Risk assessment method for non-dominant hemorrhage during perioperative period for treating intertrochanteric fracture through proximal femur PFNA
By constructing a perioperative implicated hemorrhage prediction model, using Logistic regression and Lasso regression to screen factors, the problem of difficult to identify and prevent perioperative implicated hemorrhage in patients with intertrochanteric fractures is solved, and more accurate risk assessment and management is achieved, and patient prognosis is improved.
Patent Information
- Application Number
- CN202411902358.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-23
- Publication Date
- 2025-05-16
AI Technical Summary
Patients with intertrochanteric fractures find it difficult to accurately identify and prevent implicit blood loss during the perioperative period, resulting in increased anemia and complications, affecting the patient's prognosis.
By retrospectively analyzing independent risk factors that affect implicit hemorrhage, a perioperative implicit hemorrhage prediction model was constructed, and the optimal predictors were screened out using Logistic regression model and Lasso regression, and an evaluation system was established to predict and evaluate the risk of implicit hemorrhage.
More accurately predict and evaluate the implicit blood loss after intertrochanteric fracture of the femoral, help clinicians develop management plans, reduce the risk of implicit blood loss, and improve patient prognosis.
Smart Images
Figure CN120015303A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of medical technology, and specifically to a method for assessing the risk of non-obvious blood loss during the perioperative period of proximal femoral PFNA treatment of intertrochanteric fractures. Background Art
[0002] Hip fractures, especially intertrochanteric femur fractures (ITF), have become the focus of medical attention due to their close relationship with osteoporosis. This type of fracture is not only highly valued due to its high incidence, high mortality and high disability rate, but also has a profound impact on the quality of life of patients and a large demand for medical resources.
[0003] Patients with intertrochanteric fractures not only suffer from severe pain and limited mobility, but also face a high risk of mortality. Statistics show that between 2000 and 2018, the one-year postoperative mortality rates of patients with intertrochanteric fractures and femoral neck fractures were 17.47% and 9.83%, respectively, which are significantly higher than those of the general population. In addition, the amount of blood loss after hip fracture is large, especially hidden blood loss, which far exceeds the visible bleeding during surgery and is often difficult to accurately identify. This hidden blood loss not only causes anemia in patients, but may also cause a series of complications, further increasing the risk of death. Although surgical techniques have been greatly improved in recent years, the proximal femoral nail antirotation (PFNA) for the treatment of intertrochanteric fractures has the advantages of less trauma and less intraoperative bleeding, but its average perioperative blood loss is still as high as 937 ml, of which 81.96% is hidden blood loss. However, perioperative hidden blood loss is the result of the combined action of multiple factors, which is difficult for clinicians to detect early and there is no effective means to prevent it. Summary of the invention
[0004] Purpose of the invention: The purpose of the present invention is to provide a method for assessing the risk of non-obvious blood loss during the perioperative period of PFNA for the treatment of intertrochanteric fractures of the femur. By retrospectively analyzing multiple independent risk factors affecting occult blood loss and constructing a perioperative occult blood loss prediction model based on biological and clinical variables, the problems existing in the background technology are solved.
[0005] Technical solution: The method for assessing the risk of non-obvious blood loss during the perioperative period of treating intertrochanteric fractures with a proximal femoral nail anti-rotation (PFNA) according to the present invention comprises the following steps: (1) Obtain patient data sets, preprocess them, and divide the data sets; (2) Screen the data set to obtain the optimal predictor; (3) Based on the training set, a perioperative hidden blood loss prediction model was constructed using the screened predictive factors; (4) Draw calibration graphs and decision curve analysis (DCA) to evaluate the predictive performance of the constructed model.
[0006] Furthermore, step (1) includes the following steps: (11) Obtain detailed information on patients with intertrochanteric fractures treated with proximal femoral nail anti-rotation (PFNA) from relevant medical institutions; the sample size was determined by G Power statistical analysis while controlling the type I error rate, and patients were screened according to clear inclusion criteria; (12) The collected data were cleaned, organized, and formatted comprehensively, classified according to medical diagnostic standards, and extreme values and outliers were removed. The OSTEO formula was used to accurately calculate the hidden blood loss of each patient, and HBL ³1000 ml was used as the grouping standard for the outcome variable.
