AfriHealthDesk

The Problem

Nigeria has over 50,000 community health workers (CHWs) deployed across rural and peri-urban communities. These workers are the first — and often only — point of healthcare contact for millions of Nigerians. Yet they work with paper checklists, no clinical decision support, and no structured way to record or escalate what they observe.

The numbers tell the story: Nigeria's maternal mortality ratio stands at over 1,000 per 100,000 live births according to WHO data. Out-of-pocket health spending consumes a devastating share of family income. And our DHS data reveals the ANC Paradox — women in northern Nigeria are completing antenatal care visits, yet still not delivering in health facilities. The gap between knowledge and action costs lives.

AfriHealthDesk was built to close that gap.

What I Built

AfriHealthDesk is a fully functional, mobile-first web application built entirely using MeDo — not a single line of code written manually. It runs in any phone browser, works offline, and speaks Nigerian Pidgin English.

Core features:

AI Symptom Triage — CHWs describe symptoms in plain English or Pidgin and receive a color-coded risk level (GREEN / YELLOW / RED) with specific clinical action steps. The triage engine is calibrated using real Nigeria DHS maternal mortality data per geopolitical zone — a CHW in North West gets different risk thresholds than one in South South based on actual survey data.

Pidgin English AI — The triage assistant recognizes phrases like "body dey hot", "e dey purge", "pikin dey shake", and "eye don sink" and maps them to clinical symptoms. Responses are delivered in Pidgin: "WAHALA! E bad well well! Carry am go hospital sharp sharp!"

Zone-Aware Risk Calibration — Nigeria's 6 geopolitical zones are selectable. Northern zones automatically trigger higher malnutrition and maternal risk sensitivity based on uploaded DHS data, with a note: "Risk calibrated using DHS North West maternal data."

Referral Letter Generator — After every triage, CHWs can generate a formatted, downloadable referral letter pre-filled with patient symptoms, risk level, and recommended facility type — ready to hand to facility staff.

Patient Visit Logger — Structured data capture for every visit: demographics, vitals, chief complaint, triage outcome, referral decision — exportable as CSV for program reporting.

Danger Signs Flashcards — 5 offline-capable visual cards covering Maternal Emergency, Severe Malaria, Severe Malnutrition, Severe Dehydration, and Newborn Danger Signs with large icons and bold action text.

Supervisor Dashboard — PIN-protected view showing visit trends, risk distribution charts, referral rates, and a Nigeria Health Context Panel pulling real statistics from WHO, DHS, and health financing data with full source attribution.

Offline Mode — Core features work without internet via service workers. Patient logs are saved locally and synced on reconnection.

Hausa Language Support — Key triage responses also available in Hausa for northern Nigeria CHWs.

The Data Behind It

AfriHealthDesk is not a static chatbot. It is powered by three real datasets uploaded directly into MeDo:

1. WHO Maternal Health Dataset — Nigeria MMR, ANC coverage rates, skilled birth attendance, facility delivery rates
2. Nigeria DHS Maternal Mortality Data — zone-specific MMR, ANC paradox data, facility delivery by geopolitical zone
3. Nigeria Health Financing Data — out-of-pocket spending %, government health expenditure per capita, Africa comparison

Every statistic displayed in the app is labeled with its source. The triage risk thresholds are calibrated from real survey data, not estimates.

How I Used MeDo

I built AfriHealthDesk using MeDo's conversational interface across a single comprehensive prompt session. I uploaded all three CSV datasets directly into MeDo before generation, so the app was grounded in real Nigerian health evidence from the start.

I then used targeted follow-up prompts in the same chat thread to refine specific features — the triage combination symptom detection, the Pidgin language mapping, and the header layout. Each fix cost minimal credits and was applied precisely without regenerating the entire app.

The most impressive thing MeDo generated was the combination symptom triage engine. When I described that "severe headache + persistent fever lasting 2 days" should trigger RED with specific malaria/meningitis/pre-eclampsia actions, MeDo implemented the full clinical logic including duration-based severity escalation — something that would have taken a developer significant time to code manually.

MeDo also handled the Nigerian Pidgin English recognition layer entirely from my plain English description of the phrases. I did not write any pattern matching code. I described the problem in natural language and MeDo solved it.

Real-World Impact

AfriHealthDesk is deployable today. Any CHW program coordinator, state ministry of health, or NGO field team can access it on any smartphone browser — no installation, no account, no cost. I am planning a pilot with a CHW network in Cross River State, Nigeria, and am exploring integration with DHIS2, Nigeria's national health data system.

Why Surprise Us?

No other participant in this hackathon is building for Africa's frontline health workforce. This is not a task manager, a quiz app, or a customer dashboard. This is clinical infrastructure — built in days, by one person, using a tool that requires no coding knowledge. That is exactly what MeDo was designed to enable, and exactly what Africa's health system needs.

How MeDo Was Used (separate field)

I used MeDo's multi-turn conversational interface with uploaded CSV datasets to build a complete 7-page clinical decision support application. Key MeDo capabilities used: full-stack app generation from natural language, file upload integration for real health data, multi-turn refinement prompts, one-click deployment to public URL. The entire app was built without writing any code.

Most Impressive Feature MeDo Generated

The zone-aware combination symptom triage engine — detecting that "severe headache + persistent fever + 2 days duration" means potential malaria/meningitis/pre-eclampsia and responding in Nigerian Pidgin English with specific clinical action steps calibrated to the patient's geopolitical zone using real DHS data.

Built With

• medo