Knowledge Bank

The decreasing cost of both computing power and storage has lead to the widespread adoption of Electronic Medical Records in modern health care facilities. This movement has created a valuable source of diagnostic and patient information for medical research. While these data could provide valuable insight in the study of disease, patient privacy is of utmost concern and the protected patient information must be de-identified. This de-identification process can introduce error into the data-set.

This thesis implements de-identification techniques developed by The Ohio State University Medical Center Information Warehouse to de-identify electronic medical records, and analyzes the effect the de-identification of patient visit dates has on the reliability of the resulting data-sets. In addition, we develop a tool that allows researchers to characterize the reliability of de-identified data-sets based on patient population counts.