Summarizing Behavioral Health Electronic Health Records Using a Natural Language Processing Pipeline

Doctors and nurses have limited time between patients to analyze and review a patient's documents to provide a quality assessment. This problem is supplemented by the existence of Electronic Health Records (EHR), which are essentially digital files regarding the patient. However, the length and content of each document vary greatly, reducing the effectiveness. Therefore, this research aims to reduce the need for medical professionals to manually search for crucial information about the patient's health history. We intend to accomplish our objective using various natural language processing (NLP) techniques to break down digital documents into smaller subtasks, such as event extraction and abstractive summarization, to provide a concise summary. Our proposed system intends to streamline the process and nullify the issue of having to read lengthy documents to locate essential information, which can affect the overall efficiency and quality of the patient's care. In the future, we intend to migrate to a closed-domain event extraction model and implement a timeline for easier visualization.