mabera/nigeria-malaria-health-model

Nigeria Malaria Health Model

AutoScientist Challenge 2026 | Healthcare Category

Author: Hussein Adeiza (mabera)
Role: Licensed Environmental Health Officer, Abuja Nigeria
Base Model: Llama 3.3 70B
Fine-tuned with: AutoScientist by Adaption Labs

Model Description

This is a LoRA adapter fine-tuned on Nigeria DHS malaria health data (2010–2021). It predicts and explains malaria prevalence risk from ITN coverage, immunization and child mortality indicators across Nigeria.

Training Data

• Source: DHS Nigeria national surveys (2010–2021)
• Dataset: malaria parasitemia, ITN coverage, immunization, child mortality
• Quality improvement: 163.3% (Grade D → B)
• Kaggle: https://www.kaggle.com/datasets/yunusahusseinadeiza/nigeria-malaria-risk-model-prevalence-prediction

Training Metrics

• Win rate: 65% adapted vs 35% base model
• Base model: meta-llama/Llama-3.3-70B-Instruct
• Method: LoRA (r=16, alpha=32, all-linear)
• Epochs: 1

Key Finding

Malaria prevalence rose from 36.2% in 2018 back to 39.6% in 2021 despite increased ITN coverage — signaling a sustained coverage gap. Pregnant women ITN coverage is the strongest protective factor against malaria.

Why This Matters

Nigeria carries the world's largest malaria burden. This model addresses the gap in African epidemiological AI — built by a Licensed Environmental Health Officer with real field experience in Abuja.

Credits

Powered by Adaptive Data — Adaption Labs
AutoScientist Challenge 2026