Spatiotemporal Analysis and Risk Profiling of Dengue in Lima and Callao: A Data-Driven Approach for Tailored Prevention Policies

Abstract

The recent atypical increase in dengue cases in Lima and Callao, Peru, underscores the need for a more sophisticated approach to controlling this disease in urban settings. This study employs advanced data mining techniques to conduct a comprehensive spatio-temporal analysis of dengue cases in these regions, integrating climatic, sanitation, and demographic factors at the district level. Using k-means clustering, two distinct risk profiles were identified among the districts, revealing spatial patterns related to port proximity and exposure to geological risks. Sensitivity analysis using Shapley Additive Explanations (SHAP) quantified the contribution of various variables to dengue incidence, highlighting the importance of exposure to geological risks and water management practices. The results indicate that, contrary to previous assumptions, coastal humidity is not a determining factor in the spread of dengue in the region. Instead, factors such as wind speed, landslide exposure, and water conservation practices emerged as significant predictors. Based on these findings, the study proposes specific recommendations for dengue prevention policies tailored to district risk profiles. This data-driven approach to understanding and mitigating dengue in complex urban areas offers a potentially applicable model to other regions facing similar challenges in vector-borne disease control.