(P211) ELUCIDATING HEALTH INEQUITIES AND RESEARCH GAPS IN HIV AND OTHER SEXUALLY TRANSMITTED INFECTIONS USING DATA MINING AND A LARGE LANGUAGE MODEL: A COMPUTATIONAL SCOPING REVIEW

Background:
Historical injustices such as the Tuskegee Syphilis Study and slow response to the AIDS crisis highlight a longstanding legacy of inequity in the US. Despite many advances over the last 50 years, HIV and other sexually transmitted infections (STIs) continue to pose a substantial public health challenge. We conducted a large-scale scoping review to characterize major trends and elucidate potential gaps in STI literature concerning health equity.

Methods:
Using a pre-validated search strategy, we extracted PubMed/MEDLINE articles (n=22,017 in English; published 1969-2023). We employed a three-pronged approach to characterize the literature and identified: (1) studied populations and study designs using MeSH terms, (2) condition mentions using named-entity recognition of abstracts and titles, and (3) emerging themes using a combination of BERTopic modeling and ChatGPT via prompt engineering to produce thematic labels and summaries. We also compared temporal trends in themes and studied conditions to national public health reporting of STI incidence and prevalence across population groups.

Results:
Despite a steady increase until 2013 and plateau through 2019, annual publication counts dropped precipitously over the last five years amidst a rise in STI incidence. Topic modeling identified 46 themes, 27 of which concerned HIV (e.g., voluntary testing/counseling, internet-based interventions for prevention among men who have sex with men, mobile phone-based interventions for treatment adherence, willingness to participate in vaccine trials, prevention of mother-to-child transmission, and the impact of COVID-19 on people with HIV). Rectal microbicide adherence, human papillomavirus vaccine uptake, and screening for chlamydia, syphilis, and gonorrhea were among other themes identified.

Conclusion:
STI literature concerning health equity has dramatically decreased since the COVID-19 pandemic and is dominated by HIV-related research. Coverage of different STIs and themes in the literature are highly variable over time and among minoritized populations, and do not necessarily reflect changes in incidence or prevalence.