[0007] Furthermore, step (2) includes the following steps: (21) Chi-square test or Fisher's exact test was used for comparative analysis of categorical variables in the data set; t-test or Mann-Whitney U test was used for comparative analysis of continuous variables in the data set to preliminarily screen out variables that may be related to occult blood loss; (22) All predictive factors were input into the Lasso regression model and cross-validated. The λ value with the smallest mean square error was selected to screen out the optimal predictive factor; where λ=0.0236638408027355.
[0008] Furthermore, in step (21), the variables initially screened included: demographic data (sex, age, body mass index), laboratory indicators (preoperative and postoperative hemoglobin, red blood cells, platelets, hematocrit, prothrombin time, activated partial thromboplastin time, international normalized ratio, thrombin time, fibrinogen, serum albumin, serum globulin, and blood calcium level at admission), anesthesia method, American Society of Anesthesiologists (ASA) grade, time from injury to surgery, duration of surgical operation, intraoperative bleeding, fracture type (Evans classification), whether blood transfusion was used, hypertension, and diabetes.
[0009] Furthermore, in step (22), the optimal predictive factors screened by Lasso regression included: gender, BMI kg / m², preoperative hemoglobin g / L, preoperative blood transfusion, fracture type Evans classification, hypertension, blood calcium level mmol / L on admission, and ASA grade.
[0010] Furthermore, the modeling formula of step (3) is as follows: .
[0011] Furthermore, step (4) is as follows: the prediction model established based on the training set is evaluated by ROC on the validation set, and a calibration curve and a decision curve analysis DCA diagram are drawn to evaluate the consistency of the prediction model.
[0012] The present invention provides a non-obvious blood loss risk assessment system during the perioperative period of proximal femoral intramedullary PFNA treatment of intertrochanteric fractures, comprising: Acquisition module: used to acquire patient data sets, preprocess them, and divide the data sets; Screening module: used to screen the data set to obtain the optimal prediction factor; Logistic regression module: used to perform multi-factor logistic regression modeling based on the training set using the optimal predictive factors screened out; Evaluation module: used for calibration plot and decision curve analysis DCA to evaluate the prediction performance of the constructed model.
[0013] An electronic device described in the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, it implements any one of the methods described in the method for assessing the risk of non-obvious blood loss during perioperative treatment of intertrochanteric fractures of the proximal femur using PFNA.
[0014] A storage medium described in the present invention stores a computer program, and when the computer program is executed by a processor, it implements any method for assessing the risk of non-obvious blood loss during the perioperative period of proximal femoral PFNA treatment of intertrochanteric fractures.
[0015] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: by retrospectively analyzing multiple independent risk factors affecting hidden blood loss, and constructing a perioperative hidden blood loss prediction model based on biological and clinical variables, the probability of individual clinical events occurring is graphically displayed, and the hidden blood loss after intertrochanteric fracture surgery is more accurately predicted and evaluated, thereby improving the prognosis of patients. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 is a flow chart of the present invention; Figure 2 is the LASSO regression optimization screening variable diagram of the present invention; Figure 3 The present invention is a nomogram for predicting the risk factors of hidden blood loss during the perioperative period of intertrochanteric fracture; Figure 4 is the ROC curve of the present invention; Figure 5 It is the calibration curve and decision curve analysis of the present invention. DETAILED DESCRIPTION
[0017] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.
