Scientists propose machine learning approach to solve world food insecurity

40% (<40%) when the observed prevalence is <40٪ (> 40%) as being overestimated (underestimated). Other areas are rated as low and overestimated. The solid black line indicates where the points would fall if all predicted values ​​exactly match the observed values, and the gray dashed slashes indicate a distance of ±5 spread points from it. Dashed horizontal and vertical lines in gray indicate a prevalence limit of 40%. credit: Nature Food (2022). DOI: 10.1038 / s43016-022-00587-8″ width=”800″ height=”403″/>

Machine learning can guide food security efforts when raw data is not available. Predictions that differ from the observed value by a maximum of ±5 points spread are classified as correct. An expected prevalence >40% (40%) is classified as overestimated (underestimated). Other areas are rated as low and overestimated. The solid black line indicates where the points would fall if all predicted values ​​exactly match the observed values, and the gray dashed slashes indicate a distance of ±5 spread points from it. Dashed horizontal and vertical lines in gray indicate a prevalence limit of 40%. attributed to him: Nature Food (2022). DOI: 10.1038 / s43016-022-00587-8

Researchers in a recent paper published Nature Food They propose a method that they claim will allow decision makers to make timely and informed decisions about policies and programs geared towards fighting hunger.

In 2021, 193 million people in 53 countries were acutely food insecure. This number has been steadily increasing over the past few years also as a result of the COVID-19 pandemic. To address this global issue, monitoring the situation and its evolution is key.

Governments and humanitarian organizations regularly conduct food security assessments through face-to-face and remote mobile phone surveys. However, these methods have high costs in both monetary and human resources, and therefore preliminary data on the food security situation is not always available for all affected areas. However, this information is essential for governments and humanitarian organizations.

To address this issue, researchers Nature Food The paper proposes a machine learning approach to predict the number of people with insufficient food consumption when up-to-date direct measurements are not available. Associate Professor Elisa Umudi (Department of Network and Data Science, CEU, Vienna) says: “We also propose a method for identifying the variables that drive the observed changes in expected trends, which is key to making predictions useful to decision makers.”.

The proposed method uses a machine learning algorithm to estimate the current state of food insecurity in a given region from data on the main drivers of food insecurity: conflict, extreme weather and economic shocks. The results show that the proposed methodology can explain up to 81% of the variance in inadequate food consumption.

The researchers claim that their approach opens the door to achieving near-real-time food security on a global scale, allowing decision makers to make more informed and timely decisions on policies and programs geared towards fighting hunger, in an effort to achieve Sustainable Development Goals 2 of the 2030 Agenda for Sustainable Development. .


Hunger is increasing worldwide, but women bear the brunt of food insecurity


more information:
Giulia Martini et al., Machine learning can guide food security efforts when raw data is not available, Nature Food (2022). DOI: 10.1038 / s43016-022-00587-8

Presented by Central European University

the quote: Scientists propose machine learning approach to solve world food insecurity (2022, September 19) Retrieved September 19, 2022 from https://phys.org/news/2022-09-scientists-machine-approach-world-food. html

This document is subject to copyright. Notwithstanding any fair dealing for the purpose of private study or research, no part may be reproduced without written permission. The content is provided for informational purposes only.

Leave a Reply

%d bloggers like this: