Analyzing free-form medical data such as pathology reports or physicians notes and comments presents additional challenges. Unlike with structured data (e.g., numerical, check-boxes, ICD-10 codes), there are countless ways that physicians could express the same concept in unstructured text. In this thesis, computational techniques were explored for automating the categorization of medical documents related to pediatric appendicitis. In the first project, a computational model was built to detect emergency department notes which contained features of Pediatric Appendicitis Score. This model achieved a 0.8391 F-Score compared to human performance and also outperformed the previous computational method (0.3435 F-Score). In the second project, a model was constructed to identify appendectomy pathology reports which were negative for appendicitis. This model obtained an F-Score of 0.9960. In many cases, hospitals rely on manual chart review for such tasks; this thesis presents an alternative computational approach using statistical natural language processing.