Development and validation of a predictive model for continuous renal replacement therapy in sepsis patients using the MIMIC-IV database
Abstract To develop and validate a dynamic nomogram for predicting the need for continuous renal replacement therapy (CRRT) in septic patients in the intensive care unit (ICU). Data were extracted from the MIMIC-IV 3.0 database and divided into a training set and a validation set in a 7:3 ratio. Rel...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-07-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-025-07647-6 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract To develop and validate a dynamic nomogram for predicting the need for continuous renal replacement therapy (CRRT) in septic patients in the intensive care unit (ICU). Data were extracted from the MIMIC-IV 3.0 database and divided into a training set and a validation set in a 7:3 ratio. Relevant risk factors were identified through LASSO regression, and a binary logistic regression model was subsequently developed. The CRRT risk nomogram was visualized using R language, with the DynNom package employed to create a dynamic nomogram. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), Harrell’s C-index, and calibration curves. The clinical utility of the model was evaluated via decision curve analysis (DCA). A total of 7361 septic patients were included in this study, of which 525 required CRRT. The study identified several predictive factors for CRRT, including respiratory rate, oxygen saturation, international normalized ratio (INR), activated partial thromboplastin time (APTT), creatinine, lactate, pH, body weight, renal disease, and severe liver disease. The C-index was 0.871. The AUCs for the training and validation sets were 0.87 (95% CI: 0.8535–0.8883) and 0.86 (95% CI: 0.8282–0.8887), respectively. The calibration curves demonstrated good predictive consistency. DCA confirmed the model’s significant clinical value. The dynamic nomogram is available for visualization at: https://zhong-hua-min-zu-wan-sui.shinyapps.io/CRRT_prediction_nomogram/ . We have developed a dynamic nomogram based on the MIMIC-IV database, incorporating 10 clinical features, to predict the probability of CRRT requirement in septic patients. Internal validation showed that this model exhibits robust predictive performance. |
|---|---|
| ISSN: | 2045-2322 |