Project 04 - Geo-Spatial Modelling of Vaccine Coverage

1) Project Recap

This pipeline trains on DHS/NFHS cluster points and predicts onto raster grids to produce hyper-local coverage surfaces, uncertainty layers, dropout maps, and state-level evaluation summaries.

Primary modeling targets include BCG, DPT, and MCV dose outcomes across NFHS-4 (2015-2016) and NFHS-5 (2019-2021), with direct comparison between card-only and card+maternal-recall scenarios.

2) Interactive Mesh and Sensitivity Lab

Use this panel to interactively vary state, mesh granularity, model family, and maternal-recall regime. The mesh preview and metrics update in real time to show how uncertainty and fit move with each design choice.

The 0/1 vaccination point layer is generated from DHS birth-recode pipeline outputs (summary export in data/cluster-vax-summary.json).

State boundaries are loaded from real GeoJSON shape files in data/india-states/.

DHS birth recode input: KidsRecode/IAKR7DFL.SAV filtered to age 12-23 months and alive children.
Per-child 0/1 vaccination computed exactly as card-or-maternal-recall for each vaccine indicator.
Cluster GPS merge done via DHSCLUST = ClusterID from ShapeFile/IAGE7AFL.shp, then exported to Cluster_data.csv.
This page uses those generated NFHS4_IndividualData.csv/NFHS5_IndividualData.csv plus Cluster_data.csv to build cluster-level 0/1 summaries.

State Survey Model Family Data Source Regime Vaccine (Actual 0/1 Layer) Mesh Resolution Medium Mesh Offset 0.25 Mesh Cutoff 0.08

Point legend: green = vaccinated=1 majority in cluster, red = unvaccinated majority (vaccinated=0). Radius scales with cluster sample size.

Mesh Nodes (Est.)

-

Mean Edge (km)

-

Expected RMSE

-

Uncertainty Index U

-

Coldspot Stability

-

Expected PPC95

-

Clusters (Actual Data)

-

Selected survey geolocated clusters

Children (n)

-

Selected state + vaccine

Vaccinated (Vaccinated=1)

-

Unvaccinated (Vaccinated=0)

-

Cluster Mean Coverage

-

Mean of cluster-level rates

Uncertainty by Mesh Resolution (Current State/Model/Regime)

Sparse -

Medium -

Fine -

What Varies and Why

3) Spatial Scale and Why It Works

Scale

Training: survey cluster coordinates.
Prediction: boundary-constrained raster grid.
Reporting: state-level aggregated summaries.

Signal vs Noise Tradeoff

Cluster-level inputs preserve observed survey signal. Latent spatial smoothing regularizes sparse/noisy areas before rasterized outputs are exported for planning.

4) Core Statistical Model

For cluster i, vaccine outcome v, period t, maternal-recall regime m in {0,1}:

y_{i,v,t}^{(m)} ~ Binomial(n_{i,v,t}, p_{i,v,t}^{(m)})

logit(p_{i,v,t}^{(m)}) = x_i^T beta_{v,t,m} + w_{v,t,m}(s_i) + b_{state(i)} + b_{district(i)}

where s_i = (lon_i, lat_i) and w(·) is the latent spatial field.

5) Spatial Dependence and Mesh Construction

Adjacency is encoded through coordinate-driven mesh projection, not polygon-neighbor lists.

M = (V, T), V = mesh vertices, T = triangles

A = A(loc, M) (projection matrix from cluster/raster locations to mesh basis)

w(s) approx sum_{k=1}^{K} psi_k(s) * omega_k

Raster multilevel labels are propagated through nearest-neighbor assignment for state/district factors.

6) Matérn-SPDE Formulation

(kappa^2 - Delta)^(alpha/2) w(s) = W(s)

omega ~ N(0, Q(kappa, tau)^{-1})

This produces a sparse Gaussian Markov random field representation compatible with INLA.

Boundary/Sparse Handling

Mesh clipped to raster boundary polygon.
offset, cutoff, and max.edge tuned for stable triangulation.
NA raster cells excluded from prediction domain.

7) Covariates: Spatial vs Non-Spatial

Spatial component: latent field driven by longitude/latitude through SPDE mesh.

Non-spatial component: cluster-aggregated socioeconomic and service-use predictors (wealth, education, access barriers, maternal-care proxies), with optional iid hierarchical effects for State/District/Region.

8) Uncertainty Mapping

Prediction exports include:

posterior mean
standard deviation
2.5th and 97.5th posterior quantiles
relative uncertainty ratio

U_g = sd_g / mu_g

High U_g indicates unstable estimates requiring cautious interpretation.

9) Diagnostics and Fit Comparison

Current diagnostics include coverage/error/correlation summaries and CI checks; explicit residual Moran's I/variogram diagnostics are not yet fully automated in the committed pipeline.

RMSE = sqrt((1/N) * sum_i (y_i - y_hat_i)^2)

Corr = cor(y, y_hat)

Coverage95 = (1/N) * sum_i 1[y_i in [q0.025_i, q0.975_i]]

10) Maternal Recall Sensitivity (MR=0 vs MR=1)

The pipeline explicitly tests data-source sensitivity by toggling maternal recall inclusion. For MCV1, MR=1 improved fit and reduced uncertainty in both NFHS rounds:

Survey	Metric	MR=0	MR=1
NFHS-4	RMSE	0.388	0.295
NFHS-4	Uncertainty	0.443	0.241
NFHS-5	RMSE	0.272	0.253
NFHS-5	Uncertainty	0.254	0.173

11) Persistent Coldspots Across Survey Rounds

MCV1 Persistent Low-Coverage States (MR=1)

Arunachal Pradesh, Assam, Manipur, Meghalaya, Mizoram, Nagaland, Tripura, Uttar Pradesh.

DPT3 Persistent Low-Coverage States (MR=1)

Arunachal Pradesh, Assam, Meghalaya, Mizoram, Nagaland, Tripura, Uttar Pradesh.

These persistent patterns are useful for continuity-focused policy planning across survey waves.

12) Current Gaps and Risk Sources

No fully automated prior-sensitivity sweep in the current committed workflow.
No full posterior variance-partition report for spatial-effect share yet.
Largest data-quality risk: cluster geolocation uncertainty and key harmonization mismatch.

13) Planning Integration and First-Use Map

Pipeline already exports planning-ready raster/TIFF products: coverage, uncertainty, missed-cluster and error maps.

First map to use: DPT1 to DPT3 dropout probability map under MR=1, with uncertainty overlay. This is operationally high-value because it isolates service-continuity failure after first contact.

14) One-Sentence Policy Contribution

The project provides uncertainty-aware, cluster-to-grid geospatial estimates that distinguish stable versus unstable low-coverage pockets, enabling state programs to prioritize continuity gaps and maternal-recall-sensitive hotspots with greater confidence.