Big Data can help curb the spread of chikungunya

Chikungunya is breaking out fast across Dhaka. According to the Institute of Epidemiology, Disease Control and Research, about 3,000 cases have been reported during the past two and a half months. It has no known cure, and all that governments can do is treat the symptoms and help people eradicate mosquito breeding grounds. But this happens only after the virus has begun its work.

One of the preventive solutions to the mosquito-borne virus lies in using big data. With the right information and applications, risk maps and disease outbreak forecasts can be developed. These can alert officials to the next possible outbreak and highlight zones under risk of infection.

LIRNEasia, an ICT policy and regulation think-tank working across South Asia, is already doing similar work, related to dengue, in Sri Lanka.

The main mosquito vectors of diseases such as dengue and chikungunya, aedes aegypti and aedes albopictus, travel very short distances. It is often humans who carry these diseases from one region to another, spreading them across the country. Predicting the spread of these diseases, therefore, depends heavily on the ability to predict human mobility – the patterns, causes and effects of people's movements (from work to home, etc.).

Until recently, such movement patterns were available only from census data and sample survey data, which are costly to gather and are nearly out-dated once collated. Advancements in big data technology and methods, however, allow the use of alternate information sources to develop risk models at a very high level of spatial and temporal detail. LIRNEasia is currently using mobile network big data (MNBD) in Sri Lanka, to build computer-generated models that can forecast dengue outbreaks in the country.

A Call Detail Record (CDR) is generated by the network provider whenever a subscriber makes or receives a call. This CDR provides the general location of a mobile subscriber at the time of the call. A whole series of CDRs for a single subscriber, taken across a large timespan, provides approximate movement patterns for that subscriber. By aggregating the movement patterns of millions of mobile subscribers, LIRNEasia identifies regions which act as central 'hubs' in terms of human mobility. (To protect subscribers' privacy, network operators remove all personally identifiable information from CDRs.)

The CDR data are then integrated with data on actual incidence of dengue, temperature and rainfall data, and the natural vegetation index. LIRNEasia's analyses show that human mobility is highly correlated with dengue incidences (see Fig.1) which are demonstrated in their spatial risk maps (see Fig.2) for the disease.

CDR data analysis is not without its limitations: for example, the entire population cannot be expected to have mobile phones. Nevertheless, high penetrations of mobile phones in both urban and rural areas of Sri Lanka ensure that CDR data is still a rich source of information. The research, conducted in collaboration with the University of Moratuwa and the Epidemiology Unit of Sri Lanka's Ministry of Health, has gained considerable international attention ('Impact of Human Mobility on Spread of Dengue in Sri Lanka'. NetMob Book of Oral Abstracts, 2017).

With this kind of predictive information, decision-makers are able to allocate prevention and treatment resources to regions that are most at risk. If handled effectively, timely action based on these predictions can help reduce the number of people affected by the virus.

The same methodology used to predict dengue in Sri Lanka can easily be applied to chikungunya in Bangladesh. Chikungunya spreads in the same way dengue does – through mosquitoes and human movement – and even has a similar incubation period. In addition, Bangladesh shares many characteristics with Sri Lanka, including mobile phone penetration statistics.

The only difference is that dengue is endemic in Sri Lanka while chikungunya is not yet so, in Bangladesh. According to research published in Science magazine ('Quantifying the impact of human mobility on malaria', 2012), the role of human mobility seems to be more pronounced in propagating an infectious disease to a region where it is not yet endemic. Humans have a more important role here, and studying human movement might be even more critical for chikungunya in Bangladesh.

LIRNEasia currently collaborates with the University of Dhaka's Data and Design Lab led by Dr Moinul Islam Zaber. Researchers from similar organizations can bring the power of big data analysis to bear on how the country responds to diseases such as chikungunya. With the right data and timely application, decision-makers and the public can battle this outbreak before it becomes an epidemic.