Objective To explore the application effect of the “reverse outline AI （ROAI）” in the teaching of scientific

research writing for undergraduate students in clinical medicine. Methods A single-center retrospective study design was employed. Teaching records from two academic years of the general education elective course “case decoding—exploring clinical research”（course code： EC08201） were retrieved， including a total of 81 undergraduate students. Based on the actual teaching model， students were divided into a traditional lecture group （n=39， implementing traditional lectures and after-class grading） and a ROAI teaching group （n=42， incorporating the ROAI method）. The logical scores， writing time， and teaching satisfaction of the two groups of students in their final papers were compared and analyzed. Results The logicality score of the ROAI teaching group’s papers was （89.30±4.27） points， significantly higher than that of the traditional lecture group’s （74.96±6.11） points （P<0.001）. The average total writing time of the ROAI teaching group was （215.07± 22.23） minutes， significantly shorter than that of the traditional lecture group’s （286.79±25.03） minutes （P<0.001）. The complete satisfaction rate of the teaching mode of the ROAI teaching group reached 92.86%， which was better than that of the traditional lecture group’s 74.36% （P=0.033）. Conclusion The ROAI model has re-engineered the research writing process， significantly improving writing efficiency and logical thinking， and providing an empirical paradigm for the digital transformation of new medical higher education.

Generative artificial intelligence, Reverse outline AI （ROAI）, Scientific writing, Logical reconstruction,