Africa Modelling Bridge a modelling movement
Malawi health-systems immunisation Preview

Forecasting EPI coverage drop-out following 2025–26 donor reductions

Which Malawi districts are most exposed to EPI coverage drop-out as donor financing contracts in 2025–26?

Ministry · Ministry of Health (Malawi) / EPI Programme

Difficulty
MSc
Timeline
8 months
Methods
3 listed
Budget ask
USD 1,500
Bayesian hierarchical time-series forecasting spatial smoothing

Why it matters

The Malawi Ministry of Health is preparing prioritisation decisions for the 2026 EPI cycle under a tightened external-financing environment. Pure public data (HMIS, EPI annual bulletins, DHS) is enough to identify the districts most exposed — but no one is currently doing the forecast in a way the ministry can act on.

The question

As donor-supported immunisation financing contracts in 2025–26, which Malawi districts are most exposed to EPI coverage drop-out — and what is the recommended sequencing of mitigation given a finite operational budget?

Why now

Malawi’s EPI programme has historically depended on external co-financing. The 2025–26 shifts in the Gavi co-financing landscape and bilateral aid contractions mean districts that were marginally above coverage thresholds in 2024 may slide below them in 2026. The ministry needs a forecast it can act on — not a retrospective.

Methods landscape (sketch)

A Bayesian hierarchical model with district-level random effects, a temporal trend component, and a covariate set drawn from HMIS staffing, cold-chain coverage, and historical drop-out rates. Spatial smoothing using a CAR prior on district adjacency. Validation against 2023 holdout. Not a 3D space-time GP — overkill for this question. Not a deep-learning approach — the data don’t justify it and the ministry would not be able to interpret it.

Full dossier with starter pack, data sources, supervisor and adviser publishes on v0.1.