Can you imagine if we could accurately predict a looming malnutrition crisis, in the same way, we forecast impending food security problems such as extreme hunger and famine?
CIAT is using big data and machine learning to develop a Nutrition Early Warning System (NEWS) for Africa. What may not be plainly apparent is that the success of NEWS depends on the availability of data, and to be more specific, the availability of open data – data that is freely available for reuse for other purposes, possibly different from the reason that led to the data generation. The availability of open data and open data policies will play an important role stopping malnutrition in Africa.
Sub Saharan Africa has one of the largest child malnutrition problems globally that is persistently increasing with growth in population. For instance, as at 2011, it was estimated that out of the 165 million stunted children under five worldwide, 56 million were in Africa. The child wasting prevalence is equally high in Africa; of the 52 million children under five wasted globally, 13.4 million were from Africa. This stunting and wasting crisis coupled with tropical infectious diseases such as pneumonia, measles, and diarrhea have resulted in increased deaths in the region.
In an attempt to solve this emergency, various malnutrition frameworks have been developed by different actors, using multi-sectoral pathways of variables that explain the diet and health components of nutrition. However, most of these frameworks use formalized mathematical equations with predetermined variables that do not entirely glean all the details from the given datasets.
Currently, data that is generated daily through our devices have been processed through rational systems to develop products that explain and predict various behavioral activities and their outcomes. This form of analysis is broadly known as artificial intelligence and is currently being applied all over the world in solving numerous problems. Examples of this include; data security, personal security, forex trading identification of crop diseases, identification of road routes, purchase of complementary products on online stores, customization of advertisements on various websites within the internet, and much more.
Application of these artificial intelligence algorithms in nutrition through the NEWS project can reveal critical insights from data, without any rule-based programming, and subsequently, predict malnutrition as an outcome of various population predictors. To achieve this, NEWS will train machine learning algorithms using data from existing sources. Lots of relevant data have already been availed in the form of open databases; this includes FAO data, UNICEF data, WHO database, and the WFP – Humanitarian Data Exchange. The volume and variability of these and other datasets coupled with artificial intelligence will be the driving force behind NEWS. Availability of these datasets will also be handy for comparing outcomes between countries and, identifying key variables that can predict malnutrition in each different population.
Based on our experience as working on NEWS, we conclude that; data availability is entirely tied to open access policy of the organization producing the data. The more organizations and countries embrace open access of data and makethese datasets available for re-use, the more productive the predictions from big data in agriculture projects such as NEWS are going to be.
It is clear from a practitioner’s standpoint that a significant contributor towards the success of NEWS will be data availability from open sources. As we celebrate open access week 2017, let us take some time to reflect that Open access to data will contribute to reducing malnutrition in Africa.