Mentor Match: An Artificial Intelligence–Enabled Mentor Access Platform to Augment Career Development in Emergency Medicine

Emergency medicine trainees often struggle to identify appropriate faculty mentors and research collaborators because information is fragmented and discovery relies on informal networks. This lack of transparency can delay scholarly engagement and disproportionately disadvantage trainees without prior academic exposure. Limited mentorship discovery may also constrain exposure to emerging career niches. Emerging AI tools offer an opportunity to improve mentorship discovery by lowering informational barriers and supporting more equitable access to academic opportunities.

To describe the design and early implementation of a mentorship discovery platform (MentorMatch) intended to improve access to mentorship and scholarly opportunities for emergency medicine trainees. The project focuses on facilitating transparent mentor discovery while minimizing data burden and bias, and on evaluating whether the system surfaces plausible mentorship starting points and supports more efficient navigation of mentorship pathways.

MentorMatch is a web-based platform where trainees enter free-text scholarly and career interests. An AI-powered API is used to extract and normalize relevant academic themes from trainee input. Faculty mentors are retrieved based on overlap between these themes and publicly available scholarly publications and educational roles. Results are presented with supporting evidence to aid human judgment rather than automate matching. Evaluation includes structured faculty review and pilot deployment.

This submission reports feasibility, implementation considerations, and early internal feedback. Planned evaluation includes (1) discovery validity, assessed through review of whether recommendations represent plausible mentorship starting points based on documented evidence; and (2) real-world utility, assessed through time to identify potential mentors, time to initiate contact, number of mentorship connections initiated, and learner-reported confidence. A subset of suggested mentors will provide alignment feedback to contextualize results.

Access to mentorship is a critical determinant of scholarly engagement and career development in emergency medicine, yet mentorship discovery remains largely informal, opaque, and dependent on personal networks. Trainees often struggle to identify appropriate faculty mentors because information about faculty expertise is fragmented across websites, publications, and word of mouth, disproportionately disadvantaging those without prior academic exposure. Despite increasing interest in research, innovation, and emerging career pathways, few tools exist to support structured mentorship discovery. MentorMatch was developed to address this gap by improving the transparency and accessibility of faculty expertise for trainees. Rather than predicting mentorship outcomes or assigning matches, MentorMatch focuses on improving the discovery process itself. Trainees describe their scholarly and career interests in natural language, which are processed using an AI-powered API to extract and normalize relevant academic themes and expand related concepts. Faculty mentors are retrieved based on overlap between these themes and publicly available scholarly publications and web profiles. Mentor suggestions are presented as an ordered set based on relevance signals, accompanied by supporting evidence, allowing trainees to explore alignment and initiate contact autonomously. The platform is intentionally lightweight, minimizes trainee data retention, and integrates easily into existing academic environments. Evaluation is designed to align with the educational goal of mentorship discovery and focuses on two complementary domains. Discovery validity is assessed through structured faculty review of mentor–interest pairings, examining whether the system surfaces plausible mentorship starting points based on documented evidence rather than predicting mentorship success. To provide additional context, a subset of suggested mentors are asked whether recommendations align with their current work and mentoring interests, offering a complementary self-alignment signal without serving as a determinant of discovery validity. Real-world utility is assessed through planned pilot deployment metrics, including time required for trainees to identify potential mentors, time to initiate contact, number of mentorship connections initiated, and learner-reported confidence navigating mentorship pathways. Together, these measures evaluate whether the system meaningfully reduces friction in mentorship discovery while preserving human oversight and learner autonomy. Our team is currently focused on system design, implementation considerations, and planned evaluation. MentorMatch represents a scalable, low-cost approach to improving equitable access to mentorship and academic opportunities within emergency medicine and has potential applicability across other medical specialties.