Development and validation of a machine learning-based predictive model for compassion fatigue in nursing interns: A cross-sectional study with latent profile analysis

Background Exposure to compassion fatigue during internships can significantly impact on nursing students’ future career trajectories and their intention to stay in the nursing profession. Accurately identifying nursing students at high risk of compassion fatigue is vital for timely interventions. However, existing assessment tools often fail to account for within-group variability and lack predictive capabilities. To develop and validate a predictive model for detecting the risk of compassion fatigue among nursing students during their placement. Design: A cross-sectional study design. Methods Data from 2256 nursing students in China between December 2021 and June 2022 were collected on compassion fatigue, professional identity, self-efficacy, social support, psychological resilience, coping styles, and demographic characteristics. The latent profile analysis was performed to classify compassion fatigue levels of nursing students. Univariate analysis, least absolute shrinkage and selection operator regression analysis were conducted to identify potential predictors of compassion fatigue. Eight machine learning algorithms were selected to predict compassion fatigue, and the performance of these machine learning models were evaluated using calibration and discrimination metrics. Additionally, the best-performing model from this evaluation was selected for further independent assessment. Results A three-profile model best fit the data, identifying low (55.73%), moderate (32.17%), and severe (12.10%) profiles for compassion fatigue. The area under the curve values for the eight machine learning models ranged from 0.644 to 0.826 for the training set and from 0.651 to 0.757 for the test set. The eXtreme Gradient Boosting performed best, with area under the receiver operating characteristic curve values of 0.840, 0.768, and 0.731 in the training, validation, and test sets, respectively. SHAP analysis clarified the model’s explanatory variables, with psychological resilience, professional identity, and social support being the most significant contributors to the risk of compassion fatigue. A user-friendly, web-based prediction tool for calculating the risk of compassion fatigue was developed. Conclusions The eXtreme Gradient Boosting classifier demonstrates exceptional performance, and clinical implementation of the online tool can provide nursing managers with an effective means to manage compassion fatigue.