Esther Mulu, a 69-year-old mother and grandmother from Kitui County, Kenya, has endured the hardships of food shortages for years. She relies on her small farm to feed her large family of seven children and 25 grandchildren, but her livelihood has been increasingly threatened by irregular rainfall and persistent crop failures.
Mulu’s situation is far from unique. Kitui County, along with five other regions in Kenya, is known for its frequent droughts, as reported by the African Medical and Research Foundation (AMREF). Mulu has long depended on food aid to sustain her family, but this aid has become less predictable, leaving her with fewer options for survival.
“We used to get food donations, but now they’re no longer coming,” Mulu said. “We are facing a severe drought and I don’t know where we will get food. We now eat one meal a day, mostly porridge, and wait for a larger meal that may come the following day.” Mulu, who remains hopeful for a solution to the widespread hunger crisis, also expressed her wish that technology could be developed to help end the cycle of food insecurity.
Fortunately, researchers from the University of California, with support from Microsoft, are working on an innovative AI tool aimed at helping communities like Mulu’s prepare for crop failures and malnutrition. This tool, which uses artificial intelligence to predict areas at risk for food insecurity, could be a game-changer in the fight against hunger.
The 2023 drought has wreaked havoc across the Horn of Africa, and Kenya has been particularly affected. According to the UN’s World Food Programme, an estimated 4.4 million people in Kenya are experiencing “acute food insecurity,” with roughly 1.1 million women and children facing acute malnutrition.
The new AI tool is designed to combine various data sources, including weather patterns, NASA satellite imagery of vegetation, and clinical health records from the past decade, to create computational models that predict where food insecurity will likely occur. These predictions could be used to allocate resources and interventions before hunger and malnutrition worsen.
Since its trial run in Kenya began in February of the previous year, the AI tool has proven to be effective, with accurate predictions for up to 6 months ahead, although specific figures for its success are still being compiled. Samuel Mburu from AMREF is among the experts helping to develop the software and shared how the project works.
“We’ve looked at historical data from multiple sources over the last 10 years and used that data to forecast malnutrition cases at the sub-county level,” Mburu explained. “This allows us to allocate resources in advance, so we can ensure children under five get the nutrition they need, especially in health facilities.”
The goal is to help communities better prepare for food shortages, reducing the negative impacts of droughts and crop failures. Although the tool has been focused on the six counties most prone to drought, AMREF now aims to expand it to cover the entire country. This expansion is crucial since many individuals in remote areas lack access to clinical services, and health records alone may not offer a complete picture of the situation.
In the future, AMREF hopes to incorporate community-level data into the tool for even greater accuracy. Local reports on food availability and malnutrition will be added to the AI’s calculations, providing a clearer view of food insecurity at the grassroots level.
“We now have the ability to process large volumes of data to create predictive models that bring real value,” said Mburu. “By predicting malnutrition cases, we can mobilize resources, both internally and externally, and address the problem before it becomes a crisis. However, the main challenge we face is the cost, especially with cloud computing infrastructure. Thankfully, we have received generous sponsorship from Microsoft to help make this possible.”
The AI technology not only helps predict food insecurity but also assists local governments and county officials in preparing for and responding to malnutrition crises. By predicting the number of cases in advance, officials can allocate food aid, medical supplies, and other resources in time to address the issue.
“With this technology, we can predict food insecurity 3 to 6 months in advance and make sure that we mobilize resources to meet the needs of vulnerable communities,” Mburu added. “The goal is to help improve the lives of families like Esther Mulu’s, ensuring they have enough food and nutrition during difficult times.”
As Mulu continues to face daily uncertainty, she remains hopeful that the AI tool will help prevent future generations from enduring the same struggles. She believes that technology has the potential to save lives and eradicate hunger in Kenya.
“If you can develop a technology to end hunger, I would be forever grateful,” Mulu said. “That would be a gift to all of us suffering from this hunger crisis.”
