Deep neural networks for fine-grained surveillance of overdose mortality

TL;DR

This paper introduces a deep learning named-entity recognition model that significantly improves the identification of substances in death certificates, enabling more precise overdose mortality surveillance beyond traditional ICD-10 code analysis.

Contribution

A novel deep learning model for analyzing free-text death certificates that outperforms existing methods and detects new drug mentions, enhancing overdose death surveillance.

Findings

Achieved an F1-score of 99.13% in drug identification.

Can identify misspellings and novel substances not in current tables.

Improves specificity of overdose mortality data.

Abstract

Surveillance of drug overdose deaths relies on death certificates for identification of the substances that caused death. Drugs and drug classes can be identified through the International Classification of Diseases, 10th Revision (ICD-10) codes present on death certificates. However, ICD-10 codes do not always provide high levels of specificity in drug identification. To achieve more fine-grained identification of substances on a death certificate, the free-text cause of death section, completed by the medical certifier, must be analyzed. Current methods for analyzing free-text death certificates rely solely on look-up tables for identifying specific substances, which must be frequently updated and maintained. To improve identification of drugs on death certificates, a deep learning named-entity recognition model was developed, which achieved an F1-score of 99.13%. This model can identify new drug misspellings and novel substances that are not present on current surveillance look-up tables, enhancing the surveillance of drug overdose deaths.