World Food Programme (WFP)


World Food Programme (WFP)

Description of Activities on AI

The WFP Innovation Accelerator is currently supporting several projects that use Artificial Intelligence (AI) and Machine Learning (ML) at the core of their product, including:

Project 1: SKAI

A lack of on-the-ground information at the start of a humanitarian crises is a major obstacle to a quick, effective response. SKAI leverages the power of artificial intelligence and remote sensing to assess damage within 24 hours after disasters take place.

Project 2: Informal Settlement Mapping

COVID-19 is most prevalent in highly-populated urban areas in frontier markets; however, we do not have a high-level understanding of where vulnerable populations are located in urban areas, as WFP usually operates in rural areas.

Project 3: Open-sourced Locust Movement Forecasting Tool

Locust outbreak in East Africa has greatly threatened food security. Therefore, information that provides timely locust forecasting as a preparedness measure is needed.

Project 4: DEEP (Digital Engine for Emergency Photo-analysis)

When disaster strikes, WFP and its humanitarian partners must respond as quickly and efficiently as possible. A lack of on-the-ground accurate information that can be extracted in a low-resource environment can delay emergency response. The solution, DEEP is an open-source machine learning application that works offline, on commercially available laptops to quickly quantify damage in the area. A fundamental part of DEEP is the capacity-building activity for local populations, to enable them to use the app in an emergency situation.

Project 5: Hunger Map Live

Understanding the food security situation is an intense data analysis exercise with information scattered across different data sources and platforms. HungerMapLIVE brings together streams of publicly-available information on food security, nutrition, conflict, weather and a variety of macro-economic data – including from WFP – all in one place to show a holistic picture of the food security situation. Advanced data visualization tools then convert the resulting analysis of food insecurity at the global, country, and sub-national levels, and display it on an interactive dashboard map.

Project 6: MEZA

Nutrition records for millions of malnourished children lie in remote heatlh clinics around the world. This data is difficult to surface for officers designing nutrition interventions remotely. Getting the data to HQ quickly and cheaply could improve the quality of our interventions. As a solution, WFP developed MEZA – an Optical Character Recognition technology to rapidly collect nutrition and related health data from remote, low-resource health clinics, enabling WFP and governments to have the information they need to provide high-quality, context-specific, and timely nutrition support.

Project 7: Omdena Challenge

When disasters strike, needs assessments are conducted by humanitarian experts based on the first information collected, their knowledge, and their experience. However, we do not have access to past data that could help build an innovative AI-driven logistic provision model for emergency in cyclones. We have collaborated with Omdena Challenge (crowd-sourced) to bring 40 AI experts and aspiring data scientists from around the world to build the model.

Project 8: Child Growth Monitor

Timely detection of malnutrition is critical to eliminate preventable child morbidity and mortality. Currently, malnutrition is diagnosed through taking manual measurements of weight, height/length, and mid-upper arm circumference (MUAC). The procedures for taking these anthropometric measurements are complex and require highly trained and skilled personnel as well as expensive equipment. Manual anthropometric measurements are therefore costly and time-consuming. In addition, the accuracy and precision of data are sometimes questionable, due to the limited availability of reliable equipment and human error in measurement taking and recording. At the population level, collection of poor-quality nutrition data during surveys results in inaccurate assessments of national and sub-national nutrition situations. In turn, this leads to unsound decision-making and inappropriate allocation of resources, potentially with devastating implications during emergencies.

Project 9: OPTIMUS

Despite huge efforts of WFP to fight global hunger, there are still millions of people going to bed hungry. Through an optimization of the design of food baskets in various WFP Country Offices, there is a potential to serve more people with the same resources and same nutritional value. WFP Supply Chain Planners work intensively with the various experts (across Programme, Nutrition, Procurement, Logistics, Pipeline, etc.) in the CO to map out the operation, and then used Optimus to evaluate alternative implementation plans. By combining our analytics with their field expertise, we were able to identify several improvements to the operation (e.g. implementing a mixed sourcing strategy depending on the province, and replacing the pulses with a more cost-effective alternative, and changing the mix of commodities received through Title II contributions by the USA), adding up to approx. 6M USD (full-cost recovery basis) projected savings for 2020. The necessary decisions were approved by the CD and are in the process of being implemented.

Project 10: mVAM Chatbot

In emergency and development contexts, communicating with communities is crucial to ensure that people have access to tailored information, and engage in a dialogue that reinforces their capacity to improve their livelihoods and deal with a crisis. Since 2016, WFP’s mVAM has been working to develop and roll out humanitarian chatbots. This technology helps improve communication with populations in hard-to-reach areas by complementing existing communication channels and WFP’s food security monitoring systems.

Related Sustainable Development Goals (SDGs)