Salus

Summary

Salus consolidates global health data into a single warehouse covering disease burden, health system capacity, financing flows, and population outcomes. Fifty-four asynchronous collectors pull from WHO, World Bank HNP, OWID, IHME, UNAIDS, Gavi, the Global Fund, and the UN Population Division. Four analytical engines turn the underlying observations into series-level statistics, cross-country comparators, and persistence diagnostics for policy work.

What it is

Salus is a production observatory for global health economics, designed to answer the operating questions that ministries of health, multilateral funders, and policy researchers ask repeatedly: how much does a country spend on health, where does the money come from, what does the system deliver, and how do outcomes compare to peers. The platform ingests fifty-four streams from authoritative sources, normalizes them into a SQLite warehouse with a health-specific schema (disease, age group, sex), and exposes the results through a documented FastAPI service with read-only data routes and four analytical engines.

The four engines correspond to the four pillars of health economics. The disease burden engine integrates WHO TB, malaria, NCD, AMR, and IHME GBD streams to track morbidity and mortality across age and sex strata. The health systems engine covers workforce density, hospital beds, immunization coverage, and service availability. The financing engine ties together WHO GHED, World Bank health expenditure series, Gavi disbursements, and Global Fund flows to trace public, private, and donor financing. The outcomes engine surfaces life expectancy, mortality rates, and UHC service coverage indices. Each series ships with summary statistics, year-over-year features, AR(1) persistence diagnostics, and cross-country comparators, so analysts can move from raw data to defensible findings without writing custom pipelines.

Methodology

• Async collectors built on a shared BaseCollector contract that enforces collect, validate, and store stages
• Health-specific warehouse schema with disease, age group, and sex dimensions on every series
• Annual observation grain with confidence bounds preserved where source agencies publish them
• APScheduler orchestrates staggered UTC collection windows to respect upstream rate limits
• Series-level statistics include means, extremes, compound annual growth rates, volatility, and shock detection
• AR(1) persistence and half-life estimates for every indicator to support stationarity judgments
• Cross-country comparators ranked within region and income group for peer benchmarking
• Architectural test guards prohibit mock data and enforce primary-source provenance

Data sources

• WHO Global Health Observatory, GHED, GHE, TB, malaria, NCD risk factor, UHC service coverage, AMR GLASS, WUENIC, CHOICE, air quality, child nutrition, and WASH
• World Bank Health Nutrition and Population statistics and World Development Indicators
• Our World in Data health, vaccination, mental health, maternal health, and obesity datasets
• IHME Global Burden of Disease
• UNAIDS HIV epidemiology and treatment cascade
• Gavi immunization disbursements
• The Global Fund grant and disbursement data
• UN Population Division World Population Prospects

Deliverables when used in engagements

• Documented REST API with country, series, panel, and comparator endpoints
• Four analytical engines covering disease burden, systems, financing, and outcomes
• Country and series briefs generated from the warehouse for policy audiences
• Operator dashboard with collection status, source health, and admin diagnostics
• Reproducible long-format panel extracts for downstream econometric work