Radiology reports are a rich resource for advancing deep learning applications in medicine by leveraging the large volume of data continuously being updated, integrated, and shared. However, there are significant challenges as well, largely due to the ambiguity and subtlety of natural language. We propose a hybrid strategy that combines semantic-dictionary mapping and word2vec modeling for creating dense vector embeddings of free-text radiology reports. Our method leverages the benefits of both semantic-dictionary mapping as well as unsupervised learning. Using the vector representation, we automatically classify the radiology reports into three classes denoting confidence in the diagnosis of intracranial hemorrhage by the interpreting radiologist. We performed experiments with varying hyperparameter settings of the word embeddings and a range of different classifiers. Best performance achieved was a weighted precision of 88% and weighted recall of 90%. Our work offers the potential to leverage unstructured electronic health record data by allowing direct analysis of narrative clinical notes.

Pediatric firearm-related injuries pose a significant public health problem in the United States, yet the associated financial burden has not been well described. This is the first study examining national data on the cost of initial hospitalization for pediatric firearm-related injuries. In this retrospective review, the Healthcare Cost and Utilization Project Kids Inpatient Database from the years 2003, 2006, 2009, and 2012 was used to identify all patients 18 years of age and under who were admitted with firearm-related injuries. We compared demographic and discharge-level data including injury severity score, hospital length of stay, income quartile, injury intent, and inflation-adjusted hospital costs across age groups (0-5, 6-9, 10-15, 16-18 years). There were approximately 4,753 pediatric firearm-related admissions each year, with a median hospitalization cost of $12,984 per patient. Annual initial hospitalization costs for pediatric firearm injuries were approximately $109 million during the study period. Pediatric firearm-related injuries predominately occured among older teenagers (74%, 16-18 years), males (89%), black individuals (55%), and those from the lowest income quartile (53%). We found significant cost variation based on patient race, income quartile, injury severity score, intent, hospital length of stay, disposition, and hospital region. Inflation-adjusted hospitalization costs have increased significantly over the study period (p < 0.001). Pediatric firearm-related injuries are a large financial burden to the United States healthcare system. There are significant variations in cost based on predictable factors like hospital length of stay and injury severity score; however, there are also substantial discrepancies based on hospital region, patient race, and income quartile that require further investigation.