SHM Converge Scientific Abstract ePoster Gallery-A Pilot Randomized Controlled Trial of a Novel Ai Chatbot to Improve Code Status Discussions: A Mixed-Methods Study

Pdf Summary

The poster describes a pilot feasibility study of ChatMD, an AI chatbot created to help train medical residents in code status discussions (CSDs), which are often difficult and frequently mishandled. Because patients commonly misunderstand CPR and overestimate its benefits, poor CSDs can lead to decisions that do not match patient values. Traditional training methods such as standardized patients and role play can improve skills but are costly and difficult to scale.<br /><br />To address this gap, the authors developed ChatMD and tested it in a randomized controlled study with internal medicine residents. Participants first attended a lecture, then were randomized to either ChatMD training or control. The ChatMD intervention included a two-hour session with supervised feedback during the first hour and unsupervised practice during the second. Learners could also use ChatMD before a standardized patient encounter, which was scored with an 18-item modified validated checklist covering both general communication and CSD-specific skills.<br /><br />Results suggested a small improvement with ChatMD, especially in CSD-specific items, though differences were not statistically significant due to the small sample size. General communication skills appeared to improve more naturally during residency, while CSD-specific skills seemed to require targeted training. Residents reported overwhelmingly positive experiences overall. Thematic analysis showed that ChatMD provided psychological safety and allowed repeated practice, but learners wanted more emotional realism and automated feedback.<br /><br />Overall, the study suggests AI-based simulation may be a scalable, cost-effective way to support training in difficult end-of-life conversations and improve learners’ confidence and patient-centered communication.

Asset Subtitle

Jessica Lichter

Meta Tag

Author List Alex Beutel, Beverly Forsyth, Dana Trottier, Jessica Lichter, Julie Kanevsky

Category Research

Concept Code Status Discussion

Concept ChatMD

Concept Simulation Tool

Concept Randomized Controlled Trial

Concept Patient Values

Distinguished Finalist

Presenter Organization Icahn School of Medicine at Mount Sinai and NYC Health & Hospitals/Elmhurst

Presenting Author Jessica Lichter

Track Education

Keywords

ChatMD

medical resident training

code status discussions

AI chatbot

end-of-life communication

randomized controlled study

standardized patient encounter

patient-centered communication

simulation-based learning

CPR misconceptions

Code Status Discussion

ChatMD

Simulation Tool

Randomized Controlled Trial

Patient Values