AI in Healthcare — Africa Risk Dashboard
Contextual and risk analysis of clinical-AI deployments across nine African countries, benchmarked against the United States. Explore the risk tier filter and country search below.
1 · Executive summary
- Across the 9 countries studied, 16 clinical-AI models are deployed — an order of magnitude below the US benchmark of 84 models and ~4,200 hospitals.
- Declared healthcare-AI funding sums to $790.2M, but the largest single item (Babylon Health, ~$650M+) is foreign-funded rather than domestic.
- 4 countries are at Critical risk (ETHIOPIA, KENYA, MAURITIUS, GHANA) and 4 at High risk (NIGERIA, SOUTH AFRICA, EGYPT, MOROCCO) — driven mainly by data-transparency gaps and narrow deployment footprints.
- RWANDA has the lowest composite risk (66.4/100) but still trails the US benchmark on every axis.
- Clinical-domain coverage is concentrated in radiology, triage and documentation; high-burden areas such as maternal–fetal medicine, pathology and emergency triage remain largely uncovered.
2 · Contextual landscape
Where are clinical-AI models deployed, how many hospitals use them and how much funding supports them?
3 · Risk measurement
Six normalised risk dimensions capture data transparency, deployment scale, funding sustainability, clinical-domain gaps, population equity and the aggregate benchmark gap to the US. Higher values = higher risk.
Risk-dimension glossary
- Data Transparency — Share of core metrics that are undocumented or unclear for a country's deployed AI models. Opaque reporting blocks audit, clinical validation and post-market surveillance.
- Deployment Scale — Risk that models are deployed in too few hospitals to be clinically validated or generalisable across patient populations.
- Funding Sustainability — Risk that AI models lack the funding per model needed for maintenance, retraining on drifted data and regulatory compliance.
- Clinical Domain Gap — Distance to a mature market's (US) coverage of 18 clinical domains. Narrow domain coverage leaves high-burden conditions unaddressed.
- Population Equity — Risk that too few patients benefit from clinical AI, creating access gaps between those served and those excluded.
- Benchmark Gap — Composite gap between the country and the US benchmark on models, hospitals and domains — the headline inequity indicator.
4 · Country summary table
| Country | Models | Hospitals | Domains | Transparency | Risk | Tier |
|---|---|---|---|---|---|---|
| MAURITIUS | 1 | — | 1 | 0.0% | 98.7 | Critical |
| GHANA | 1 | — | 1 | 0.0% | 98.7 | Critical |
| KENYA | 1 | 1 | 1 | 33.3% | 93.1 | Critical |
| ETHIOPIA | 2 | 2 | 3 | 33.3% | 90.6 | Critical |
| MOROCCO | 2 | 21 | 2 | 0.0% | 83.8 | High |
| SOUTH AFRICA | 3 | — | 4 | 0.0% | 77.6 | High |
| NIGERIA | 3 | 38 | 3 | 33.3% | 73.8 | High |
| EGYPT | 2 | 2 | 3 | 33.3% | 73.0 | High |
| RWANDA | 1 | — | 2 | 66.7% | 66.4 | Moderate |
5 · Country-by-country context
Filter by risk tier or search for a country.
NIGERIA High Risk 73.8
3 indigenous models; strongest population reach (11k+) but small per-model funding (~$900k avg). Triage, preventive care and clinical documentation focus.
SOUTH AFRICA High Risk 77.6
Radiology-heavy; Qure.ai alone reports 10M+ patients, but hospital counts are undocumented — hidden footprint risk.
EGYPT High Risk 73.0
Dominated by Vezeeta's telehealth platform (10M users, 78 cities); Baheya breast-cancer AI is narrowly deployed (2 hospitals).
ETHIOPIA Critical Risk 90.6
Early-stage ecosystem; AXIR-CX data mainly from research (n=1,579), generalisability risk is high.
MOROCCO High Risk 83.8
Mostly foreign-sourced tech (Da Vinci) without local funding trail; endocrinology + surgery, narrow domain coverage.
KENYA Critical Risk 93.1
Single AMR-stewardship model in one hospital (Narok) — critical scale risk despite a high-impact clinical use case.
RWANDA Moderate Risk 66.4
Relies on Babylon Health (foreign funded, $650-675M+) — sustainability tied to an external commercial roadmap.
MAURITIUS Critical Risk 98.7
DRRIYA only; no documented hospitals, users or funding — near-total transparency gap.
GHANA Critical Risk 98.7
minoHealth radiology AI system; almost every contextual field undocumented — critical transparency risk.