Liver transplantation recipient sarcopenia classification method and device based on peripheral blood and CT

Abstract

The invention discloses a peripheral blood and CT-based liver transplantation recipient sarcopenia classification method and device, and the method comprises the steps: carrying out the single-factor COX regression preliminary screening of collected peripheral blood biochemical indexes of liver transplantation recipients, and obtaining potential prediction factors related to sarcopenia; performing LASSO regression screening on the potential predictive factors to obtain core predictive factors related to sarcopenia; multiplying each core predictive factor by a regression coefficient calculated in LASSO regression, performing accumulation calculation to obtain an LASSO + COX model score, and dividing liver transplantation recipients into different risk groups according to the size of the LASSO + COX model score; and carrying out definite diagnosis on sarcopenia according to a risk group division result in combination with a CT image examination result. According to the method, a new strategy of peripheral blood prediction preliminary screening and high-risk CT confirmation is realized, the radiation risk, cost and detection threshold are reduced, and the sarcopenia management efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the field of medical artificial intelligence technology, and specifically relates to a method and device for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT. Background Art

[0002] Sarcopenia refers to a syndrome characterized by decreased muscle mass, decreased muscle strength, and reduced physical function caused by factors such as age, malnutrition, and chronic diseases. Recent studies have shown that sarcopenia is closely associated with many adverse clinical outcomes, including poor postoperative recovery, increased risk of infection, decreased quality of life, and increased mortality, which seriously affect patients' clinical prognosis and quality of life. Especially in the liver transplant patient population, sarcopenia has been proven to be an important risk factor affecting short-term and long-term prognosis after transplantation. Liver transplant patients with sarcopenia have significantly increased postoperative complications and significantly shortened survival, and even seriously affect the success rate of transplantation surgery. Therefore, early and accurate identification and monitoring of the sarcopenia status of liver transplant patients is of great clinical significance for optimizing surgical plans, predicting postoperative risks, formulating individualized rehabilitation plans, and improving patients' clinical prognosis and quality of life.

[0003] Currently, the assessment and monitoring of sarcopenia in clinical practice primarily relies on imaging studies, with computed tomography (CT) as the current gold standard. CT measures the cross-sectional area of ​​skeletal muscle at the level of the third lumbar vertebra (L3) and calculates the skeletal muscle index (SMI), providing an objective and quantitative basis for assessing muscle mass. This method, with its high accuracy and good reproducibility, has been widely used in both clinical and scientific research. However, CT examinations also have significant drawbacks. First, its ionizing radiation exposure poses a significant risk. Although the radiation dose received by patients with each examination is within a safe range, frequent use can significantly increase cumulative radiation exposure, increasing the risk of future cancer, especially for liver transplant patients requiring long-term follow-up. Second, CT examinations are expensive, placing a significant financial burden on patients, and require specialized personnel for operation, limiting their widespread use in primary care settings and for frequent follow-up. Furthermore, as patients' awareness of radiation safety increases, their compliance with high-frequency CT follow-up has decreased. Therefore, exploring an alternative or auxiliary monitoring method that can maintain the accuracy of sarcopenia assessment while reducing radiation exposure and lowering the medical economic burden has become an urgent clinical need.

[0004] To address the shortcomings of CT examinations, in recent years, some studies have attempted to assess muscle status through peripheral blood biochemical indices, blood biomarkers (such as creatinine, albumin, prealbumin, etc.), or physical function tests (such as grip strength, walking speed), hoping to replace or assist CT examinations. These methods have significant advantages such as non-invasiveness, low cost, and ease of frequent repeated testing, providing new ideas for monitoring sarcopenia. These methods have the advantages of being non-invasive, low cost, and ease of frequent repeated testing. However, single indicators have poor sensitivity and specificity, and are easily interfered with by factors such as inflammation, abnormal liver and kidney function, and nutritional status, making it difficult to independently and reliably assess muscle mass and functional status. Therefore, an effective clinical application system has not yet been established, and it cannot play a leading role in the assessment and monitoring of sarcopenia. Summary of the Invention

[0005] In view of the above, the purpose of the present invention is to provide a method and device for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT. By constructing the LASSO+COX model score and combining multiple peripheral blood indicators, high-risk sarcopenia patients can be screened without the need for frequent CT examinations, realizing a new strategy of peripheral blood prediction initial screening + high-risk CT confirmation, thereby reducing radiation risks, costs and detection thresholds, and improving the efficiency of sarcopenia management.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT, comprising the following steps:

[0008] The peripheral blood biochemical indicators collected from liver transplant recipients were screened by univariate COX regression to obtain potential predictors related to sarcopenia. LASSO regression was then performed on the potential predictors to obtain core predictors related to sarcopenia.

[0009] Each core predictor was multiplied by its regression coefficient calculated in LASSO regression and the results were added together to obtain the LASSO+COX model score. Liver transplant recipients were divided into different risk groups according to the LASSO+COX model score.

