Effect of AI-Supported Virtual Patient Chatbot Training on Nursing Students' Self-Perceived Clinical Decision-Making and Gynecological Cancer Knowledge

Participants received access to five sequential AI-supported educational virtual patient scenarios addressing cervical, endometrial, ovarian, vulvar, and vaginal cancers. The scenarios were completed during a two-week access period. Participants responded in free text, received immediate individualized formative feedback, and were allowed up to three response attempts per scenario. After completing all five training scenarios, participants completed the common virtual patient assessment and the immediate post-intervention outcome assessments.

A web-based educational platform integrating structured gynecological oncology scenarios, AI-supported virtual patient interaction, predefined performance criteria, sequential progression, and immediate individualized formative feedback. Participants completed five training scenarios during a two-week access period. The platform used the Claude Sonnet 4 model, API identifier claude-sonnet-4-20250514, throughout the study.

Participants did not receive access to the AI-supported training scenarios during the outcome assessment period. They completed the baseline assessments, the same common virtual patient assessment as the intervention group, and the immediate post-assessment knowledge and clinical decision-making measures. After completing all study outcome assessments, they were granted access to the AI-supported educational scenarios.

The CDMNS-Tr is a 40-item scale assessing self-reported clinical decision-making. Items are rated on a 5-point Likert scale ranging from 1 (never) to 5 (always), with eighteen negatively worded items being reverse-scored. Total scores range from 40 to 200, with higher scores indicating a higher self-perception of clinical decision-making abilities.

The researcher-developed test consists of 15 multiple-choice items assessing the pathophysiology, clinical manifestations, diagnosis, treatment, complications, and nursing care of gynecological cancers. Correct answers are scored as 1 and incorrect or unanswered items as 0. Total scores range from 0 to 15, with higher scores indicating greater knowledge. Change was evaluated by comparing baseline and immediate post-intervention total scores.