This study investigates a geographical targeting approach towards active case-finding for tuberculosis using artificial intelligence software. We conducted this study through a large pragmatic, stepped-wedge cluster randomized trial in 68 districts of Pakistan. This dataset includes anonymoized, aggregate data collected from TB active case finding interventions called chest-camps, that was used in the analysis of the trial and for sample size calculations. Camps have been identified as being conducted using the AI software (intervention) or by field teams (control).