Incarceration is a significant social determinant of health, contributing to high morbidity, mortality, and racialized health inequities. However, incarceration status is largely invisible to health services research due to inadequate clinical electronic health record (EHR) capture. This study aims to develop, train, and validate natural language processing (NLP) techniques to more effectively identify incarceration status in the EHR.

The study population consisted of adult patients (≥ 18 y.o.) who presented to the emergency department between June 2013 and August 2021. The EHR database was filtered for notes for specific incarceration-related terms, and then a random selection of 1,000 notes was annotated for incarceration and further stratified into specific statuses of prior history, recent, and current incarceration. For NLP model development, 80% of the notes were used to train the Longformer-based and RoBERTa algorithms. The remaining 20% of the notes underwent analysis with GPT-4.

There were 849 unique patients across 989 visits in the 1000 annotated notes. Manual annotation revealed that 559 of 1000 notes (55.9%) contained evidence of incarceration history. ICD-10 code (sensitivity: 4.8%, specificity: 99.1%, F1-score: 0.09) demonstrated inferior performance to RoBERTa NLP (sensitivity: 78.6%, specificity: 73.3%, F1-score: 0.79), Longformer NLP (sensitivity: 94.6%, specificity: 87.5%, F1-score: 0.93), and GPT-4 (sensitivity: 100%, specificity: 61.1%, F1-score: 0.86).

Our advanced NLP models demonstrate a high degree of accuracy in identifying incarceration status from clinical notes. Further research is needed to explore their scaled implementation in population health initiatives and assess their potential to mitigate health disparities through tailored system interventions.

To improve maternal health outcomes, increased diversity is needed among pregnant people in research studies and community surveillance. To expand the pool, we sought to develop a network encompassing academic and community obstetrics clinics. Typical challenges in developing a network include site identification, contracting, onboarding sites, staff engagement, participant recruitment, funding, and institutional review board approvals. While not insurmountable, these challenges became magnified as we built a research network during a global pandemic. Our objective is to describe the framework utilized to resolve pandemic-related issues.

We developed a framework for site-specific adaptation of the generalized study protocol. Twice monthly video meetings were held between the lead academic sites to identify local challenges and to generate ideas for solutions. We identified site and participant recruitment challenges and then implemented solutions tailored to the local workflow. These solutions included the use of an electronic consent and videoconferences with local clinic leadership and staff. The processes for network development and maintenance changed to address issues related to the COVID-19 pandemic. However, aspects of the sample processing/storage and data collection elements were held constant between sites.

Adapting our consenting approach enabled maintaining study enrollment during the pandemic. The pandemic amplified issues related to contracting, onboarding, and IRB approval. Maintaining continuity in sample management and clinical data collection allowed for pooling of information between sites.

Adaptability is key to maintaining network sites. Rapidly changing guidelines for beginning and continuing research during the pandemic required frequent intra- and inter-institutional communication to navigate.

Oxytocin has been proposed to mediate amygdala dysfunction associated with altered emotion processing in schizophrenia, but the contribution of oxytocin pathway genes is yet to be investigated.

To identify potential different contributions of three oxytocin receptor polymorphisms (rs53576, rs237902 and rs2254298) between patients with schizophrenia spectrum disorders (SCZ), affective spectrum disorders (AD) and healthy controls (HC).

In a total of 346 participants (104 with SCZ, 100 with AD, and 142 HC) underwent genotyping and functional magnetic resonance imaging (fMRI) during an emotional faces matching paradigm. Genetic association analyses were performed to test the possible effects on task-induced BOLD amygdala response to fearful/angry faces.

In participants with SCZ, the rs237902 G allele was associated with low amygdala activation (left hemisphere: b= −4.99, Bonferroni corrected P=0.04) and interaction analyses showed that this association was disorder specific (left hemisphere: Bonferroni corrected P=0.003; right hemisphere: Bonferroni corrected P=0.03). There were no associations between oxytocin polymorphisms and amygdala activation in the total sample, among AD patients or HC.

Rs237902 was associated with amygdala activation in response to fearful/angry faces only in patients with SCZ. Our findings indicate that the endogenous oxytocin system could serve as a contributing factor in biological underpinnings of emotion processing and that this contribution is disorder specific.