Beyond the AI Hype: From AI promise to practical healthcare impact in Nigeria

A few thoughts on how we’re framing the AI in Nigerian healthcare question at Plural.

The opportunity gaps are obvious.

Structural staff shortages with entrenched nurse & doctor emigration, demonstrated by our 1:4,000-10,000 doctor-to-population ratio compared to WHO’s recommended 1:600.
Clinical reasoning variability, with diagnostic accuracy potentially as low as 36% in PHCs and with sector-wide fatal consequences.
Fragmented patient data, with siloed systems having adverse impact on health outcomes and costs – and AI providing a different way to approach this problem.
Missed drug–drug interactions (DDIs), with studies indicating up to 95% of potential DDIs missed, up to 43% of admissions due to drug-related issues and adverse drug reactions often underreported but increasing cost of care.
Maternal mortality, with Nigeria accounting for approx. 14% of global maternal deaths despite having 2.6% of the world’s population.
Zero-dose vaccinations, with Nigeria having the highest global burden of zero-dose (ZD) children (approx. 2.1 million children under the age of one not having received any routine vaccinations).
Childhood malnutrition: approx. 32%–37% of children under five stunted and 10.8%–18% wasting.

The headlines tend to focus on “AI will fix healthcare.” The real question is much harder:

Which specific, agentic AI use cases meaningfully solve our most consequential problems — for the actual healthcare workers on the ground?

Community health officers in PHCs.
Midwives managing antenatal care and labour.
Nurses.
Doctors.
Pharmacists.
Laboratory professionals synthesising fragmented patient data.

Where can AI safely augment workflows?
Where can parts of the job genuinely be handed off?

Some use cases are obvious — and still underexplored:

Medical scribing
Intelligent billing
Chronic disease follow-up and monitoring
Coordinating care for patients with multimorbidity across multiple facilities

But then comes the commercial reality.

Using $200/month AI subscriptions as a benchmark:

Why would a small private hospital owner in Ibadan pay for AI billing automation when they can hire a human for ₦50,000/month?

Why pay for AI-generated clinical notes when junior doctors are inexpensive (read “grossly undervalued”)?

If the promise is efficiency or upside, can that upside be demonstrated before the cheque is written?

Healthcare executives are solving for two questions:

Does this increase revenue?
Does this reduce cost?

In a low-labour-cost environment, the second question is less compelling than Silicon Valley assumes. Then layer in infrastructure realities:

Limited in-country data centre capacity
GPU constraints
Power and internet reliability
Regulatory preferences for in-country hosting
Cost implications of Azure/AWS/GCP GPU rentals
Model selection trade-offs (proprietary vs open-source, geopolitical considerations)

This is not a hackathon problem. So the real question for any serious African healthtech founder building in AI is:

Is my problem space clinically consequential — and can I demonstrate commercially relevant pilot data that scales affordably within our infrastructure and policy constraints?

Not activity.
Not noise.
Results.

At Plural, we’ve chosen our first bite carefully — a Nigeria-relevant problem space where the clinical stakes are real, the commercial logic is sound, and the infrastructure realities are acknowledged upfront.

To eat an elephant, you start with a bite.

We’ve picked ours.

Watch this space.