[0010] Sarcopenia was diagnosed based on the risk group division results combined with CT imaging results.

[0011] Preferably, the significance level P<0.1 is set as the initial screening criterion in the univariate COX regression.

[0012] Preferably, a total of 14 potential predictive factors were screened, including recipient age, alpha-fetoprotein level, tumor number, anti-HBc antibody level, microvascular invasion, albumin / globulin ratio, international normalized ratio, intraoperative blood loss, platelet count, mean corpuscular hemoglobin, lymphocyte percentage, triglycerides, total cholesterol, and diagnosis of non-alcoholic steatohepatitis.

[0013] Preferably, the optimal performance is achieved when log(λ)=-2.57 is set in the LASSO regression screening, where λ represents a penalty parameter.

[0014] Preferably, a total of 7 core predictive factors are screened, including: diagnosis of non-alcoholic steatohepatitis, mean corpuscular hemoglobin, lymphocyte percentage, intraoperative blood loss, anti-HBc antibody level, albumin / globulin ratio, and total cholesterol.

[0015] Preferably, each core predictor is multiplied by its regression coefficient calculated in LASSO regression and the results are added together to obtain the LASSO+COX model score, which is calculated as follows:

[0016] LASSO+COX model score = (α1×diagnosis of nonalcoholic steatohepatitis) + (α2×mean corpuscular hemoglobin) + (α3×lymphocyte percentage) + (α4×intraoperative blood loss) + (α5×anti-HBc antibody level) + (α6×albumin / globulin ratio) + (α7×total cholesterol), where α1, α2, α3, α4, α5, α6, and α7 represent the regression coefficients corresponding to each core predictor, respectively.

[0017] Preferably, the liver transplant recipients are divided into different risk groups according to the LASSO+COX model score, including:

[0018] Liver transplant recipients whose LASSO+COX model scores were less than the risk threshold were divided into the low-risk group for sarcopenia;

[0019] Liver transplant recipients with LASSO+COX model scores greater than or equal to the risk threshold were divided into a high-risk group for sarcopenia.

[0020] Preferably, the risk threshold is set to 3.5.

[0021] Preferably, the diagnosis of sarcopenia based on the risk group division results combined with CT imaging results includes:

[0022] Liver transplant recipients classified as low-risk for sarcopenia will undergo continuous follow-up evaluation and monitoring, while liver transplant recipients classified as high-risk for sarcopenia will undergo further CT examinations and sarcopenia will be confirmed based on the CT imaging results.

[0023] In a second aspect, to achieve the above-mentioned purpose of the invention, an embodiment of the present invention further provides a sarcopenia classification device for liver transplant recipients based on peripheral blood and CT, which is implemented using the above-mentioned sarcopenia classification method for liver transplant recipients based on peripheral blood and CT, and includes: a predictor screening module, a risk classification module, and a sarcopenia diagnosis module;

[0024] The predictor screening module is used to perform a single factor COX regression primary screening on the peripheral blood biochemical indicators collected from liver transplant recipients to obtain potential predictors related to sarcopenia, and perform LASSO regression screening on the potential predictors to obtain core predictors related to sarcopenia;

[0025] The risk classification module is used to multiply each core prediction factor by its regression coefficient calculated in LASSO regression and cumulatively calculate the LASSO+COX model score, and divide the liver transplant recipients into different risk groups according to the size of the LASSO+COX model score;

[0026] The sarcopenia diagnosis module is used to diagnose sarcopenia based on the risk group division results combined with CT imaging examination results.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] The present invention constructs a LASSO+COX model score through peripheral blood biochemical indicators to achieve an initial screening of the patient's muscle status. After abnormalities or high-risk groups are found, a precise CT examination is performed to confirm the diagnosis. This solution can not only avoid the abuse of CT and effectively reduce radiation exposure, but also save medical costs, improve patient compliance and screening frequency, reduce blind examinations and clinical burdens, and thus meet the needs of long-term follow-up and chronic disease management. This comprehensive new diagnosis and treatment model can be adapted to multiple scenarios such as postoperative follow-up, telemedicine, and geriatric disease management. It can dynamically monitor the risk of sarcopenia on a non-invasive basis and can also be promoted and applied to multiple clinical scenarios such as liver disease management, chronic disease management, and health monitoring of the elderly population. It provides a new paradigm for the future assessment and management of sarcopenia, and has broad clinical application potential and socioeconomic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0030] Figure 11 is a flow chart of a method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT provided by an embodiment of the present invention;

[0031] Figure 2 1 is a schematic diagram of a calibration curve evaluation based on LASSO+COX model scoring provided in an embodiment of the present invention;

[0032] Figure 3 is a scatter plot of the correlation between the LASSO+COX model score and the CT image body composition index provided by an embodiment of the present invention;

[0033] Figure 4 This is an ROC curve analysis diagram of the LASSO+COX model score in predicting CT indicators provided by an embodiment of the present invention;

[0034] Figure 5 This is a schematic diagram of a Kaplan-Meier survival analysis of patient prognosis based on refined stratification based on LASSO+COX model scores combined with CT imaging indicators, provided in an embodiment of the present invention;

[0035] Figure 6 3 is a schematic structural diagram of a sarcopenia classification device for liver transplant recipients based on peripheral blood and CT provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and do not limit the scope of protection of the present invention.

