A semantically annotated Verbal Autopsy corpus for automatic analysis of cause of death

This paper presents a method employed in building a semantically annotated corpus of 11,741 Verbal Autopsy documents, each annotated with Cause of Death, based on verbal records of deaths of mothers, stillbirths, and infants up to 1 year of age, captured for analysis in Ghana between December 2000 and July 2010. Verbal Autopsy is a technique which involves interviewing individuals (such as relatives or caregivers) who were close to the deceased, and if possible, who cared for the individual around the time of death, to document events that may have led to the individuals’ death. The Verbal Autopsy technique is recommended by the World Health Organisation as a pragmatic substitute for a clinical autopsy to establish cause of death in regions such as sub-Saharan Africa where death may occur well away from clinical services. An evaluation is carried out based on established criteria to demonstrate that the Verbal Autopsy corpus possesses the qualities of many referenced corpora. The experiences drawn from the methods employed, with alternative approaches, may lead to a more efficient and cost effective corpus development framework. ICAME Journal No. 37