contacts, only local contacts.
I am doubtful enough people will download the NHSX app for it to be helpful but on the plus side, with a centralised system, you can do more clever things.
The Joint Committee on Human Rights commented that, because of the unprecedented data gathering, there are ‘significant concerns regarding surveillance and the impact on other human rights which must be addressed.’
The Ada Lovelace Institute, an independent institute dedicated to ensuring that data and AI work for society, recommends the introduction of primary legislation to regulate data processing for digital contact tracing and to impose strict limitations, adding that premature deployment of ineffective apps could undermine public confidence.
is the nhs app private and secure?
Update, 18 June 2020: the UK government announced that ‘we will now be taking forward a solution that brings together the work on our app and the Google/Apple solution.’ Both approaches are described below.
Yes, according to Ian Levy, technical director of the National Cyber Security Centre, adding that it is designed to strike a balance between individual, group and national privacy.
The app developed by NHSX detects and records when you’re near other app users and can later reveal if you if you’ve been in ‘high-risk contact’ with someone who has symptoms.
Rather than adopt a decentralised model’ when the public health authority – by design – knows almost nothing about ill people, the NHS went for a centralised model, where an ill user reports their symptoms and discloses their anonymous contacts to the public health authority.
Using this anonymous data, the health authority can use risk modelling to decide which contacts are most at risk, and then notify them to take some action, such as self-isolation to start with. Or the health authority could discover that a particular anonymous person seems to be a super-spreader.
As well as providing a clear picture of which anonymous app users have been in close contact with each other (to understand how the virus is spreading and where hotspots might be arising) a centralised system offers flexibility in how to notify users, NHSX says it ‘will help us give the most accurate advice and to better spot false alarms and unnecessary alerts asking people to self-isolate.’
how does the nhs app work?
When you run the app, your phone is identified by a large random number. Using Bluetooth, it does the digital equivalent of a handshake when another app user comes close enough. The apps sample and record the Bluetooth signal strength, as a proxy for distance, along with duration and time. Then, if you wake up with a fever, the public health authority can work out the devices that have been in ‘risky contact’ with you.
A notification system asks them to self-isolate for a couple of days until the results of tests are known or too few of your contacts report symptoms to statistically suggest that you were probably infectious. There’s more detail in this paper.
‘Digital contact tracing is new and no-one’s done it at scale before,’ said Ian Levy. ‘Using Bluetooth to measure distance in the real world hasn’t been done at this scale before. Interoperating between massive numbers of different devices in ways that weren’t originally conceived hasn’t been done before. And we’ve not battled a pandemic like this before.’
what can we learn from south korea’s digital detectives?
South Korea is a “fine example to us and we should emulate what they achieved”, Deputy chief scientific adviser Professor Dame Angela McLean said at a UK government briefing last week.
With the help of Channel 4’s ‘The Country that Beat the Virus: What can Britain Learn?’ programme (broadcast this Wednesday at 9pm), I talked to Dr Alice Tan, an internal medicine specialist at the MizMedi Women’s Hospital.
She told me how South Korea ‘crushed the curve’ when protecting its population of 52 million people from COVID-19 (the UK population is 67 million).
Previously, Dr Tan was a leading clinician at the Samsung Medical Centre in Seoul, the ‘ground zero’ for the 2015 outbreak of another kind of coronavirus, called Middle East respiratory syndrome (MERS) which, like COVID-19, likely originated from bats.
Her edited answers are in italic to distinguish them from my commentary.
We had a huge super-spreader event in our hospital during MERS when one patient, a big gentleman with a very forceful cough, infected 82 other patients within a short period. The hospital had to shut down for a month. It was horrible and we learnt many lessons. I have a very clear memory of this, and it was a game-changer for the country.
That experience of MERS taught South Korea the importance of rapid testing, prevention and control. They even conducted a tabletop pandemic testing exercise last December. South Korea was well prepared.
The KCDC, the Korean Centres for Disease Control, raised the alert to level one on January 3rd, just a few days after the first cluster of cases of pneumonia of unknown cause in China. Quarantine and fever detection methods – thermal cameras – were enhanced at airports right away. That’s actually how South Korea discovered our first case, on January 19.
After the diagnosis of our fourth case, the alert was upgraded to level 3 on January 28. By January 31 testing (direct testing for the virus, using a technology based on PCR) was available in 18 validated labs throughout Korea.
By then we had only 16 cases but we were already at level three, tracking and tracing very aggressively. For every positive case that was diagnosed, hundreds of close contacts were isolated at the time, and they were monitored closely.
(Today there are more than 638 testing centres nationwide, conducting tests in 6-8 hours, and delivering the results in 24 hours. South Korea also introduced the world’s first drive-through and walk through testing facilities. Cases have dropped, so no more than 4000 tests are now being done per day of the daily capacity of 20,000. Last week, social distancing was eased).
when did the epidemic start to get out of control?
We did not suspect transmission among people in the community until ‘Patient 31’, a 61-year-old woman. She had been in a car accident and was admitted on 7 February to a hospital in Daegu – a city of 2.5 million people in the South. There she developed a fever on February 10.
In the interim, on February 16, even though she had pneumonia, she visited the Shincheonji Church of Jesus, which I had never heard of at the time. The next day she was tested and the day after diagnosed for COVID-19.
We are still not sure if she is the index case (the origin of this cluster) but she had visited the church three times, and she was probably contagious for all three visits.
In the church services we estimated she had come into contact with over 1,000 people by the time that she was diagnosed. We then knew the outbreak was going to be massive. I was shocked to be honest, because for a few days the week prior, there were actually no new cases. It was a depressing day for the whole country.
