ARTIFICIAL INTELLIGENCE–ASSISTED PARTICIPANT RECRUITMENT FOR PSYCHIATRIC CLINICAL TRIALS: A COMPARATIVE EVALUATION OF LEAD QUALITY AND ENGAGEMENT

Background

Efficient participant recruitment remains a persistent challenge in psychiatric clinical trials. Conventional recruitment vendors frequently distribute shared pools of leads to multiple clinical sites with limited customization of screening criteria, potentially reducing the efficiency of identifying eligible participants. Artificial intelligence assisted recruitment strategies coupled with state-of-the-art media buying powers may improve participant identification by integrating targeted digital advertising with automated prescreening and engagement assessment.

Methods

Over a 16-week observational evaluation, CalNeuro Research Group assessed recruitment outcomes associated with an AI-driven recruitment vendor, Trial For Good. Trial For Good utilizes proprietary strategies to advertise psychiatric clinical research opportunities and direct prospective participants to an AI-guided prescreening interface. Participants provided self-reported clinical and demographic information, which was analyzed using AIbased algorithms designed to assess engagement patterns and estimate participant intent and anticipated reliability in site communication. Recruitment metrics, including engagement and lead quality, were compared with those obtained through centrally managed vendor campaigns previously utilized by the site.

Results

AI-assisted recruitment was associated with increased lead volume and higher participant engagement relative to traditional vendor campaigns. Leads generated through the AI-guided prescreening process demonstrated greater responsiveness and communication with the clinical site. Additionally, the AI-based model reduced the proportion of unqualified participants referred for site-level screening, thereby decreasing the prescreening burden on site staff and allowing greater allocation of resources to enroll participant visits and ongoing study procedures. Three metrics were used to quantify the success of Trial For Good: leads generated, screening visits, and randomizations. Over a 16-week period from November 3, 2025, to March 1, 2026, Trial For Good generated 353 leads, compared with an average of 68 leads from common vendors such as Radius365-Flex and Splash Clinical. From these Trial For Good leads, 123 subjects came onsite for a screening visit, whereas an average of 50 subjects from Radius365-Flex and Splash Clinical leads attended onsite screening during the same period. Trial For Good resulted in 6 randomizations, compared with an average of 2 randomizations from the other vendors. Additionally, 4 other subjects from Trial For Good met all screening eligibility criteria but were unable to randomize due to external circumstances.

Conclusions

AI-assisted recruitment and automated prescreening may enhance the efficiency of participant identification in psychiatric clinical trials. By evaluating engagement signals and participant-reported clinical information prior to site referral, AI systems may improve lead quality and reduce screening burden for high-enrolling research sites. Further evaluation is warranted to determine the impact of AI-driven recruitment strategies on enrollment timelines and study completion rates.