Discipline: Computer Sciences & Information Management
Subcategory: STEM Research
Shanta Chowdhury - Prairie View A&M University
Co-Author(s): Xiangfang Li and Lijun Qian, Prairie View A&M University
Electronic Medical Record (EMR) is a digital version of storing patients’ medical history in textual format. It has shaped medical domain in such a promising way that it can gather all information into one place for physicians and patients. In order to handle these overwhelming EMR data, Named Entity Recognition (NER) is aimed at retrieving the entity terms which are related to disease, test, symptom, genes terms etc. Such type of information extraction can be a relief for providers and medical specialists to extract information automatically by avoiding unnecessary and unrelated information in EMR. However, challenges like incomplete syntax, numerous abbreviations make the recognition task very difficult. In this research project, novel deep learning methods are explored to address these challenges. Specifically, a multitask bi-directional recurrent neural network model is proposed for NER and the results show that it outperforms the state-of-the-art. The proposed model will be improved further by using a joint loss function and joint optimizer to reduce the training time and increase the prediction accuracy.
Funder Acknowledgement(s): NSF
Faculty Advisor: None Listed,