Nomogram Prediction Model on hospital mortality in patients after surgery with non-traumatic intracerebral hemorrhage: a retrospective analysis of the MIMIC-IV database

Background The study aimed to investigate the association of clinical data in intensive care unit (ICU) with the hospital mortality, and constructed a prediction model on the prognosis in patients with non-traumatic intracerebral hemorrhage (ICH). Methods We extracted all clinical information in ICU of patients with non-traumatic ICH from the Medical Information Mart for Intensive Care IV (MIMIC-IV). Logistic regression analysis was used to analyzing the risk factors. A nomogram for predicting the hospital mortality was constructed with R software and validated. Afterwards, the effectiveness and accuracy of the model were tested and evaluated. Results In the MIMIC IV database, 2004 patients with non-traumatic ICH were included in the study, with 1403 in training cohort and 601 in validation cohort. The mortality of hospital was 20.0% and 20.1% in the training and validation cohort, respectively. Age, Sequential Organ Failure Assessment Score (SOFA score), maximal mean blood pressure (Mbp), maximal temperature, glumbg and white blood cell count(WBC) were independent risk factors. The patients with higher glumbg and WBC were more likely to have poor prognosis (glumbg: p < 0.001, OR 1.018, 95% CI 1.010–1.025; WBC: p = 0.006, OR 1.057, 95% CI 1.016–1.099). The nomogram for predicting model exhibited a good performance and calibration. The AUC was 0.8022(0.7741–0.8303). The calibration curve indicated a satisfactory level of fit. Conclusion Non-traumatic ICH was associated with high rate of mortality (20%). The nomogram constructed in the study included age, SOFA score, maximal Mbp, maximal temperature, glumbg and WBC. Therefore, dynamic monitoring of these factors may be beneficial in assessing the risk and predicting the prognosis of the patients admitted to ICU with non-traumatic ICH.