African Community Health Worker Interaction Dataset

This is a synthetic dataset generated from high-quality expert-labelled seed data. All records are algorithmically derived — statistical distributions, inter-field correlations, and annotation characteristics faithfully replicate real-world patterns from the source data, while ensuring no real individual, organisation, or transaction can be identified or reconstructed.

The African Community Health Worker Interaction Dataset contains 500K+ structured visit records generated by community health workers (CHWs) operating in rural and peri-urban communities across Nigeria, Ghana, and Kenya. CHWs recorded each household visit using a standardised digital data collection tool deployed on low-end Android smartphones, capturing maternal health indicators, child nutrition and growth metrics, immunisation status, symptom flags, and the referral decision made at the point of care.

Maternal health records cover antenatal visit compliance, delivery location, postnatal check scheduling, and danger sign screening for conditions including pre-eclampsia, postpartum haemorrhage, and neonatal infection. Child records track MUAC measurements, weight-for-age Z-scores, breastfeeding status, and vaccination history against the national immunisation schedule. Each visit record is linked to a household identifier enabling longitudinal panel analysis of health trajectories across multiple CHW visits.

The dataset is purpose-built for CHW decision-support AI — systems that suggest next-best actions, flag high-risk households for supervisor follow-up, and optimise visit routing. It is also used for supply-chain forecasting (predicting commodity needs based on caseload trends) and for training multilingual voice-based data collection assistants that reduce CHW administrative burden.