E&T Innovation Awards: controlling dengue fever outbreaks

E&T: First of all, could you give us an idea of the scale of the problem? What is dengue fever, how dangerous is it, and where does it occur?

Dr Quillon Harpham: Dengue fever is a serious problem. It’s a viral disease. It’s spread by aedes mosquitoes; aedes mosquitoes are the small ones – so not the same mosquitoes that spread malaria, for example. Dengue gives you very severe flu-like symptoms. So it gives us severe headaches, very high fever, nausea, vomiting and aching all over your body – it is a very unpleasant disease to catch indeed. And one thing that makes it difficult to manage is that there are four distinct serotypes, or strains, of the disease. They’re closely related, but they’re different. And so the first time you catch dengue fever, you will be immune to the serotype that you catch, but the second infection will be a different serotype and that can lead to a much more severe fever, which has a 2.5 per cent mortality rate. So it’s more severe the second time and you’re more likely to die from it.

The global annual cost of controlling dengue fever is about $9bn per year. To put that into a little bit of context, cholera stands at $3bn and gastroenteritis just $2bn so dengue’s $9bn is considerably more than both of those put together. Dengue is also considerably on the rise. In fact, in 1970, there were nine countries that reported cases of dengue fever, maybe 15,000 cases per year. Today, there’s over 140 countries reporting cases of dengue fever, maybe 400 million cases per year. And that’s not just in south-east Asia; it’s in Africa, the Americas, even the eastern Mediterranean and western Pacific regions. Forty thousand people die from dengue fever every year, and about four billion people are at risk of the infection.

E&T: One of the problems, as I understand it and taking Vietnam as an example, is that the outbreaks are occurring apparently at random. Once they have occurred it is too late to tackle them effectively. How important is it to catch an outbreak at an early stage or even forecast it before it does happen?

QH: It’s very important to do that if possible. There’s no specific antiviral treatment for dengue fever. Companies are developing vaccines as quickly as they can, but at the moment vaccination is restricted to seropositive individuals – people who have had the virus or have the antibodies to it. In Vietnam between 2017 and 2019, there were over 200,000 cases reported every year to the Ministry of Health and, of course, that represents considerable suffering, a number of cases of mortality, as well as a huge economic burden – maybe $100m per year at the moment. So preventing dengue fever outbreaks, or at least curtailing them before they get too serious, is certainly better than trying to cure them when they happen.

E&T: If an outbreak has been forecast, or identified early, what steps can be taken to minimise its impact?

QH: Different activities can be done at different lead times. Immediately before a potential outbreak, maybe a week before [and] up to a couple of months before, the community can take action. They can cover their water containers; a lot of people in these countries store water in water butts, but that’s where mosquitoes can breed and thrive, so they cover those. [They can] reduce the standing water or puddles that are around; they can sleep under nets or wear the right clothing to prevent getting bitten.

A little bit further into the future, maybe two weeks to two months, the local Ministry of Health can take action. They can spray insecticide in places where they know mosquitoes are and they can issue community alerts to the communities to take action.

And then maybe five to six months in advance, the National Ministry of Health can do budget planning and allocation. So areas that are forecast to be more severe this year can have a higher budget allocation for dealing with it in advance.

E&T: This brings us to D-Moss. Perhaps you could start by telling us how the project came about and who are the partners within it. It is obviously more than just HR Wallingford?

QH: HR Wallingford itself does research and consultancy into water in the water cycle. Anything to do with water, we do. And we realise that the water cycle touches a lot of areas of life. Some years ago we realised that the water cycle has connections to mosquitoes and mosquito-borne diseases since they thrive in standing water. So the more water there is in the system, the more likelihood there is of mosquito-borne diseases – that’s the theory.

We began discussions with other related parties and that led to dengue fever, because dengue fever is on the rise so much and because it’s classed these days as a neglected tropical disease. In spite of its impact, it doesn’t get much attention, so we formed a team and applied for funding from the UK Space Agency’s International Partnership Programme, and the team consists of UK partners and local partners. In the UK, for example, we work with the London School of Hygiene and Tropical Medicine, the UK Met Office, and Oxford Policy Management.

The Vietnam implementation of D-Moss includes the World Health Organisation and the United Nations Development Programme, the government ministries of General Department of Preventative Medicine, the Pasteur Research Institutes and the Institutes of Hygiene and Epidemiology, and the Institute of Meteorology, Hydrology and Climate Change.

The Malaysia implementation includes the International Medical University and the Institute for Medical Research, and in Sri Lanka, we’re working directly with the Ministry of Health there, so there’s always a UK and a local component to any D-Moss implementation.

E&T: That’s a big team you’ve got there. Could describe what D-Moss actually is? What technologies do you use?

QH: D-Moss stands for the ‘Dengue forecasting MOdel Satellite-based System’. So there are a few clues there in the name. It’s the first functional dengue fever early warning system of its kind. It uses satellite Earth-observation data, in-situ observations and seasonal climate forecasts, and it issues regular forecasts of dengue fever, up to six months in advance. It’s based on one of the 10 years of research... now maybe 15-20 years of research... from ourselves at HR Wallingford looking at informatics and surface water availability modelling.
The London School of Hygiene and Tropical Medicine has been researching into statistical epidemiology and the UK Met Office has done a lot of research over the years into seasonal forecasting. These three research strands have come together to give us D-Moss.

