A Machine Learning-Modified Novel Nomogram to Predict Perioperative Blood Transfusion of Total Gastrectomy for Gastric Cancer

Abstract

BackgroundPerioperative blood transfusion reserves are limited, and the outcome of blood transfusion remains unclear. Therefore, it is important to prepare plans for perioperative blood transfusions. This study aimed to establish a risk assessment model to guide clinical patient management.MethodsThis retrospective comparative study involving 513 patients who had total gastrectomy (TG) between January 2018 and January 2021 was conducted using propensity score matching (PSM). The influencing factors were explored by logistic regression, correlation analysis, and machine learning; then, a nomogram was established.ResultsAfter assessment of the importance of factors through machine learning, blood loss, preoperative controlling nutritional status (CONUT), hemoglobin (Hb), and the triglyceride–glucose (TyG) index were considered as the modified transfusion-related factors. The modified model was not considered to be different from the original model in terms of performance, but is simpler. A nomogram was created, with a C-index of 0.834, and the decision curve analysis (DCA) demonstrated good clinical benefit.ConclusionsA nomogram was established and modified with machine learning, which suggests the importance of the patient’s integral condition. This emphasizes that caution should be exercised regarding transfusions, and, if necessary, preoperative nutritional interventions or delayed surgery should be implemented for safety.

Prediction of intraoperative red blood cell transfusion in valve replacement surgery: machine learning algorithm development based on non-anemic cohort

“…As we investigated, the optimal model, CatBoost algorithm, was superior to LR with an AUC of 0.752 (CatBoost) vs. 0.666 (LR), implying intraoperative RBC transfusion is not simply linearly related to preoperative risk factors based on our dataset. Undoubtedly, with the proficiency of artificial intelligence in handling nonlinear prediction tasks, more and more machine learning models are being used to predict transfusions in a series of different diseases, offering new approaches to improve clinical outcomes ( 22 , 23 ).…”

Naïve Bayes is an interpretable and predictive machine learning algorithm in predicting osteoporotic hip fracture in-hospital mortality compared to other machine learning algorithms

“…A variety of ML algorithms, including regression-based methods, decision-tree based methods (i.e. decision trees [DT], Random Forests [RF], eXtreme Gradient Boosting [XGB] implementation), neural networks (NN), Naïve Bayes and support vector machines (SVM) have been used in the prognostication of patients in the general peri-operative [11][12][13][14][15][16] and peri-HF [17][18][19][20] period with varying degrees of success. However, most of these tools require data that is not readily available on admission (such as intra-operative data and laboratory data), much like the tools developed from traditional statistical methods and most do not predict short-term in-hospital mortality following HF.…”