Architecture Overview

VADEMOS Shiny App Architecture Overview

1. Introduction

The VADEMOS Shiny App is a web-based decision-support tool designed to estimate vaccine demand for Foot-and-Mouth Disease (FMD). It integrates forecasting models, geospatial mapping, and database queries to provide real-time analytics for policymakers and researchers.

---

2. Architecture Components

The VADEMOS architecture consists of three main layers:

A. User Interface (UI)

• Frontend: Built using Shiny's UI functions (fluidPage, tabPanel, sidebarLayout).
• Customization: Uses a custom CSS file (styles.css) for branding and UI enhancements.
• Interactive Elements: Includes dropdowns, buttons, sliders, and tables to allow user input and control over model parameters.
• Map Visualization: Uses leaflet to display vaccination density maps dynamically.

B. Server Logic

• Reactive Programming: Shiny’s reactive expressions handle dynamic updates when user inputs change.
• Data Processing: Utilizes R’s dplyr and tidyr packages for data transformation.
• Model Execution:
  • Uses AutoEnsemble models to forecast livestock populations.
  • Computes prophylactic vaccination requirements using historical trends and user-defined inputs.
  • Simulates emergency vaccination needs using geospatial density and user-selected radius.
• Database Interaction: Queries AWS-hosted MySQL databases to retrieve and update datasets dynamically.

C. Data Sources & Backend

• AWS RDS (MySQL): Stores:
  • FAO livestock population data
  • FMD control policies (PCP stages)
  • Vaccination coverage parameters
  • Geospatial density data (Grided Livestock Model)
• REST API (ESRI ArcGIS): Fetches country boundary shapefiles for visualization.
• FAO & WOAH Datasets: Integrated for real-world livestock and outbreak data.

---

3. Workflow & Data Flow

1. User selects input parameters (country, species, year, vaccination policies).
2. App queries the database to fetch:
   • Forecasted livestock population (AutoEnsemble model).
   • Predefined vaccination schedules.
   • Historical PCP stage data.
3. Prophylactic vaccination demand is calculated based on:
   • Forecasted livestock numbers.
   • PCP-based vaccination coverage.
4. Emergency vaccination is estimated based on:
   • User-selected radius.
   • Livestock density from FAO Grid Livestock Model.
5. Leaflet renders an interactive map with density-based coloring.
6. User can download results as CSV, Excel, or PDF reports.

---

4. Key Technologies & Libraries

Component	Technology Used
Frontend UI	Shiny, HTML, CSS
Database	MySQL (AWS RDS)
Data Processing	dplyr, tidyr
Forecasting Models	AutoEnsemble, ARIMA
Geospatial Analysis	leaflet, sf, geojsonsf
API Integration	httr, jsonlite
Export & Reporting	DT, shinyjs, shinyscreenshot

---