olatunjijohnson/loa-loa-evt-risk-mapping

Bayesian joint geostatistical + extreme-value model for high-intensity Loa loa microfilaraemia risk (ivermectin MDA safety), south-eastern Nigeria # Loa loa high-intensity risk mapping with extreme-value theory Code for a Bayesian **joint geostatistical and extreme-value** model that predicts where dangerously high *Loa loa* microfilarial loads (>20,000 mf/mL) are appreciably common, to support the safety of ivermectin mass drug administration (MDA) for onchocerciasis and lymphatic filariasis in loiasis co-endemic areas. The analysis is applied to a survey of south-eastern Nigeria (Abia, Anambra, Delta, Ebonyi, Imo). The model 1. maps infection **prevalence** (logistic geostatistical regression), 2. maps microfilarial **intensity** among positives (lognormal body), the two fields linked by a **bivariate Gaussian process** (intrinsic coregionalisation), and 3. characterises the **upper tail** of intensity with a peaks-over-threshold **generalised Pareto distribution (GPD)** — necessary because the threshold of concern (20,000 mf/mL) lies beyond the largest observed density. For each Local Government Area (LGA) it reports `P(q_g [Authors]. *Predicting where few people carry dangerously high Loa loa > microfilarial loads: a Bayesian joint geostatistical and extreme-value analysis to > support safe ivermectin mass treatment in south-eastern Nigeria.* (in preparation / > under review). > **DOI: _to be added on publication._** Closely related prior work using geostatistics for loiasis MDA safety assessment: Johnson O, Giorgi E, Fronterrè C, et al. *Geostatistical modelling enables efficient safety assessment for mass drug administration with ivermectin in Loa loa endemic areas...* PLOS Negl Trop Dis. 2022;16(2):e0010189. ## Repository layout ``` code/ NGA_common.R Model definitions: Stan hurdle + ICM GP, GPD tail, kriging, exact per-LGA prediction, helpers (CFG, load_nga) NGA_fix_data.R Data-relabelling step (two misgeocoded villages) NGA_gee_extract.py Google Earth Engine extraction of environmental covariates NGA_validation.R Adequacy gate: convergence, calibrat …