Artificial Intelligence Causal Effect Estimation of Oral Rehydration Solution Uptake Under Variable Household Survey Recall Windows: A Target Trial Emulation Evaluation of Healthcare Diarrhoeal Mortality Equity

Household surveys that measure oral rehydration solution (ORS) uptake for childhood diarrhoea typically ask caregivers to recall events over windows that differ across survey programmes and rounds, and the resulting heterogeneity is usually treated as measurement noise rather than as a structural feature of the estimand. The framework is specified as a sequence of definitions, assumptions and identification results, and it is calibrated against the real DHS Program indicator 'Households possessing a computer' (HC_HEFF_H_CMP), which was observed for 27 African countries across 56 country-survey observations between 2015 and 2025, with a mean of 7.893 and a standard deviation of 5.144. The recall window enters the identification proof as a bias term whose magnitude is bounded by the window length and the within-window hazard of the outcome, and that the target trial emulation recovers the mortality contrast only when the window is either held fixed across compared arms or explicitly modelled. The contribution is a methodological framework and a set of derived propositions, not an empirical estimate of ORS effectiveness; the DHS indicator is used to characterise the survey infrastructure within which such estimation would be conducted.