Stellamarries-Syombua/Spatial_Modelling

Spatial modelling and mapping of Diabetes Prevalence in Kenya # 🌍 Spatial Modelling & Clustering of Diabetes Prevalence in Kenya ## 📌 Project Overview This repository presents an analysis of **diabetes prevalence in Kenya** using advanced spatial modelling and clustering techniques. Diabetes is a growing public health challenge in Kenya, projected to rise from **3.3% prevalence to 4.5% by 2025** without effective interventions. The project applies **Bayesian Hierarchical Models**, **Geographically Weighted Regression (GWR)**, and **Spatial Scan Statistics** to identify **diabetes hotspots**, assess determinants, and explore spatial heterogeneity across Kenyan counties. --- ## 🎯 Objectives ### General Objective - To identify distinct clusters, patterns, and determinants of diabetes prevalence in Kenya. ### Specific Objectives - 🔍 Identify **diabetes hotspot counties** in Kenya using spatial scan statistics and mapping. - 📊 Link the occurrence of diabetes to its determinants using **Hierarchical Spatial Models**. - 🧮 Select appropriate **spatial smoothing techniques** (based on Deviance Information Criterion, DIC). - 🗺️ Apply **Geographically Weighted Regression (GWR)** to analyze varying relationships between diabetes and risk factors across counties. --- ## 🏥 Background & Problem Statement - Globally, **422 million adults** live with diabetes; **14.2 million** in Africa (aged 20–79). - In Kenya, **prevalence is ~3.3%**, expected to rise to **4.5% by 2025**. - Diabetes poses economic and social burdens due to high **mortality, management costs, and lost productivity**. - County-level management is underfunded, understaffed, and under-resourced. This study provides critical data to guide **health policy and interventions** targeting non-communicable diseases in Kenya. --- ## 📚 Literature Review (Highlights) - Spatial smoothing techniques help reveal distinct disease prevalence patterns (Ogunsakin & Ginindza, 2022; Fang et al., 2006). - Bayesian mapping applied in malaria (Kazembe, 2007) and cancer (Rosenger et al., 20 …