Room 129 / IM-APHP Team : Rémi Flicoteaux, Vassilina Nikoulina, Thomas Séjourné, Claire Vinet, Pierre Zweigenbaum, Cyril Grouin, Tarik Namik, Yohan Taillé, Mickael Elbaz
Increasing large volumes of healthcare data are regularly generated in the form of Electronic Medical Records (EMR). One major issue that can be approached by capitalizing on the routinely. generated textual data is the automation of diagnosis code assignment to medical notes. The task involves characterizing patient’s hospitalstay (symptoms, diagnoses, treatments, etc.) by a small number of codes, usually derived from the International Classification of Diseases (ICD). Comprehensive MIMIC database that spans more than a decade with detailed information about individual patient care offer a great opportunity for development and evaluation of automatic ICD code assignment pipelines that will ensure reproducibility of training and test methods.
Team : Medical information AP-HP.
Detail of the project : https://github.com/IM-APHP/mimic_icd9_code_assignment/blob/master/mimic_icd9_classification.ipynb