Nomograms Predicting Self-Regulated Learning Levels in Chinese Undergraduate Medical Students

The purpose of this study was to construct a multi-center cross-sectional study to predict self-regulated learning (SRL) levels of Chinese medical undergraduates.

We selected medical undergraduates by random sampling from five universities in mainland China. The classical regression methods (logistic regression and Lasso regression) and machine learning model were combined to identify the most significant predictors of SRL levels. Nomograms were built based on multivariable models. The accuracy, discrimination, and generalization of our nomograms were evaluated by the receiver operating characteristic curves (ROC) and the calibration curves and a high quality external validation.

There were 2052 medical undergraduates from five universities in mainland China initially. The nomograms constructed based on the non-overfitting multivariable models were verified by internal validation (C-index: learning motivation: 0.736; learning strategy: 0.744) and external validation (C-index: learning motivation: 0.986; learning strategy: 1.000), showing decent prediction accuracy, discrimination, and generalization.

Comprehensive nomograms constructed in this study were useful and convenient tools to evaluate the SRL levels of undergraduate medical students in China.