Radiology Clinical Notes Mining Using Weighted Association Rules

Electronic health record (EHR) serves to capture the patients' medical conditions and detailed visits information. The structured part of EHR can be used to serve administrative and financial management of patients. However, the unstructured part, that contains the interventions applied to the patient and transcribed in textual format, have unexplored knowledge due to the narrative format. Unlike conventional data mining techniques that can be applied to structured databases, applying large and automated analysis on clinical notes can potentially provide better support for medical decision making for an individual patient or for collective of patients. It can also allow for discovery of emerging associations to explore relationships among the patients using the data stored in the unstructured text of the EHR.