WHAT WAS LOST: tracing unmet demand in contraceptive supply chains
Context
At the EURO 2025 conference in Leeds, we present our work on reconstructing true demand in family planning supply chains—where stockouts routinely censor observed data and distort decision-making. Standard forecasting tools treat these absences as lack of demand, leading to understocking and reinforcing supply failure.
We introduce a novel approach that integrates a Truncated Conformal Kalman Filter (TCKF) with simulation-based inventory evaluation. By correcting for both partial and full censorship and layering conformal prediction for distribution-free uncertainty, we recover latent demand more accurately and translate it into better ordering policies.
Through synthetic experiments and real data application, we show how ignoring censored demand underestimates both need and risk. Our results point toward a scalable framework for inventory management in fragile public health systems—where demand isn’t lost, just buried under broken assumptions.
Because “zero demand” doesn’t mean “zero need”—it often just means the shelves were empty.