Who Makes It Through the Funnel? Sociodemographic Drivers of Recruitment Completion into an AI-Ready Digital Speech Bank

BackgroundDigital recruitment methods offer promising opportunities to address persistent challenges in clinical research participation, particularly in specialized fields like neurology. However, the impact of digital approaches across different socioeconomic and demographic groups remains inadequately understood. This study analyzed participant recruitment pathways in a digital neurology research study to identify sociodemographic factors associated with participation outcomes. MethodsWe conducted a longitudinal analysis of 5,846 patients invited to participate in a remote speech capture study for neurological disease research between March and July 2024. Using data from Qualtrics, PTrax, and our recording platform, we tracked participant progression through multiple recruitment checkpoints. Socioeconomic status was assessed using the Housing-based Socioeconomic Status (HOUSES) index and Area Deprivation Index (ADI). We examined associations between participation pathways and demographic factors including age, sex, geographic location, and socioeconomic indices using Kruskal-Wallis and Wilcoxon rank-sum tests. ResultsOnly 415 participants (7.1%) completed all study requirements. Participants from neighborhoods with higher socioeconomic disadvantage (higher ADI national ranks) were significantly less likely to express interest in initial invitations (median ADI 45.0 vs. 42.0 for responders, p<0.001). Urban participants completed enrollment faster than those from rural areas or urban clusters (median 32.0 days vs. 41.0 and 40.0 days, p=0.011). Contrary to expectations, younger participants were more likely to drop out at multiple recruitment stages, with the median age increasing from 63 years in the invited cohort to 66.3 years among completers. Female participants required more time to complete enrollment compared to males (median 38.5 days vs. 32.0 days, p=0.010). While neighborhood-level socioeconomic status significantly influenced participation, individual housing circumstances showed no significant association across recruitment stages. ConclusionsDigital recruitment methods in neurological research do not automatically overcome traditional barriers to participation and may introduce new disparities related to the digital divide. The significant associations between participation outcomes and sociodemographic factors--particularly neighborhood socioeconomic status, geographic location, age, and sex--highlight the need for targeted recruitment strategies. Researchers should implement multi-channel approaches, design age-specific engagement strategies, address geographic disparities, and consider socioeconomic factors to enhance the inclusivity and effectiveness of digital recruitment in neurological research.