1.     Schistosomiasis is a parasitic disease that affects over 250 million people worldwide, with the majority living in rural areas of sub-Saharan Africa. The parasite relies on freshwater snails of the genus Biomphalaria as intermediate hosts. Mapping snail distribution is vital for identifying disease transmission hotspots. However, expert-led monitoring is often constrained by limited resources and restricted access to remote areas, highlighting the need for scalable and cost-effective alternatives.2.     This study evaluates the effectiveness of citizen science in predicting Biomphalaria spp. presence by comparing models built from citizen- and expert-collected data. We tested two scenarios: the first one assumed perfect detection and focused on environmental and geomorphological predictors, while the second accounted for imperfect detection to explore discrepancies between citizen observations and expert-derived detection probabilities.3.     In the perfect detection scenario, the expert and citizen models identified site type and NDVI as significant environmental predictors of snail presence. Although both models demonstrated low marginal R² values (~16-17%), indicating limited explanatory power of broad-scale environmental predictors, conditional R² values exceeded 65%, suggesting that fine-scale, site-specific habitat characteristics are critical determinants of Biomphalaria presence. For the imperfect detection scenario, models showed minimal discrepancies, primarily explained by individual observer variability and differences in sampling effort. Increased sampling effort consistently reduced false negatives and led to unexpected observations of snail presence by the citizens (i.e. observed presence in sites predicted unsuitable by expert model). 4.     Our findings demonstrate that citizen science, when properly structured and statistically accounted for, can generate ecological data with accuracy comparable to expert-led surveys. We highlight the importance of accounting for observer variability, providing calibrated training, and optimizing sampling strategies to enhance data quality. This study presents a transferable and cost-efficient framework for participatory ecological monitoring in resource-limited and under-sampled regions.