Semantic based Clinical Notes Mining for Factual Information Extraction

The existence of a substantial amount of clinical notes has raised significant demand for clinical text processing and information extraction. Clinical notes are one of the most common forms of clinical documentation and an abundant source of patient information. A list of diseases mentioned in the patient clinical notes is one of the essential factual information for experts. However, checking this information manually is a cumbersome and time consuming task. Therefore, we propose an automatic mechanism that extracts disease relevant information from clinical notes using natural language processing techniques. The methodology extracts disease information that is diagnosed in the patient and neglects the negated disease names presented in the notes. The initial evaluation results of the methodology on MTSamples provided dataset with 97.81% accuracy show its effectiveness and applicability for the mentioned goal. The methodology benefits and assists human experts by extraction disease relevant information from patient notes in a minimum time frame.