Combining deep neural networks, a rule-based expert system and targeted manual coding for ICD-10 coding causes of death of French death certificates from 2018 to 2019

• Published in: International journal of medical informatics (Shannon, Ireland) Ireland), 2024-08, Vol.188, p.105462-105462, Article 105462
• Main Authors: Zambetta, Elisa, Razakamanana, Nirintsoa, Robert, Aude, Clanché, François, Rivera, Cecilia, Martin, Diane, Hebbache, Zina, Flicoteaux, Rémi, Coudin, Elise
• Format: Article
• Language: English
• Subjects: Aged; Cause of Death; Causes of death; Child; Child, Preschool; Clinical Coding - methods; Clinical Coding - standards; Coding; Death Certificates; Deep learning; Encoder/decoder; Expert Systems; Female; France; Humans; ICD-10; Infant; International Classification of Diseases; Male; Mortality; Neural Networks, Computer; NLP; Transformer
• ISSN: 1386-5056; 1872-8243
• DOI: 10.1016/j.ijmedinf.2024.105462

•Causes of death in death certificates are usually coded in ICD-10 either by rule-based expert systems or by human experts.•Transformer-type deep neural networks are trained on already coded data to complement modes of coding causes of death in a fully controlled way.•DNNs are integrated into the regular statistical production of the CoD database through interaction with other coding modes.•The allocation of DCs between the different modes of coding can be optimized to achieve best quality under the constraint of limited human resources. For ICD-10 coding causes of death in France in 2018 and 2019, predictions by deep neural networks (DNNs) are employed in addition to fully automatic batch coding by a rule-based expert system and to interactive coding by the coding team focused on certificates with a special public health interest and those for which DNNs have a low confidence index. Supervised seq-to-seq DNNs are trained on previously coded data to ICD-10 code multiple causes and underlying causes of death. The DNNs are then used to target death certificates to be sent to the coding team and to predict multiple causes and underlying causes of death for part of the certificates. Hence, the coding campaign for 2018 and 2019 combines three modes of coding and a loop of interaction between the three. In this campaign, 62% of the certificates are automatically batch coded by the expert system, 3% by the coding team, and the remainder by DNNs. Compared to a traditional campaign that would have relied on automatic batch coding and manual coding, the present campaign reaches an accuracy of 93.4% for ICD-10 coding of the underlying cause (95.6% at the European shortlist level). Some limitations (risks of under- or overestimation) appear for certain ICD categories, with the advantage of being quantifiable. The combination of the three coding methods illustrates how artificial intelligence, automated and human codings are mutually enriching. Quantified limitations on some chapters of ICD codes encourage an increase in the volume of certificates sent for manual coding from 2021 onward.