E&T: So it’s not just a satellite data, it’s very much a local project as well?

QH: Local in terms of stakeholder engagement, yes, that’s extremely important. And also we use some on-the-ground information. We use recent recorded cases of dengue fever from the Ministry of Health. And if in-situ device data exists, that measures things like hydro-meteorological parameters, such as air temperature or precipitation, then we use that as well.

E&T: When D-Moss is up and running, who actually operates it and where? Is it HR Wallingford in the UK, or is it by local teams in the countries that are affected?

QH: D-Moss is almost completely automatic, so there are data harvesters, data scalars, standardisers, formatters, and databases that all run automatically. They collect data from the satellite services when they’re available.

The models to produce the forecasts all run automatically and update the user interface automatically. The whole system is hosted on the cloud and accessed by users via a private website. So HR Wallingford provides the technical support, and ministries of health provide the latest dengue cases, which is a semi-automatic update. We then format it, add it to the system, and that triggers a new forecast.

E&T: How are you judging the success of the project? Do you have details to demonstrate how D-Moss has helped in the fight against dengue fever?

QH: Yes, that’s been really important to us. Anybody can say ‘here’s a forecast of dengue fever’. But it’s really important to us to make sure that it’s actually working and adding value. We’ve done detailed accuracy assessments of the forecast to make sure that the forecast is accurate. So we understand where it’s more accurate, when it’s more accurate, and how accuracy deteriorates as we go forward in time and the forecast stretches further into the future. We know it is extremely accurate for the first few months and then gradually deteriorates, but it’s still value added right up to six months.

Vietnamese agencies have been using it now for several years as part of their budget planning and intervention and processes, and it is changing Vietnam’s reactive approach to think about prevention and control to something far more proactive so they can deal with dengue in advance.

We have user testimonies and collected a lot of quotes about how users have been using it – they’ve been using it to prevention‑control dengue fever in advance, before the outbreaks happen. We’ve done an extensive anonymous survey of the local users, and they estimated that long-term usage of D-Moss would reduce dengue fever by 15 per cent and reduce the amount of control resources by 15 per cent. The cost-benefit ratio that we’ve calculated for the operational system is quite overwhelming – it’s between 30-1 and 222-1 – so there’s a huge cost-benefit case for the system, far in excess of the usual investment thresholds.

We also get occasional reports from the users – Malaysian users reported to us last year that following the forecast of an outbreak, they were able to take action and reduce the number of cases.

E&T: Fantastic results. What have been the standout achievements of D-Moss from a technical point of view, the things that would have impressed the judges so much at the end of last year?

QH: I’m tempted to ask you how long have you got! D-Moss is absolutely steeped in innovation from those three research strands that I mentioned earlier. A lot seems to have come together to give us the system, so I’ll list a few as they come to me.

We discovered new links between parameters describing surface water and dengue fever. It’s been known for some time that meteorological conditions such as temperature, or precipitation, are related to dengue fever, but we discovered that surface-water parameters, things like soil moisture or run-off, relate to the fever as well.

The modelling approach was used on the system as a statistical super-ensemble – and it incorporates a whole variety of different parameters – so the surface water parameters I mentioned earlier, as well as meteorological parameters, even things like El Nino in the Pacific Ocean, have been shown to affect dengue fever in south-east Asia. All of these go into this statistical super-ensemble model and the whole system works as a machine-learning algorithm. So it learns as it gets more data; it recalibrates itself as it gets more data and so, as it goes on, it gets more and more accurate all the time.

The whole system was designed to be able to be rapidly redeployed and easily scaled. It’s just a matter of configuring out-of-the-box components to reimplemented somewhere else, because of a lot of informatics and standards research that we’ve done and built into the system.

We’ve also introduced new techniques for post-processing Earth-observation data, so gap filling, ground truth, and with in-situ data bias correcting forecasts. Forecasters know that their forecasts have biases in certain places at certain times of the year. These have been corrected as well.

And I suppose last but not least, D-Moss produces a water availability forecast as a by-product. So we forecast water availability first, and then the water-availability forecast goes into the dengue fever forecast. But the water-availability forecast, of course, is of use in its own right – it can be used by water resource managers for climate change adaptation measures in traditional water-management sets as well.

E&T: You must have been very pleased with your success at the E&T Innovation Awards last year. Does such reward and recognition help with the work that you’re doing?

QH: Absolutely. We were delighted with our success in these awards. It’s really important for promotion and profile raising. We haven’t been shy to tell people about our success on social media and in interviews. It’s absolutely fantastic and we are delighted.

E&T: It was probably equally nice for internal recognition for the team that worked on the project.

QH: Absolutely. I mean, we do enjoy the awards. It's a lot of fun to put together; it’s a lot of fun to go through and attend the awards themselves. And it’s been, I think, a real career highlight for many of the people in the team and will go on all of our CVs!