Approximately 25% of the millions of veterans (est. 8.92 million FY 2013) enrolled for care in Veterans Health Administration (VHA) have diabetes mellitus, and diabetic macular edema (DME) is the leading cause of vision loss in the adult diabetic population world-wide. Although diabetic retinopathy has been well-studied, comparatively little is known about the burden of DME. In fact, only two national prevalence studies and no national study on the incidence of DME in persons with type 2 diabetes have been conducted. Similarly many risk factors have been characterized for DR, but no large studies have established predictors for DME. Beyond the Medicare claims database, the VHA National Patient Care Database (NPCD) contains standardized administrative data for several aspects of patient care including diagnoses, procedures, medications, lab test results, vital signs, clinical text notes, and mortality. Because the VA uses teleretinal screening as routine clinical care for all patients with diabetes with these results included in the NPCD, the NPCD is an ideal source for studying the epidemiology of and risk factors for DME. This study proposes to determine the burden of diabetic macular edema, establish risk factors, and examine treatment outcomes in a previously extracted dataset on 1.98 million veterans who have undergone diabetic retinopathy screening at least once since 2004. Currently invaluable ophthalmic data are encoded in unstructured clinical encounter notes in the Computerized Patient Record System (CPRS), and no validated automated extraction method exists to capture these data elements. An automated extraction method using natural language processing will be created and validated to unlock key ophthalmic variables. These text extraction methods will be applicable to extracting ophthalmology data from not only notes of patients with DME but also any ophthalmology clinical note. This will enable future large scale studies in ophthalmology using NPCD and be immediately valuable to the research community at large. The candidate, Dr. Aaron Lee, MD MSCI, is an ophthalmologist with subspecialty training in retina surgery with a strong background in computer science and epidemiology. His career goal is to become an independent clinician scientist studying diabetic eye disease with large-scale electronic medical record extracted data. While he possesses the foundational skills, he seeks to gain training in advanced statistics and natural language processing to unlock the data captured in unstructured clinical encounter notes. He has assembled an outstanding mentorship team under the primary mentor, Dr. Edward Boyko, MD MPH. This mentorship team includes renowned experts in clinical epidemiology, health informatics, ophthalmology, and natural language processing. This K23 will provide Dr. Lee the structured coursework, mentorship, and applied learning needed to acquire new research skills. He will leverage key local resources to carry out th...