A huge outbreak in Daegu, was unleashed. Our highest nationwide alert – level four – was announced a few days later, on February 23. Cases in Daegu climbed from one to 6000 in a single month.
how did you bring covid-19 under control?
South Korea’s containment strategy involved tracking down everyone a patient had spent time with since they were infected by using GPS coordinates, which mobile phone companies are required to divulge along with credit card details and CCTV footage, so an app was not necessary.
Close contacts, which are classed as people within a given distance for a given time, were quarantined at home or, if with mild symptoms, in specialised centres. All their contacts were immediately tested. If they tested positive the process was repeated.
Text alerts were sent to the phones of people who lived close by and people could check for themselves if they’d come into contact with someone who’d been infected. Non-compliance had legal consequences, from fines to imprisonment.
An app was, however, developed for incoming travellers in quarantine to log their symptoms, and to manage the large number of quarantined members of the Shincheonji church.
February 29 saw a peak of 909 cases. In the days that followed, when we looked at the distribution of new cases throughout the country, we could see that the fire was still burning, so to say, in some provinces.
However, there were other regions and provinces that recorded no new cases, sometimes a few days in a row, and that was encouraging.
The first week of March is when we started to turn the corner. I remember sending a message to all my Facebook friends: ‘Hey guys, it’s (social distancing) working.’
is the worst over in south korea?
We’re going to see a prolonged war, not a quick victory. South Korea warned of a second wave of the new coronavirus yesterday as infections rebounded to a one-month high, just as the authorities were starting to ease restrictions. “It’s not over until it’s over,” President Moon Jae-in told the nation in an address in which he vowed to turn the crisis into an opportunity.
The Korea Centers for Disease Control and Prevention reported 34 new infections, the highest since April 9. The outbreak centred around nightclubs in the Itaewon district of Seoul visited by a 29-year-old man before testing positive. Officials are tracking down more than 7000 people who visited the clubs, and the mayor of Seoul, Park Won-soon, has ordered the capital’s bars and clubs to shut.
Many people who are contagious feel so great that they can engage in all sorts of “normal” activities such as club-hopping until 04:30 in the morning. There are 5,726 tests that are pending results, and the yield has been less than 1% positive, so I expect another 50 or so positive cases in the near future.
There have been no fatalities in positive cases under 30, and the club-going crowd is generally young, so other than the additional strain on our already exhausted medical workers and the loss of income of the owners of these establishments, the impact hopefully will not be so big. But I am a little nervous, and I stayed at home voluntarily all day yesterday.
some argue the uk has a relatively high death rate because london is a major conurbation
Seoul has a population of around 12 million people and we ended up with 669 COVID-19 patients and two fatalities because we managed it well. It is not about the size of the city.
COVID-19 has two faces: a meek face, where there are symptom-free and mild cases, and a horrific face in the elderly population, killing one-quarter of those over 80.
COVID-19 will sneak into a community. A person might not know they are sick and yet there are high levels of viral shedding two days before symptoms, unlike other respiratory illnesses. It spreads quietly, and if you are not monitoring with aggressive testing, it goes from one to several thousand cases in a few weeks because it is so transmissible. Then you get infections in high-risk populations, such as nursing homes.
the korean system requires an invasion of privacy – is this too high a price?
There’s always a balance to strike between personal rights and the greater public good. Technology can enhance traditional tracking and tracing, which can be slow and incomplete if you rely on a patient’s memory alone. Using their data, we can now analyse the movements of a positive person within ten minutes.
When it comes to COVID-19, lives are at stake and time is of the essence. We had to get this under control and save as many lives as we could, subject to our Prevention of Contagious Disease Act. Compared to what’s going on in other countries, like the UK, we realise our sacrifices have been small compared to what could have happened.
what other covid apps are there?
Daily symptoms have been logged by three and a half million users of the COVID Symptom Study app, a mobile application launched in March. This particular kind of app could help to predict geographical hotspots of COVID-19 incidences up to a week in advance of official public health reports.
The app enables users to self-report data on COVID-19 exposure and infection. It was developed by the Coronavirus Pandemic Epidemiology Consortium, a multinational collaboration, and it found that tests for the disease were often predicted by combinations of three or more symptoms, including fatigue and cough, followed by diarrhoea, fever, and loss of smell.
Using data from users in Southern Wales, they successfully predicted two spikes in the number of confirmed COVID-19 cases in advance of public health authorities.
Another study by the same team found that of the 18,401 who had undergone a SARS-CoV-2 test, the proportion of participants who reported a loss of smell and taste was three times higher in those with a positive test result than in those with a negative test result.
The results suggest that loss of sense of smell and taste should be included as part of routine screening for COVID-19 and added to the symptom list of the World Health Organisation.
Lead researcher Tim Spector from King’s College London, said: “Accurate real-time data is essential if we are to beat this disease. The more people we can get logging their symptoms on the app, the quicker we will be able to really understand this disease. We would like to thank every single person who is already participating, and would urge everyone else to download the app and check in every day, whether you are experiencing any symptoms or feeling fine.”
are there any more covid-19 apps?
Apple and Google are joining forces on decentralised COVID-19 contact tracing technology and, because the NHSX centralised app has to use different ‘keepalive’ strategies to run in the background on Android and iOS devices, might still be an option for the UK.
what is the state of the pandemic?
The latest picture of how far the pandemic has spread can be seen on the Johns Hopkins Coronavirus Resource Center or Robert Koch-Institute. You can check the number of UK COVID-19 lab-confirmed cases and deaths, and at the Office of National Statistics.
The Science Museum Group is collecting objects and ephemera to document this public health emergency for future generations.
The Country that Beat the Virus: What can Britain Learn? is on Channel 4 on Wednesday 13 May at 9pm, and then available on All 4.