Racial-ethnic minorities (REM) and lesbian, gay, bisexual, transgender, and queer (LGBTQ) individuals experience high levels of psychological distress. Psychological treatments can be effective in addressing mental health concerns, but disparities in quality of care still exist. Although systemic and institutional factors contribute to disparities in care, mental health providers are also critical to examine. A primary focus of efforts to understand and reduce provider contributions to mental health care disparities has been to examine cultural competency (CC), which involves a provider?s ability to navigate the cultural aspects of clinical interactions. Patient ratings of CC are generally associated with treatment outcomes and therapeutic processes. While patient perceptions of provider CC are important, a reliance on retrospective patient ratings limits what we know about how cultural identities are discussed, and the language that constitutes culturally sensitive care. Many studies of provider CC also require observers or patients to make complex judgments based on internal provider characteristics that are not reliably observable (e.g. rate provider awareness of their own cultural values). More studies are needed that examine patient-provider interactions in treatment in order to assess the impact of specific provider behaviors, and how they relate to perceptions of provider CC. Recently, Natural Language Processing (NLP) models have been applied to psychotherapy conversations to automatically capture the use of evidence based treatments, topics of conversation, empathy, and emotional expression. Prior research demonstrating the feasibility of automatically identifying topics of conversation in psychotherapy suggest that NLP models could be trained to automatically identify specific moments in sessions where patients and providers are talking about cultural issues. NLP models could allow researchers to not only examine how specific patterns of provider-patient interactions drive CC, but might also provide rapid feedback to providers, and in turn help address disparities in care. The purpose of the current study is to do the foundational work to develop and evaluate NLP tools that capture the cultural content of provider-patient interactions among REM and LGBTQ patients. First, utilizing 32,436 labeled talk turns from 200 psychotherapy sessions we will evaluate the accuracy of NLP models in recognizing the discussion of cultural topics in psychotherapy. Second, we will use NLP models to explore differences in the content of 1,235 psychotherapy sessions that were rated as highly positive or negative on a measure of cultural competence.
Although disparities in the quality of mental health treatment for racial-ethnic minority (REM) and lesbian, gay, bisexual, transgender, and queer (LGBTQ) patients are well known, to date there are no tools that can identify specific patterns of provider-patient interactions that drive disparities in care. This project will evaluate the ability of Natural Language Processing (NLP) models to recognize discussion of cultural topics in psychotherapy among REM and LGBTQ patients, and explore differences in patient-provider interactions with low and high patient ratings of provider cultural competency.