Spatial machine-learning analysis of child indicators

Description du projet

Apprentissage automatique spatial pour la surveillance des enfants vulnérables

L’initiative Horizon Europe s’aligne sur les objectifs de développement durable (ODD) des Nations unies. La commission OMS-UNICEF-Lancet met particulièrement l’accent sur les enfants, reconnaissant leur rôle essentiel dans la réalisation de ces objectifs, car ils constituent l’un des groupes les plus vulnérables et les plus marginalisés au monde. S’appuyer uniquement sur des données d’enquête pour calculer les indicateurs relatifs aux enfants présente des limites, car les enquêtes peuvent nécessiter des mises à jour périodiques ou ne couvrir que certaines régions d’un pays. Le projet SMALACI, financé par le programme MSCA, utilise l’imagerie satellitaire et des techniques d’apprentissage automatique pour identifier les enfants vulnérables dans les pays dont les données d’enquête sont obsolètes ou incomplètes. L’objectif global est de développer des outils permettant de suivre les progrès accomplis dans la réalisation des ODD, de concevoir des interventions ciblées et de réduire les coûts associés aux programmes de réduction de la pauvreté.

Objectif

Horizon Europe seeks to achieve the UN’s Sustainable Development Goals (SDGs). The WHO–UNICEF–Lancet Commission suggests placing children at the center of the SDGs, because children are among the world’s most vulnerable and marginalized population. Child indicators calculated with surveys are limited because surveys can be outdated and do not cover all the regions of a country. In this context, the objective of the project is to develop a methodology to overcome this limitation and identify vulnerable young children in countries with outdated, incomplete, and low-quality survey information. The methodology is based on a spatial machine-learning analysis of child indicators and integrates 3 disciplines: child development, spatial analysis of satellite images, and machine-learning. Satellite images and households’ simulations will complement the missing or outdated information of surveys, and machine learning will identify children that could be left behind during the development process due to the intersectionality of gender with biological diversities, ethnicity, socio-economic status, and geographical location. The project will create new tools for monitoring the progress towards the SDGs and will help to formulate targeted interventions that increase the social impact and reduce the economic costs of poverty-reduction and development programs. If the project is funded, Dr. Rolando Gonzales Martinez will carry out the fellowship at the University of Groningen, under the supervision of Prof. Dr. Hinke Haisma and with support from Prof. Dr. Dimitris Ballas. A short visit to UNICEF is planned as a secondment for Dr. Gonzales Martinez, so he can work with UNICEF on improving the policy impact of the project. The transfer of knowledge between the research fellow and the host organization, the teaching activities, and the short visit to UNICEF will increase the career prospects and employability of Dr. Gonzales Martinez within and outside academia.