The problem – Data incompleteness and transcription error are possible sources of the inconsistent quality in data that are sometimes collected from resource-poor settings. By collecting large amounts of data and applying predictive algorithms to place uncertainty bounds on the value of the missing or errant data, the effect of perturbations and omissions can be reduced. Machine learning (or data mining) algorithms provide a way to determine patterns in data by separating the signal from the noise so policy makers are able to make inferences from predictive models.

The way forward - Machine learning algorithms were applied to national-level and regional data to predict cause of death and disability, and to determine the determinants of diarrheal illness and their relative importance.

Contribution – Developing algorithms; analyzing data; publishing and presenting findings.