Lessons Learned from Designing an Artificial Intelligence-Assisted Recruitment Strategy of Young Adults into ANCHOR: a Cannabis Nurse-Led Cannabis Use Disorder Harm Reduction Pilot

Bright Yaw Tamakloe, BSN, RN
University of Miami

Co-authors: Bright Tamakloe¹, Renessa Williams¹, Bria-Necole Diggs¹, Ciné Brown¹, Denise C. Vidot^1
1University of Miami

Background: Population-based data suggests that young adults between the ages 18-35 have the highest prevalence of daily cannabis use and may face elevated risk for problematic cannabis use. Stigma, limited access to cannabis-informed clinicians, and distrust of conventional outreach can impede engagement in harm-reduction services. Artificial intelligence (AI) may improve recruitment efficiency, but its use requires transparent, nurse-governed safeguards when  informing clinical care and practice.

Objective(s): To describe lessons learned and the feasibility of an AI-assisted recruitment workflow for Advancing Nurse-led Cannabis Harm Reduction (ANCHOR), a  nurse-led, technology-delivered brief cannabis intervention for young adults who self-medicate with cannabis and are enrolled in the Herbal Heart Study (HHS).

Methods: An artificial intelligence (AI)-assisted recruitment workflow was designed to identify, engage, screen, and schedule potentially eligible young adults from an established cohort of 150 daily cannabis consumers. The design process incorporated the AAMC’s Principles of Trustworthiness and was completed alongside the National Cannabis Nurse Task Force (CNTF). The workflow included four linked components: (1) AI-assisted drafting of recruitment messages using nonjudgmental, harm-reduction language; (2) tailoring of messages according to participant characteristics, including cannabis-use patterns, CUDIT-R scores, reasons for use, and desire to reduce consumption; (3) automated preliminary eligibility screening using prespecified study criteria; and (4) scheduling and follow-up support through standardized invitations, reminders, and response pathways. The CNTF and HHS staff reviewed AI-generated language, verified eligibility, and retained responsibility for participant communication.

Results: The design process produced a structured, reproducible recruitment workflow that integrates personalized outreach, preliminary screening, scheduling, and follow-up within a single recruitment pathway. AI assistance improved recruitment efficiency by reducing the manual effort required to draft and adapt outreach messages, supporting identification of potentially eligible participants, and standardizing routine screening and scheduling tasks. AI-generated messages required nurse review to ensure clinical appropriateness, accuracy, and alignment with harm-reduction principles. Tailoring was most useful when limited to study-relevant participant characteristics and approved message templates. Automated screening increased efficiency but did not replace staff verification of eligibility or informed consent. Additional priorities included minimizing the personal information entered into AI systems, maintaining an auditable record of message revisions, monitoring differential response across participant groups, and clearly disclosing the role of automation in recruitment.

Conclusions: AI-assisted, nurse-governed recruitment may efficiently connect high-need young adults with cannabis harm-reduction services. Transparent messaging, human oversight, and monitoring for differential recruitment yield are essential implementation safeguards.