Usefulness of Random Forest Algorithm in Predicting Severe Acute Pancreatitis

Abstract

Background and AimsThis study aimed to develop an interpretable random forest model for predicting severe acute pancreatitis (SAP).MethodsClinical and laboratory data of 648 patients with acute pancreatitis were retrospectively reviewed and randomly assigned to the training set and test set in a 3:1 ratio. Univariate analysis was used to select candidate predictors for the SAP. Random forest (RF) and logistic regression (LR) models were developed on the training sample. The prediction models were then applied to the test sample. The performance of the risk models was measured by calculating the area under the receiver operating characteristic (ROC) curves (AUC) and area under precision recall curve. We provide visualized interpretation by using local interpretable model-agnostic explanations (LIME).ResultsThe LR model was developed to predict SAP as the following function: -1.10-0.13×albumin (g/L) + 0.016 × serum creatinine (μmol/L) + 0.14 × glucose (mmol/L) + 1.63 × pleural effusion (0/1)(No/Yes). The coefficients of this formula were utilized to build a nomogram. The RF model consists of 16 variables identified by univariate analysis. It was developed and validated by a tenfold cross-validation on the training sample. Variables importance analysis suggested that blood urea nitrogen, serum creatinine, albumin, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, calcium, and glucose were the most important seven predictors of SAP. The AUCs of RF model in tenfold cross-validation of the training set and the test set was 0.89 and 0.96, respectively. Both the area under precision recall curve and the diagnostic accuracy of the RF model were higher than that of both the LR model and the BISAP score. LIME plots were used to explain individualized prediction of the RF model.ConclusionsAn interpretable RF model exhibited the highest discriminatory performance in predicting SAP. Interpretation with LIME plots could be useful for individualized prediction in a clinical setting. A nomogram consisting of albumin, serum creatinine, glucose, and pleural effusion was useful for prediction of SAP.

Development of a Novel Nomogram Incorporating Red Blood Cell Distribution Width-Albumin Ratio for the Prediction of 30-day Mortality in Acute Pancreatitis Patients

“…In recent years, a number of diagnostic and prognostic markers, both alone and in combination, have been tested on individuals diagnosed with AP [16][17][18]. After elevated RDW was established as a predictor of poor outcomes in chronic heart failure in 2007, multiple studies have shown that it is signifcantly related to short-term mortality outcomes in a range of infammatory disorders [19,20].…”

Predictive value of machine learning for the severity of acute pancreatitis: A systematic review and meta-analysis

“…These newly identified studies were screened based on the inclusion and exclusion criteria. Finally, a total of 33 original studies [ [14] , [15] , [16] , [17] , [18] , [19] , [20] , [21] , [22] , [23] , [24] , [25] , [26] , [27] , [28] , [29] , [30] , [31] , [32] , [33] , [34] , [35] , [36] , [37] , [38] , [39] , [40] , [41] , [42] , [43] , [44] , [45] , [46] ] were included. The literature screening process is illustrated in Fig.…”

“…Ten [ 23 , 24 , 26 , 27 , 30 , 31 , 34 , 35 , 40 , 41 ] studies were multicenter studies, while two studies [ 18 , 20 ] collected subjects from databases. Eleven studies [ [14] , [15] , [16] , 18 , [20] , [21] , [22] , 25 , 26 , 31 , 35 ] considered overfitting, and k-fold cross-validation was primarily used. The original studies collectively constructed 55 new machine-learning models and evaluated three primary clinical scales: APACHE II, BISAP, and Ranson.…”

“…In terms of statistical methods, ten studies [ 17 , 18 , 25 , 26 , [39] , [40] , [41] , 44 , 45 ] had an EPV <10 in the training set, and nine studies [ [20] , [21] , [22] , 24 , 30 , 33 , 34 , 36 , 46 ] were unable to calculate EPV. In addition, eleven studies [ 14 , 16 , 17 , 21 , 22 , 24 , 29 , 30 , 36 , 37 , 39 ] had no independent validation sets or had independent validation sets with fewer than 100 cases.…”

Application of machine learning in prediction of bone cement leakage during single-level thoracolumbar percutaneous vertebroplasty

“…The relationship between model errors and fitted variables is depicted by cross-validation curves. Multiple groups of different training and verification are performed on the model in this study by using different training sets and verification set division through the implementation of ten-fold cross-validation, to deal with the problems of one-sided test results and insufficient data [20].…”