Natural Language Processing to aid in genomic diagnostics

Electronic Health Records (EHRs) have been widely used to store health information related to patients digitally. EHR data usually consists of information written in textual form by the practitioner during each session with each patient. However, this unstructured free text data is challenging for computers to effectively extract valuable information to help practitioners make the best treatment plans for their patients. 

To address this challenge, this project aims to leverage Natural Language Processing (NLP) and Machine Learning (AI) techniques to effectively deal with unstructured EHR clinical data. More specifically, we are interested in revealing insight into EHRs which can accurately link phenotypic descriptions of disease with genomic diagnosis in order to further enhance analysis, prediction, and decision-making for better healthcare for epileptic patients. The data would be accessed from the Irish Epilepsy Electronic Patient Record, hosted by Beaumont Hospital, under the necessary agreements, including informed consent.