PROJECT SUMMARY/ABSTRACT Primary Immune Deficiency (PID) is a debilitating condition that affects one in 1,200 persons in the US. Although PID has been historically perceived to predominantly affect non-Hispanic white population, emerging evidence suggests stark disparities in the diagnosis of PID among racial and ethnic minorities. Of note, while past reporting of PID found that the majority of patients were non-Hispanic whites, implementation of newborn screening for certain types of PID found no difference in disease prevalence in any ethnic group. Differential access to diagnostic testing and specialty care, as well as diagnostic bias rooted in the prevailing belief that PID primarily affects non-Hispanic white population, may have contributed to the underdiagnosis of PID among minority populations. To date, there remains scant data on the risk factors of diagnostic delay in minority patients with PID, and there are currently no published studies investigating impediments to diagnosis and how they can be addressed. Delay in the treatment of PID can result in serious health problems, including organ damage and death. There is therefore an urgent need to address disparities in the diagnosis of PID. Our long-term goal is to improve timely diagnosis and treatment of PID in underserved populations. To achieve this goal, we propose the following specific aims: (1) Identify patterns of diagnostic delay in PID among racial and ethnic minorities; (2) Identify barriers to early diagnosis of PID among racial and ethnic minorities; and (3) pilot a targeted intervention to improve awareness of disparities in PID diagnosis. We will combine analyses of electronic health record (EHR) data (Aim 1) with qualitative analysis of patients’ lived experience and real- world perspectives from healthcare providers to understanding barriers to early diagnosis of PID (Aim 2). Additionally, we will apply advanced machine learning analysis as an innovative approach to enable a more comprehensive understanding of the patterns of diagnostic delay. PID is a complex group of diseases with highly variable clinical manifestations. We anticipate that the application of machine learning methods to EHR data can facilitate identification of under-recognized patterns of diagnostic delay and will enable us to learn from large clinical datasets in a scalable manner. Integrating knowledge from these analyses, we will then develop and evaluate an educational outreach program targeting healthcare providers to raise awareness of the disparities in PID diagnosis experienced by minority groups (Aim 3). The study will be conducted at 2 major healthcare systems in Massachusetts: Mass General Brigham and Boston Medical Center. This body of work represents the first systematic effort to investigate the patterns of, and barriers to, early recognition of PID among minority populations in the US. To the best of our knowledge, the proposed pilot educational outreach program will be the first educa...