Enhancing rule-based text classification of neurosurgical notes using filtered feature weight vectors

Clinicians need to record clinical encounters in written or spoken language, not only for its work-flow naturalness but also for its expressivity, precision, and capacity to convey all required information, which codified structured data is incapable of. Therefore, the structured data which is required for aggregation and analysis must be obtained from clinical text as a later step. Specialised areas of medicine use their own clinical language and clinical coding systems, resulting in unique challenges for the extraction process. Rule-based information extraction techniques have been used effectively in commercial systems and are favoured because they are easily understood and controlled. However, there is promising research into the use of machine learning techniques for extracting information, and this research explores the effectiveness of a hybrid rule-based and machine learning-based audit coding system developed for the neurosurgical department of a major trauma hospital.