- Type
- Automation concept · public health surveillance
- Role
- Solo builder: survey design, automation, AI prompts, reporting
- Timeline
- 2026
- Client context
- Ministry of Health malaria program team & supporting NGOs
- Architecture
- Collect → Store → Analyze → Report → Share via n8n and LLM
“Facility data misses the majority of malaria treatment that happens outside facilities. The only way to see it is to ask the public directly.”
The problem
Official malaria statistics in Nigeria rely on facility data, but most symptomatic Nigerians never reach a facility. They buy medication at patent medicine stores, treat themselves at home, or use traditional remedies like agbo. That entire tier of care is invisible to the Ministry, NGOs, and donors who allocate RDT kits, ACT medications, and education materials.
Before MalariaPulse
Decision-makers planned interventions based only on confirmed clinic cases. Informal treatment, self-diagnosis, and traditional medicine usage were not measured, so resource allocation could not reflect actual health-seeking behaviour across states.
What I built
A five-stage pipeline: public survey intake, structured storage, n8n sanitization, LLM analysis, and automated report generation and delivery.
MalariaPulse establishes a direct-to-consumer survey channel, then uses an LLM to turn raw behavioural responses into structured, publication-grade reports. The design intentionally separates epidemiological analysis from executive report drafting so each prompt can be tuned independently.
- Public intake. A mobile-optimized Google Form distributed through social media, SMS, NGOs, and community groups.
- Response ledger. Responses land in a structured Google Sheets/Airtable ledger with a migration path to PostgreSQL or Supabase.
- n8n orchestration. A backend workflow sanitizes entries, groups them by state or LGA, and prepares the payload for analysis.
- Two-stage AI analysis. One prompt performs epidemiological analysis; a second prompt drafts the executive report, with explicit instructions not to overstate certainty.
- Automated delivery. Reports are saved to Google Drive and emailed to program officers, with a log entry for each run.
The outcome
Built as a public-health automation concept, MalariaPulse turns self-reported behaviour into decision-ready intelligence. A sample batch of 20 responses surfaced a 65% informal-treatment rate and multiple critical risk flags for banned or unverified drugs. The intended outcome is better allocation of diagnostics, medication, and education by decision-makers who can finally see the full treatment picture.
Stack
Next improvements
- Interactive dashboards. Add Looker Studio or Metabase dashboards for live trend visualization.
- WhatsApp intake. Add a WhatsApp-based survey channel for low-bandwidth and rural populations.
- Human-in-the-loop. Add a public-health reviewer step before any report is distributed externally.
Automation flow snapshot
MalariaPulse reporting flow
The public-health survey header, n8n workflow engine, example email notification with PDF attachment, and Google Drive storage.