Methodological Framework

Methodological Commitment

Africa Health Intelligence (AHI) is committed to the highest standards of scientific excellence, methodological rigor, and analytical reliability. Every indicator undergoes independent validation to ensure objective, consistent, and reproducible assessments before being incorporated into the platform. Indicators that fail to meet the required validation standards are refined, recalibrated and revalidated before inclusion in subsequent assessment cycles.

Validation Measure	Standard	Purpose
Intraclass Correlation Coefficient (ICC)	≥ 0.90	Ensures a high level of consistency in numerical scoring among independent evaluators.
Cohen's Kappa	≥ 0.85	Confirms strong agreement between two independent evaluators.
Fleiss' Kappa	≥ 0.85	Confirms strong agreement among three or more independent evaluators.

Core Principles

Health systems are political systems

power and stewardship shape outcomes.

Historical path-dependence matters

colonial legacies and past epidemics condition present capacity.

Lived functionality outweighs formal design

paper strategies are not enough.

Predictive capacity outperforms hindsight

early signals matter more than annual reports.

Scoring System

80–100 Strong

60–79 Functional

40–59 Strained

20–39 Weak

0–19 Critical

0100

80–100 Strong
60–79 Functional
40–59 Strained
20–39 Weak
0–19 Critical

Inversion Logic for Risk Indicators

For negative indicators (e.g., stock-outs, corruption, outbreaks), scores are inverted so that higher always means better.

Four Intelligence Layers

Historical Systems Layer

1950–PRESENT

Tracks governance structures, financing models, disease burden, infrastructure, workforce, and donor penetration over time. Establishes baselines and trajectories.

Structural Capacity Layer

CURRENT STATE

What exists today? Laws, facilities, workforce numbers, supply chain infrastructure.

Functional Performance Layer

REAL-TIME ANALYSIS

What actually works? Stock-out rates, referral completion, waiting times, guideline adherence.

Anticipatory Risk Layer

PREDICTIVE INTELLIGENCE

AI-driven outbreak probability, supply chain failure risk, workforce collapse risk, fiscal stress.

AI + Human Workflow

01

STEP 1
AI pre-fills each indicator using approved sources and proposes a score.
02

STEP 2
Analyst reviews, edits with justification, or flags for additional sourcing.
03

STEP 3
Optional external evaluator review.
04

STEP 4
Final score published with revision history.

AI and human expert collaborating on health intelligence workflow

Data Sources & Confidence

Data Categories

01
National administrative systems DHIS2, HMIS, budget portals
02
International repositories WHO, World Bank, UNICEF, UNAIDS
03
Remote sensing and environmental datasets flooding, temperature, rainfall
04
NGO and philanthropic program data
05
Expert and institutional surveys

Confidence Levels

High

multiple independent sources align

Medium

single reliable source

Low

model estimated or partial

Low confidence indicators are visually flagged throughout the system