Mastering the processes of model design and exploration is essential to transform observations and empirical knowledge into theories and tools that can support understanding and decision-making. In all approaches where agent-based modeling and simulation are used, researchers aim to generalize processes in order to validate them and transfer them to other contexts. In the long term, this should lead to the development of empirically validated and numerically explored modules in GAMA, reusable in other situations. This theme, which lies at the core of UMMISCO’s expertise, will be implemented in Senegal through the development and exploration of mathematical and computational models to study marine socio-ecosystems and Sahelian agro-sylvo-pastoral systems in a context of climate change and deep transformations in local practices. The Sahel has been facing ecosystem degradation since the second half of the 20th century, due to both climatic and anthropogenic factors. West African savannas could play an important role in climate change mitigation, provided that dialogue is established with local stakeholders to co-construct viable carbon neutrality scenarios aligned with sustainable development challenges. This participatory approach will be implemented within the Ferlo-Sine project, in two territories representative of land use in the Great Green Wall (GGW) intervention area in Senegal. Participatory workshops (“scenario labs”) will lead to co-constructed and modeled scenarios using the MAELIA platform. The most relevant scenarios will be discussed with local authorities, NGOs, government services, and communities to define territorial action plans aimed at achieving carbon neutrality by 2035. These results will also contribute to low-carbon scenarios at the scale of the GGW (IRN RESET-GMW project). Another key issue is artisanal fisheries, which play a major role in Senegal’s economy and regional food security. However, catches have declined in recent years due to overexploitation. Research conducted during UMMISCO 3 highlighted the underlying bioeconomic dynamics and the need for decision-support tools. This is the objective of the “Lolli” simulator, developed within the H2020 “Habitable” project, focusing on the mobility of artisanal fishers in Senegal. Implemented in GAMA, this simulator enables the exploration of climatic and socio-economic scenarios. It will be further developed by: extending it to inland fisheries at Lake Guiers (Santés & Territoires project), exploring the impacts of spatial management measures such as marine protected areas, in collaboration with the Senegalese Ministry of Environment and Sustainable Development and the Stockholm Resilience Centre (FORA project).

Artificial intelligence, based on deep learning techniques, is driving major transformations in society. However, its rapid development often leads to opaque and biased tools that are difficult to transfer to different contexts. Developing local expertise in this field is therefore essential in countries of the Global South, to avoid dependence on algorithms developed and trained on data from the Global North. UMMISCO West Africa focuses in particular on the use of AI for monitoring human health and socio-ecosystems. Early disease detection, a key factor in improving public health, often relies on visual analysis of medical images by doctors or midwives in rural areas. The limited availability of specialists is therefore a major constraint. At the same time, hospitals generate large annotated image datasets. To leverage this resource, UMMISCO West Africa has established collaborations with major hospitals (Le Dantec, Abass Ndao, Dalal Jamm) and the Faculty of Medicine in Dakar to develop AI-based diagnostic tools. Targeted pathologies include cervical cancer, strabismus, skin diseases, and heart conditions. Some diseases are linked to environmental conditions, such as schistosomiasis, for which deep learning is being explored to detect prevalence hotspots around Lake Guiers (Santé et Territoires project). In fisheries and marine biodiversity monitoring, AI contributes to continuous monitoring systems for socio-ecosystem parameters. Annotated aerial image datasets of pirogues have enabled the development of detection algorithms currently being integrated into operational tools (Habitable project). For biodiversity monitoring, a large dataset of underwater acoustic recordings has been collected. However, annotating these data and developing fish sound detection models remains a challenge and is the subject of ongoing doctoral research. New international collaborations are also being developed, including with a Colombian bioacoustics team from the Humboldt Institute. These competencies may later support AI-based monitoring of terrestrial biodiversity, particularly within the IRN RESET-GMW project.

The quality and quantity of available data in developing countries are often limiting factors for predictive modeling and AI training. The high cost and deployment constraints of industrial sensors are major barriers. To address these challenges, UMMISCO West Africa has invested in the development of the “CoFab in Dakar” within the IRD-UCAD international campus. The objective is to foster local expertise in designing and deploying frugal sensors adapted to local needs. These tools will support ongoing research projects within the center and partner institutions. Their industrialization could lead to the creation of a start-up during UMMISCO 4. Within the “Santés et Territoires” project, the CoFab developed a prototype autonomous station for water quality monitoring, deployed around Lake Guiers. This system integrates multiple sensors capable of transmitting data in real time. Following successful testing, the next step will be to deploy additional stations and train participants in living labs to interpret data and produce narratives based on measurements. For underwater acoustic data collection, expensive hydrophones are required, limiting their use in participatory monitoring systems. In collaboration with ESP and LMI ECLAIRS, a “made in Senegal” hydrophone prototype has been developed and successfully tested at Lake Guiers. A more robust version is planned for UMMISCO 4.

By placing participatory approaches at the core of its research strategy, UMMISCO West Africa adopts a modern and inclusive approach. Through co-construction of knowledge and mediation between different knowledge systems, projects aim not only to advance scientific research but also to promote sustainable and equitable territorial development. The center has long-standing expertise in these approaches, particularly in Sahelian agro-sylvo-pastoral systems, where it continues to organize living labs bringing together researchers and local stakeholders. UMMISCO West Africa aims to strengthen these interactions through collaborations with research groups such as ComMod (UMR SENS), UMR ASTRE, and UMR G-EAU. These approaches will be applied across projects, drawing on two complementary frameworks: ComMod and Living Labs. The Companion Modeling (ComMod) approach transforms the relationship between researchers, models, and stakeholders. By working closely with stakeholders, researchers must integrate multiple perspectives, ensuring that model outcomes meet both scientific and stakeholder expectations. In the Ferlo-Sine project, this process is truly reciprocal: stakeholders contribute to the scientific process while researchers support local reflection. This dual support aims to enhance the transformative impact of participatory modeling. Living labs, on the other hand, foster user-centered innovation by exploring, co-creating, testing, and evaluating solutions in real-world contexts. By involving diverse stakeholders, they support the development of sustainable and resilient practices. Within the “Santés et Territoires” project, living labs have been implemented in Mbane and Keur Momar Sarr, and this approach will be extended in future projects during UMMISCO 4.