Abstract Improving the Diagnosis of Liver Disease in Primary Care Patients with Abnormal Liver Function Tests Through Predictive Modeling Reducing diagnostic error has been identified by the Institute of Medicine as a top national priority. Diagnostic errors pervade all of healthcare, with the average individual experiencing one major error during their lifetime. Therefore, improving the diagnostic process and reducing diagnostic error is not only highly appropriate for all patients, but will play a crucial role in optimizing the quality and value of healthcare delivery in the United States.1 Liver disease, with complications including acute liver failure, cirrhosis, and liver cancer ranks as a leading cause of death in America and over recent years has had a significant climb in age-adjusted mortality, while death rates from heart disease and cancer have fallen.2 Despite the increasing preventability of liver-related conditions through early recognition and treatment, the toll of chronic and end stage liver disease continues to rise.3 The traditional diagnostic process, a synthesis of information gathered from history, physical exam, and laboratory testing, performs poorly in the detection of early liver disease.4,5 Instead, clinicians rely more heavily on laboratory studies, and liver function tests (LFTs) in particular.6 Abnormal LFTs are among the most frequently encountered findings in medicine.7,8 Currently, primary care clinicians currently lack the ability to consistently identify liver-related disease from these abnormalities.9-12 Preliminary data in primary care emphasize the immense scope of the problem; in studies from Europe, LFTs have been found elsewhere to be abnormal in nearly 1 in 5 people.13,14 In our preliminary studies, we have up to 40% of patients seen in an academic primary care clinic possessed at least one abnormal LFT. Further, these abnormal liver tests are inappropriately or inadequately followed-up. These data and our own experience indicate that primary care physicians (PCPs) lack the resources to reliably identify and accurately diagnose liver-related diseases amongst these many abnormal LFTs. In this proposal, the candidate and his mentorship team seek to harness inter-professional teamwork and information technology to reduce diagnostic error. They will identify clinical and demographic variables of patients with abnormal LFTs associated with specific liver-related diagnoses in primary care (Aim 1). Additionally, they will develop and validate a predictive model to identify patients with abnormal LFTs at risk for liver-related diagnoses (Aim 2). Lastly, they will create a decision support tool application to aid PCPs confronted with abnormal LFTs to promptly and accurately diagnose liver disease (Aim 3).