Mining medical events in EHRs

keywords: 
Neural networks, classical ML methods (CRFs, graphs based…), medical event detection
Description: 
The main idea in this work is the development of a system to perform text mining in electronic health records in order to identify events between medical entities. Some examples can be, events between drugs and disorders representing adverse drug reactions (ADR) or treatments, events between diseases when a disease generates another one, or events between for example, disorders and the location or part of the body where they occur. There are some avalaible materials that could be used in this proposal: a manually annotated corpus, a thesis in ADR detection using machine learning and medical entity recognition tools.
Objectives: 
Extraction of medical events from electronic health records in Spanish
Task: 
i) State of the art analysis in medical event detection using different techniques as neural networks and ML methods, ii) design and development of a system for event detection, iii) implementation iv) result analysis and discussion and iv) dissemination of results.
References: 
Casillas A., Pérez A., Oronoz M., Gojenola K., Santiso S. (2016). Learning to extract adverse drug reaction events from electronic health records in Spanish. Expert Systems with Applications, Volume 61, 1 November 2016, Pages 235-245 K.B. Cohen, D. Demner-Fushman. Biomedical Natural Language Processing. Natural Language Processing, John Benjamins Publishing Company (2014)
Team: 
Arantza Díaz de Ilarraza, Koldo Gojenola, Alicia Pérez, Arantza Casillas, Aitziber Atutxa, Olatz Pérez de Viñaspre, Maite Oronoz
Profile: 
Informatikaria
contact: 
maite.oronoz[abildua|at]ehu.eus
other: 
Real scenario (hospital)
Date: 
2017