[0018] like Figure 1 As shown, an embodiment of the present invention provides a method for assessing the risk of non-obvious blood loss during the perioperative period of proximal femoral PFNA treatment of intertrochanteric fractures, comprising the following steps: (1) Obtain patient data sets, preprocess them, and divide the data sets; including the following steps: (11) With the approval of the local ethics committee, detailed data of 231 patients with intertrochanteric fractures treated with proximal femoral nail anti-rotation (PFNA) were obtained from relevant medical institutions. The sample size was determined through rigorous GPower statistical analysis to ensure sufficient statistical power (1-β=0.80) to detect the potential association between independent variables and hidden blood loss (HBL), while strictly controlling the type I error rate (α=0.05). Patients were screened according to clear inclusion criteria (etc.) to maintain the homogeneity of the research subjects and the reliability of the data. Exclusion criteria included potential confounding factors such as perioperative bleeding, infection, and chronic anemia to minimize interference with the research results; (12) The collected data were cleaned, sorted and formatted comprehensively, classified according to relevant medical diagnostic standards, and extreme values and outliers were removed; the OSTEO formula was used to accurately calculate the hidden blood loss of each patient, and HBL³1000ml was used as the grouping standard for the outcome variable. The 95% confidence interval was determined using R software (specifically, the Confint() function was used to calculate the profilelikelihood CI), and the data were then randomly split into a training set and a validation set in a ratio of 8:2 (with a random seed number of 1234) to ensure the independence and validity of model training and validation.
[0019] (2) Screening the data set to obtain multivariate predictors; including the following steps: (21) Chi-square test or Fisher's exact test was used for comparative analysis of categorical variables in the data set; t-test or Mann-Whitney U test was used for comparative analysis of continuous variables in the data set to preliminarily screen out variables that may be related to hidden blood loss; the preliminarily screened variables included: demographic data (gender, age, body mass index), laboratory indicators (hemoglobin, red blood cells, platelets, hematocrit, prothrombin time, activated partial thromboplastin time, international normalized ratio, thrombin time, fibrinogen, serum albumin, serum globulin, blood calcium level at admission before and after surgery), anesthesia method, American Society of Anesthesiologists (ASA) grade, time from injury to surgery, duration of surgical operation, intraoperative bleeding, fracture type (Evans classification), blood transfusion, hypertension, diabetes. Design nomogram model (such as Figure 3 The nomogram model includes multiple variables and their corresponding score ranges, including score scale, patient gender, BMI, serum calcium concentration at admission, preoperative hemoglobin level, ASA grade, Evans classification of intertrochanteric fracture, hypertension, perioperative blood transfusion, etc. In actual clinical work, medical staff substitute the patient's various indicators into the model for assignment, add up the total score, and convert the total score to the reading on the risk axis to obtain the risk coefficient of hidden blood loss during the perioperative period of the patient receiving PFNA treatment. For example, taking a specific patient with intertrochanteric fracture as an example, the clinical data are as follows: male gender (assigned value 7.5), BMI is 24 (assigned value 21.25), blood calcium concentration at admission is 2.25mmol / L (assigned value 15), preoperative hemoglobin level is 135 g / L (assigned value 60), ASA grade is grade III (assigned value 0), intertrochanteric fracture Evans classification is type II (assigned value 20), no hypertension (assigned value 7.5), perioperative blood transfusion (assigned value 100). Substituting these data into the nomogram model, the total score is calculated to be 231.25. The total score is converted to the reading on the risk axis, and the risk coefficient of hidden blood loss during the perioperative period of PFNA treatment is about 0.85. Based on this risk coefficient, clinicians can formulate corresponding perioperative management plans to reduce the risk of hidden blood loss in patients.
[0020] (22) Figure 2 As shown in the figure, all predictive factors were input into the Lasso regression model to establish a cross-validation model, and the λ value with the smallest mean square error was selected to screen out the optimal predictive factor; where λ=0.0236638408027355; the optimal predictive factors included: gender, BMI kg / m², preoperative hemoglobin g / L, preoperative blood transfusion, fracture type Evans classification, hypertension, blood calcium level mmol / L on admission, and ASA grade.
[0021] (3) Based on the training set, the optimal predictive factors screened out were used to construct a perioperative hidden blood loss prediction model; the modeling formula is as follows: ; Among them, log[p ̂ / (1-p ̂)] represents the logarithmic ratio of the probability of an event occurring (p ̂) to the probability of not occurring (1-p ̂) (i.e., log odds), also known as logit. The right side of the formula is the various independent variables (predictors) and their corresponding coefficients, which together determine the probability of an event occurring.
