Development and External Validation of a Nurse-Friendly Machine Learning Model for Early Identification of Intradialytic Hypotension in ICU Patients Receiving Renal Replacement Therapy

Background Renal replacement therapy (RRT) is essential for critically ill patients in the intensive care unit (ICU), yet hypotension remains its most common complication, increasing the risk of organ hypoperfusion and poor outcomes. No reliable tool currently exists to predict hypotension risk before RRT initiation, particularly one designed for nursing integration. This study aimed t o develop and validate a nurse-friendly machine learning model using readily accessible pre-RRT variables to predict hypotension risk in ICU patients. Methods Eligible ICU patients were identified from the MIMIC-IV database and randomly split (7:3) into training and testing sets, with external validation from a Chinese ICU cohort. Twelve candidate predictors were selected based on meta-analysis evidence, clinical relevance, and ease of acquisition by nursing staff in routine ICU practice. Ten machine learning models were developed using predictors selected by the least absolute shrinkage and selection operator. Performance was assessed by AUROC, Brier score, and other metrics. SHapley Additive exPlanations ensured model interpretability, and the best model was deployed online. Results A total of 1,342 patients from the MIMIC-IV database and 133 patients from the external cohort were included in the analysis. The GBM demonstrated the most consistent overall performance across all datasets, with an AUROC of 0.801 in the external validation cohort. Key predictors included RRT modality, vasopressor use, mean and systolic blood pressures, age, lactate level, and the interval from ICU admission to RRT initiation. The final model was deployed using the Streamlit framework to facilitate real-time clinical interpretation and visualization. Conclusions The GBM-based model demonstrated strong predictive performance, nursing applicability, and clinical utility. Designed with a nursing-friendly approach, the model enables early risk stratification and supports proactive hemodynamic management before RRT initiation. Trial registration: Not applicable