Today, the 28th of February, marks the international day of rare (aka orphan) diseases. I would like to take the opportunity and discuss big data epidemiology. That is, how the big data generated by various providers, such as telcos, can be used to prevent (not so rare) diseases such as the influenza.
When we feel ill, we stay at home and/or we visit a hospital. This common phenomenon can be used to research the spread of a disease in real time. How?
Current modern technology has enabled us to create so-called „movement patterns“. You can imagine such a pattern on the example of a smaller city residential area: Common population composition in this area during a business day has certain characteristics (age, sex, mobile behaviour, media behaviour, energy consumption, etc.).
There is a limitless volume of data that can describe or characterise a certain area. These can be mobile phones, sensors in intelligent home appliances and a heating system, IPTV (internet protocol TV) boxes, payment terminals in shops and so on. Technological advancements in recent years have enabled more effective work with such data which, because of its sheer volume, made any use practically impossible in the past.
The current speed of technological development will soon enable us to monitor deviations within the aforementioned characteristics of locations and subsequently extrapolate this information to help identify the cause of such deviations. The respective cause could for example be the spread of an influenza epidemic.
Understanding the spread can mean possibilities of its prevention
Another model is created at such a moment – a disease spread model. Such a model provides key information on possibilities how to curb the current disease and how to proceed with disease prevention in the future. Additionally such a model provides a better understanding if an uncommon disease is concerned.
Procedures for research and creation of disease spread models are, of course, an older issue. However, there is a substantial problem – the price & time needed to develop these models. Research and creation of such a model in a “classic“ way present a very expensive procedure, which means that these are only performed in case of the most serious diseases and often when these diseases are a burning issue.
The simplicity of research of a disease‘s spread by means of the population analysis is several times simpler and faster. All that has to be done in this respect is to analyse, understand and reproduce the data which are already available. In this respect big data epidemiology has much to offer. It will be up to those who generate this data to put it to good use. It will also be up to regulators to create an ideal balance between making sure the privacy is protected & the general good that comes from the research possibilities big data provides. But it will also be up to the general public to encourage those who analyse this data to use it well.
Image: Incidence of flu-like syndromes in France (www.openhealth.fr)