Background: Machine-learning techniques are useful for creating prediction models in clinical practice. This study aimed to construct a prediction model of postoperative 30-day mortality based on an automatically extracted electronic preoperative evaluation sheet. Methods: We used data from 276,341 consecutive adult patients who underwent non-cardiac surgery between January 2011 and December 2020 at a tertiary center for model development and internal validation, and another dataset from 63,384 patients between January 2011 and October 2021 at another center for external validation. Postoperative 30-day mortality was 0.16%. We developed an extreme gradient boosting (XGB) prediction model using only variables from preoperative evaluation sheets. Results: The model yielded an area under the curve of 0.960 and an area under the precision and recall curve of 0.216, which were 0.932 and 0.122, respectively, in the external validation set. The optimal threshold calculated by Youden’s J statistic had a sensitivity of 0.885 and specificity of 0.914. In an additional analysis with balanced distribution, the model showed a similar predictive value. Conclusion: We presented a machine-learning prediction model for 30-day mortality after non-cardiac surgery using preoperative variables automatically extracted from electronic medical records and validated the model in a multi-center setting. Our model may help clinicians predict postoperative outcomes.

Restrictive fluid management has been recommended for thoracic surgery. However, specific guidelines are lacking, and there is always concern regarding impairment of renal perfusion with a restrictive policy. The objective of this study was to find the net intraoperative fluid infusion rate which shows the lowest incidence of composite complications (either pulmonary complications or acute kidney injury) in open thoracotomy. We hypothesized that a certain range of infusion rate would decrease the composite complications within postoperative 30 days. All patients (n = 1,031) who underwent open thoracotomy at a tertiary care university hospital were included in this retrospective study. The time frame of fluid monitoring was from the start of operation to postoperative 24 hours. The cutoff value of the intraoperative net fluid amount was 4-5 ml. kg −1 .h −1 according to the minimum p-value method, thus, patients were divided into Low (≤3 ml.kg −1 .h −1), Cutoff (4-5 ml.kg −1 .h −1) and High (≥6 ml.kg −1 .h −1) groups. The Cutoff group showed the lowest composite complication rate (19%, 12%, and 13% in the Low, Cutoff, and High groups, respectively, P = 0.0283; Low vs. Cutoff, P = 0.0324, Bonferroni correction). Acute respiratory distress syndrome occurred least frequently in the Cutoff group (7%, 3%, and 6% for the Low, Cutoff, and High groups, respectively, P = 0.0467; Low vs. Cutoff, P = 0.0432, Bonferroni correction). In multivariable analysis, intraoperative net fluid infusion rate was associated with composite complications, and the Cutoff group decreased risk (odds ratio 0.54, 95% confidence interval: 0.35-0.81, P = 0.0035). In conclusion, maintaining intraoperative net fluid infusion at 4-5 ml.kg −1 .h −1 was associated with better results in open thoracotomy, in terms of composite complications, compared to more restrictive fluid management. Postoperative pulmonary complications are the leading cause of morbidity, mortality, and prolonged hospital stay after thoracic surgery 1. Fluid overload is one of the major risk factors 2-4 and is receiving significant attention as a preventable risk factor. Therefore, fluid restriction is usually recommended during and after thoracic surgery 2-6. Despite the well-known risk of fluid overload, few studies have provided a guideline for intraoperative fluid management. Some experts insist on extreme restriction, and even zero fluid balance, for thoracic surgery 5. However, evidence for the efficacy of zero balance only exists in the context of intensive care units (ICUs) 7. Guidelines for ICU patients may not be applicable to patients undergoing a major operation because of anesthesia-induced vasodilation, increased insensible loss from the open thoracic cavity, surgical trauma induced fluid shift, and hemodynamic instability related to surgery. In addition to the lack of solid evidence, impairment of perfusion of major organs, especially the kidney, is another major concern with restrictive fluid management.