JMIR Publications

ABSTRACT

Kenya has made substantial progress in tuberculosis (TB) case detection, yet a persistent proportion of incident TB remains undiagnosed or unreported. Active case finding (ACF) can help close this gap but is resource-intensive, making accurate geographical targeting increasingly important. Through Tamatisha TB programme, the Centre for Health Solutions Kenya (CHS), EPCON and Kenya's National Tuberculosis, Leprosy and Lung Disease Programme (NTLD-P) have begun implementing the Epi-control platform, an artificial intelligence-enabled geospatial decision-support system designed to support data-informed ACF planning. The platform integrates routine TB programme data with demographic, socioeconomic, environmental and health-system indicators within a common spatial framework. A ward-level machine-learning model is being developed to estimate TB positivity and identify priority areas for ACF, while a higher-resolution sub-ward framework supports contextual visualisation and future micro-geographic modelling. Implementation has been accompanied by co-creation, stakeholder engagement and structured capacity building at national and county levels. This viewpoint describes the platform's development and early implementation experience, including data integration, modelling, stakeholder engagement and anticipated challenges. It also outlines the pathway toward geolocated, sub-ward modelling and prospective evaluation of whether model-informed targeting improves ACF yield, notification, geographic coverage and resource efficiency. The Kenya experience highlights that successful adoption of AI-enabled public health tools depends not only on predictive performance, but also on data quality, governance, interoperability, programme ownership, local analytical capacity and the operational ability to act on model outputs.