[0022] The following is an explanation of each parameter: Constant term -10.439: This is the baseline value of logit when all independent variables are 0 (or reference category). In practical applications, since it is difficult for all independent variables to be 0, this value is more used as a calculation benchmark.
[0023] 0.523 × (Male): If the individual is male, add 0.523 to the logit. A positive coefficient indicates that males are more likely to have the event occur than the reference group (females).
[0024] 1.594 × (BMI ≥ 24 kg / m²): Individuals with a BMI greater than or equal to 24 kg / m² have an increase of 1.594 on the logit. This indicates that a higher BMI is associated with an increased probability of the event occurring.
[0025] 0.061×(preoperative hemoglobin level): For every unit increase in preoperative hemoglobin level, the logit increases by 0.061. The coefficient is small, indicating that its direct impact on the probability of the event may be relatively small, but it is still a positive impact.
[0026] 7.481×(preoperative transfusion): If the individual received a transfusion before surgery, 7.481 is added to the logit. This is a very large coefficient, indicating that preoperative transfusion greatly increases the probability of the event.
[0027] 1.52× (fracture type Evans type II), 0.452× (fracture type Evans type III), 1.316× (fracture type Evans type IV): These coefficients represent the contribution of different Evans types of fractures to the logit relative to an unspecified reference fracture type (Evans type I). Positive values indicate that these types of fractures increase the probability of the event.
[0028] -0.594 (hypertension): If the individual has hypertension, then 0.594 is reduced on the logit. A negative coefficient indicates that hypertension reduces the probability of an event.
[0029] -1.099×(admission blood calcium >2.25mmol / L): Individuals with blood calcium levels above 2.25 mmol / L on admission have a logit reduction of 1.099. This means that higher blood calcium levels are associated with a lower probability of the event occurring.
[0030] -1.401×(anesthesia grade ≥ 3): If the anesthesia grade was 3 or above, it was reduced by 1.401 on the logit. This suggests that higher anesthesia grade (which may mean a riskier surgery) is associated with a reduced probability of an event, but this may also reflect other aspects of surgical complexity and patient health.
[0031] It should be noted that since the logistic regression model outputs logit values, to obtain the actual probability value p̂, it is necessary to use the formula p ̂ = 1 / (1 + e^(-logit)) for conversion. In addition, these explanations are based on the direct interpretation of the model coefficients. In practical applications, factors such as the model fit and multicollinearity of variables must also be considered.
[0032] (4) Draw a calibration graph and decision curve analysis DCA to evaluate the prediction performance of the constructed model. Specifically: Perform ROC evaluation on the validation set based on the prediction model established on the training set, draw a calibration curve and decision curve analysis DCA graph to evaluate the consistency of the prediction model. Figure 4 As shown in the figure, the ROC curve of the nomogram model is shown, and the discriminative ability of the model is quantified by calculating the area under the ROC curve (AUC). Figure 5 As shown, calibration curve and decision curve analysis of the training cohort and internal test cohort are presented to further validate the accuracy and reliability of the model.
[0033] The embodiment of the present invention further provides a risk assessment system for non-obvious blood loss during perioperative treatment of intertrochanteric fractures by intramedullary PFNA of the proximal femur, comprising: Acquisition module: used to acquire patient data sets, preprocess them, and divide the data sets; Screening module: used to screen the data set to obtain multi-factor predictors; Logistic regression module: used to perform multi-factor Logistic regression modeling based on the training set using the screened multi-factor predictors; Evaluation module: used for calibration plot and decision curve analysis DCA to evaluate the prediction performance of the constructed model.
[0034] An embodiment of the present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is loaded into the processor, it implements any one of the methods for assessing the risk of non-obvious blood loss during perioperative treatment of intertrochanteric fractures of the proximal femur using PFNA.
[0035] An embodiment of the present invention also provides a storage medium storing a computer program, which, when executed by a processor, implements any one of the methods for assessing the risk of non-obvious blood loss during perioperative treatment of intertrochanteric fractures by proximal femoral PFNA.
