RESUMO
【Objective】 To predict the risk factors of intraoperative blood transfusion by establishing a random forest algorithm prediction model, and to evaluate its prediction performance in clinical. 【Methods】 A total of 48 176 patients who underwent surgery from January 2014 to December 2017 in the First Medical Center of the Chinese People′s Liberation Army General Hospital were collected and divided into a blood transfusion group(n=5 035) and a non-transfusion group(n=43 141) according to whether blood was transfused or not during the operation, and the age, gender, weight, blood routine, coagulation test indicators, surgical grade, number of operations and anesthesia methods, and preoperative blood transfusion history between the two groups were compared and analyzed. All cases were randomly divided into training set(n=33 723) and the test set(n=14 453), using the sklearn function package in the computer programming language(Python V 3.9.0) to introduce the random forest algorithm, with 2 groups of different factors incorporated into the random forest algorithm to build the model, and the model was evaluated using the operating curve(ROC). 【Results】 1) There were statistically significant differences between the blood transfusion group and the non-transfusion group in terms of gender, age, blood routine, coagulation function, surgical grade, and preoperative blood transfusion history(P0.05); 3) In the established intraoperative blood model, the blood routine, coagulation function and general anesthesia had a great influence, with the cumulative importance > " 0.90" ; 4) The ROC analysis showed that the area under the ROC curve of the random forest model was 0.91 and 0.82 in the training set and the test set, which demonstrated a good predictive ability. 【Conclusion】 The intraoperative blood, using prediction model based on random forest method, can predict intraoperative blood use and blood transfusion risk factors.
RESUMO
【Objective】 To develop a prediction model of allogenic blood transfusion in elective patients based on machine learning, so as to guide clinicians to prepare blood for perioperative patients more reasonably. 【Methods】 Relevant data of all surgical patients from 2012 to 2018 were extracted from the big data integration platform of our hospital, to construct the surgical blood database based on Python V3.8.0. All data were analyzed using Excel and SAS, and the prediction model was developed based on SPSS Modeler 18.0. 【Results】 1) There was a negative correlation between preoperative Hb and BMI and intraoperative blood transfusion rate, with Pearson correlation coefficient (R) as -0.168 and -0.046, respectively. The transfusion rate of patients under 1 year old was the highest, up to 15.63%. The transfusion rate of female patients was higher than that of male patients (P>0.05), as cardiac surgery rated at the highest 11.38%, but their per capita blood transfusion was lower than that of males (P<0.01). 2) The AUC range corresponding to the prediction model for transfusion probability was 0.67~0.88, and when the AUC reached the highest, the hit ratio, coverage rate and specificity of Model 9 was 10.7%, 85.76% and 75.4%, respectively. 3) The main factors contributing to the prediction model for transfusion volume in surgery were weight, Hb, total protein(TP), etc. 【Conclusion】 The prediction efficiency of the successfully constructed prediction model for perioperative blood use was better than that of MSBOS.
RESUMO
【Objective】 To establish a blood transfusion outcome prediction model for comprehensivel evaluation of coagulation function of patients with upper gastrointestinal bleeding by thrombelastogram (TEG) and blood coagulation indicators. 【Methods】 The data of 101 patients with upper gastrointestinal hemorrhage, admitted to the Department of Gastroenterology of Zhejiang Provincial People′s Hospital and its Chun′an Branch from June 2018 to June 2021, were collected through Tongshuo blood transfusion management system and His system. Those patients were divided into blood transfusion group (n=56) and non-transfusion group (n=45), and into cirrhosis group (n=74) and non-cirrhosis group (n=27), and 40 patients, with non-upper gastrointestinal bleeding, were enrolled as the control. The results of TEG indicators (R, K, α, MA), coagulation function (PT, INR, APTT, TT, Fib), blood routine (Hb, Plt, WBC, NEUT%) and biochemical detection(Alb, SCr, ALT, AST, GGT) before transfusion were compared between groups and the correlation between TEG indicators and traditional coagulation parameters was analyzed. Single-factor and multi-factor analysis were used to screen blood transfusion-related factors to establish a predictive model. 【Results】 The comparisons of paremeters between transfusion and non-transfusion group were as follows, K (min), α (°), and MA (mm) was 3.86±3.12 vs 2.50±1.47, 54.00±14.08 vs 61.05±10.88, and 51.12±13.37 vs 58.26±11.08, respectively (P<0.01); PT (s) and Fib (g) was 16.36±7.45 vs 13.44±1.50 and 1.59±0.87 vs 2.35±1.09 (P<0.01); NEUT% and Hb (g/L) was 0.75 ±0.13 vs 0.66±0.15 and 68.04±14.49 vs 100.73±22.92 (P<0.01); Alb (g/L) and SCr (nmol/L) was 29.73±6.08 vs 33.73±7.19 and 99.50±53.55 vs 76.25±19.28 (P<0.01). Correlation analysis showed that APTT was positively correlated with R and K values, and negatively correlated with α and MA. Fib was negatively correlated with K values, and positively correlated with α and MA. Plt was negatively correlated with K values, and positively correlated with α and MA (P<0.01). Eight pre-transfusion indicators as K, MA, PT, Fib, NEUT%, Hb, Alb, and SCr were subjected to Logistic regression to establish a blood transfusion prediction model. The optimal ROC curve of blood transfusion threshold (blood transfusion predictive value of patients), sensitivity, specificity and AUC were 0.448, 92.9%, 88.9%, and 0.969, respectively. 【Conclusion】 The establishment of Logistic regression model by integrating detection indicators of TEG, coagulation function, blood routine and biochemistry in patients with upper gastrointestinal bleeding have showed significant correlation with blood transfusion prediction, and good clinical practicability.