| Summary: | Objective: To investigate the application value of the random forest algorithm in identifying key risk factors associated with adverse nursing events, to thereby provide clinical evidence for the early recognition and intervention of potential risks. Methods: A retrospective analysis was conducted on 659 cases of adverse nursing events among inpatients at a tertiary grade A hospital. Using Python programming language integrated with the Scikit-learn library and implemented via the Jupyter environment, a random forest prediction model was developed. The dataset was partitioned into training and validation sets at a 3 : 1 ratio, followed by model training and evaluation to assess predictive performance. Results: The incidence rate of adverse nursing events was 0.352%. Random forest analysis revealed that the patient's fall risk score, Barthel index, and months of clinical experience among newly recruited nurses exhibited the higher variable importance. Additional influencing factors included patient age, department, shift time, nurse qualification level, patient compliance, nurse decision-making capability, and hospital ward area. The model achieved an accuracy of 74.7%, a recall rate of 75%, an F1 score of 72%, and an area under the receiver operating characteristic curve of 0.89. Conclusions: The application of the random forest algorithm in analyzing critical risk factors enables targeted early warning strategies. It way offer valuable insights for the prevention and management of adverse nursing events. This approach also provides a pathway for advancing high-quality productivity in nursing practice.
|
|---|