Detecting Conversation Topics in Recruitment Calls of African American Participants to the All of Us Research Program Using Machine Learning: Model Development and Validation Study
Bibliographic record
Abstract
BACKGROUND: Advancements in science and technology can exacerbate health disparities, particularly when there is a lack of diversity in clinical research, which limits the benefits of innovations for underrepresented communities. Programs like the All of Us Research Program (AoURP) are actively working to address this issue by ensuring that underrepresented populations are represented in biomedical research, promoting equitable participation, and advancing health outcomes for all. African American communities have been particularly underrepresented in clinical research, often due to historical instances of research misconduct, such as the Tuskegee Syphilis Study, which have deeply impacted trust and willingness to participate in research studies. With the US population becoming increasingly diverse, it is crucial that clinical research studies reflect this diversity to improve health outcomes. However, limited data and small sample sizes in qualitative studies on the inclusion of underrepresented groups hinder progress in this area. OBJECTIVE: The goal of this paper is to analyze recruitment conversations between research assistants (RAs) and potential participants in the AoURP to identify key topics that influence enrollment. By examining these interactions, we aim to provide insights that can improve engagement strategies and recruitment practices for underrepresented groups in biomedical research. METHODS: Our study design was an observational, retrospective approach using machine learning for content analysis. Specifically, we used structural topic modeling to identify and compare latent topics of conversation in recruitment calls by Morehouse School of Medicine RAs between February 2021 and April 2022 by estimating expected topic proportions in the corpus as a function of enrollment and participation in AoURP. RESULTS: In total, our model estimated 45 topics of which 12 coherent topics were identified. Notable topics, that were more likely to occur in conversations between RAs and participants that enrolled and participated, include closing or following up to schedule an appointment, COVID-19 protocols for in-person visits, explaining precision medicine and the need for representation, and working through objections, including concerns about costs, insurance, care changes, and health fears. Topics among potential participants who did not enroll include technical challenges and describing physical measurement visits (eg, collection of basic physical data, such as height, weight, and blood pressure). CONCLUSIONS: Using an approach that leverages machine learning to identify topical structure and themes with limited human subjectivity is a promising strategy to identify gaps in, and opportunities to improve, the recruitment of underserved communities into clinical trials.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.030 | 0.049 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.003 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.003 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".