# Verbal Autopsy in Civil Registration and Vital Statistics: The Symptom-Cause Information Archive

**Authors:** Samuel J. Clark, Martin W. Bratschi, Philip Setel, Carla Abouzahr, Don de Savigny, Zehang Li, Tyler McCormick, Peter Byass, Daniel Chandramohan  
**arXiv:** [1910.00405](https://arxiv.org/abs/1910.00405)  
**Date:** 2019-10-01  
**Summary page:** https://www.summarizepaper.com/en/paper/1910.00405v2/

## Key points

- The burden of disease is important for public health interventions
- Calculating the burden of disease requires cause of death information
- Verbal autopsy is a method to determine cause of death through interviews
- Verbal autopsy plays an increasingly important role as civil registration and vital statistics systems improve in developing countries
- Machine-based algorithms are needed to automate cause assignment from verbal autopsy data
- A global symptom-cause archive is proposed to facilitate the development of machine-based algorithms
- The archive would be used for training and validation of automated cause assignment algorithms
- The establishment of such an archive aims to improve comparability and accuracy in cause assignment

## Summary

The burden of disease is a crucial aspect of public health interventions as it helps in understanding, prioritizing and monitoring the impact of various health conditions. Calculating the burden of disease requires information on the cause of death which is often not available in many parts of the developing world due to undetected or unassigned deaths. To address this issue, verbal autopsy has emerged as a feasible method for determining the cause of death by conducting interviews with individuals who were responsible for caring for the deceased person. As civil registration and vital statistics systems improve in developing countries, verbal autopsy plays an increasingly important role in providing valuable insights into causes of death and overall disease burden. To enhance accuracy and comparability across different contexts, there is a need to develop machine-based algorithms that can automate cause assignment from verbal autopsy data. To facilitate this development process, it is proposed to create a global symptom-cause archive containing reference deaths which includes both verbal autopsy data and causes assigned through an independent mechanism. This archive would be used as training and validation data to refine and test machine-based algorithms for automated cause assignment. The establishment of such an archive would not only improve the comparability of machine-assigned causes but also provide a means to fine-tune individual cause assignment within specific contexts. Ultimately, this initiative aims to enhance our understanding of disease burdens worldwide by ensuring accurate and standardized cause-of-death information through improved verbal autopsy practices and automated algorithms.