We used NLP Machine learning classifier-based models and feature extraction to predict opioid overdose patients from discharge summaries collected from the publicly accessible Medical Information Mart for Intensive Care III (MIMIC III) database. The goal of this experiment is to evaluate the performance of machine learning based systems to automatically map clinical notes to opioid overdose.

Learning Objective: Evaluate machine learning approaches to classify opioid overdose clinical narratives free-text from electronic health records.

Authors:

Md Rashedul Hasan (Presenter)
US Food and Drug administration

Mitra Ahadpour, US Food and Drug administration
Henry Francis, US Food and Drug administration
Alfred Sorbello, US Food and Drug administration

Presentation Materials:

• Abstract/Manuscript - 1