Claims
1. A method for assessing the risk of non-obvious blood loss during the perioperative period of proximal femoral PFNA treatment of intertrochanteric fractures, characterized in that: The following steps are involved: (1) Obtain patient data sets, preprocess them, and divide the data sets; (2) Screening the data set to obtain multivariate predictors; (3) Based on the training set, a perioperative hidden blood loss prediction model was constructed using the screened multivariate predictors; (4) Draw calibration graphs and decision curve analysis (DCA) to evaluate the predictive performance of the constructed model.
2. A method for assessing the risk of non-obvious blood loss during perioperative treatment of intertrochanteric fractures by proximal femoral PFNA according to claim 1, characterized in that: Step (1) includes the following steps: (11) Obtain detailed information on patients with intertrochanteric fractures treated with proximal femoral nail anti-rotation (PFNA) from relevant medical institutions; the sample size was determined by G Power statistical analysis while controlling the type I error rate, and patients were screened according to clear inclusion criteria; (12) The collected data were cleaned, organized, and formatted comprehensively, classified according to medical diagnostic standards, and extreme values and outliers were removed. The OSTEO formula was used to accurately calculate the hidden blood loss of each patient, and HBL ³1000 ml was used as the grouping standard for the outcome variable.
3. A method for assessing the risk of non-obvious blood loss during perioperative treatment of intertrochanteric fractures by proximal femoral PFNA according to claim 1, characterized in that: Step (2) includes the following steps: (21) Chi-square test or Fisher's exact test was used for comparative analysis of categorical variables in the data set; t-test or Mann-Whitney U test was used for comparative analysis of continuous variables in the data set to preliminarily screen out variables that may be related to occult blood loss; (22) All predictive factors were input into the Logistic regression model and a 10-fold cross-validation was performed. The λ value with the smallest mean square error was selected to screen out the multi-factor predictive factors; where λ=0.0236638408027355.
4. A method for assessing the risk of non-obvious blood loss during perioperative treatment of intertrochanteric fractures by proximal femoral PFNA according to claim 3, characterized in that: In step (21), the variables initially screened included demographic data, laboratory parameters, anesthesia method, American Society of Anesthesiologists (ASA) grade, time from injury to surgery, duration of surgical operation, intraoperative bleeding, fracture type, blood transfusion, hypertension, and diabetes.
5. A method for assessing the risk of non-obvious blood loss during perioperative treatment of intertrochanteric fractures by proximal femoral PFNA according to claim 3, characterized in that: In step (22), the multivariate predictors included: sex, BMI kg / m², preoperative hemoglobin g / L, preoperative blood transfusion, fracture type Evans classification, hypertension, blood calcium level mmol / L on admission, and ASA classification.
6. A method for assessing the risk of non-obvious blood loss during perioperative treatment of intertrochanteric fractures by proximal femoral PFNA according to claim 1, characterized in that: The modeling formula of step (3) is as follows: 。 7. A method for assessing the risk of non-obvious blood loss during perioperative treatment of intertrochanteric fractures by proximal femoral PFNA according to claim 1, characterized in that: Step (4) is as follows: The prediction model established based on the training set is evaluated by ROC on the validation set, and the calibration curve and decision curve analysis DCA diagram are drawn to evaluate the consistency of the prediction model.
8. A risk assessment system for non-obvious blood loss during perioperative treatment of intertrochanteric fractures by intramedullary PFNA of the proximal femur, characterized in that: include: Acquisition module: used to acquire patient data sets, preprocess them, and divide the data sets; Screening module: used to screen the data set to obtain multi-factor predictors; Logistic regression module: used to perform multi-factor Logistic regression modeling based on the training set using the selected multi-factor predictors; Evaluation module: used for calibration plot and decision curve analysis DCA to evaluate the prediction performance of the constructed model.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is loaded into the processor, it implements the method for assessing the risk of non-obvious blood loss during perioperative treatment of intertrochanteric fractures by proximal femoral PFNA according to any one of claims 1 to 7.
10. A storage medium storing a computer program, characterized in that: When the computer program is executed by the processor, it implements the method for assessing the risk of non-obvious blood loss during perioperative treatment of intertrochanteric fractures by proximal femoral PFNA according to any one of claims 1 to 7.