[0037] The inventive concept of the present invention is: to address the problems of insufficient accuracy of sarcopenia assessment and monitoring methods in the existing technology, as well as the high cost, low compliance and unsuitability for frequent long-term follow-up due to reliance on precise CT examinations, the embodiments of the present invention provide a method and device for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT. The LASSO+COX model score is constructed by using peripheral blood biochemical indicators to achieve initial screening of the patient's muscle status. After abnormal or high-risk groups are found, a precise CT examination is used to confirm the diagnosis. This can avoid the abuse of CT and effectively reduce radiation exposure, while saving medical costs, improving patient compliance and screening frequency, reducing blind examinations and clinical burdens, and thus meeting the needs of long-term follow-up and chronic disease management.

[0038] Figure 1 FIG. 1 is a flow chart of a method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT provided by an embodiment of the present invention. Figure 1 As shown, the embodiment provides a method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT, comprising the following steps:

[0039] S1. The peripheral blood biochemical indicators collected from liver transplant recipients were initially screened by univariate COX regression to obtain potential predictors related to sarcopenia. The potential predictors were then screened by LASSO regression to obtain core predictors related to sarcopenia.

[0040] S1.1, Patient inclusion criteria.

[0041] In the embodiment, patients who were pathologically diagnosed with hepatocellular carcinoma (HCC) after liver transplantation were selected, and whose preoperative peripheral blood biochemical indicators were complete, CT imaging data were complete and could be used for muscle mass assessment, and who completed at least 90 days of follow-up.

[0042] S1.2, collection of peripheral blood indicators.

[0043] In the examples, peripheral blood samples were collected from patients 1-2 days before liver transplantation for testing of biochemical data such as blood routine, liver function, coagulation function, tumor markers, and metabolic indices. In addition, some perioperative parameters (such as intraoperative blood loss and operation duration) were obtained from surgical and anesthesia records after surgery as supplementary variables and incorporated into the analysis during model development. The acquisition of these intraoperative variables does not affect the pre-feature acquisition process of the LASSO+COX model, but is used to enhance the interpretation of clinical outcomes and supplement modeling during subsequent model training.

[0044] (1) Routine blood test parameters: hemoglobin (Hb), mean corpuscular hemoglobin (MCH), red blood cell distribution width (RDW), platelet count (PLT), lymphocyte percentage, etc.;

[0045] (2) Liver function indicators: albumin (ALB), total protein, globulin, albumin / globulin ratio (AG), alanine aminotransferase (ALT), aspartate aminotransferase (AST), AST / ALT ratio, etc.;

[0046] (3) Coagulation function: international normalized ratio (INR);

[0047] (4) Tumor markers: alpha-fetoprotein (AFP);

[0048] (5) Metabolic indicators: total cholesterol (TC), triglycerides (TG), high-density lipoprotein (HDL), low-density lipoprotein (LDL);

[0049] (6) Others: perioperative parameters such as intraoperative blood loss and operation duration.

[0050] S1.3, Univariate Cox regression analysis.

[0051] In the embodiment, all collected peripheral blood indicator variables were subjected to univariate COX regression analysis to screen potential predictive factors associated with sarcopenia, wherein the significance level P<0.1 was set as the initial screening criterion, and a total of 14 potential predictive factors were screened, including: recipient age, alpha-fetoprotein (AFP) level, number of tumors (>3), anti-HBc antibody level, microvascular invasion (MVI), albumin / globulin ratio (AG), international normalized ratio (INR), intraoperative blood loss, platelet count, mean corpuscular hemoglobin (MCH), lymphocyte percentage, triglycerides, total cholesterol, and non-alcoholic steatohepatitis (NASH) diagnosis.

[0052] S1.4, LASSO regression analysis.

[0053] In the embodiment, LASSO regression was used to further optimize the variables. The optimal λ value was determined by 10-fold cross validation. The optimal performance was achieved when log(λ) = -2.57. Under this parameter, the LASSO regression model achieved the best prediction accuracy and the least number of variables. After LASSO regression screening, 7 core predictive factors were finally determined, including:

[0054] (1) Diagnosis of nonalcoholic steatohepatitis (binary variable)

[0055] Regression coefficient α1: 1.4975;

[0056] HR (95% CI): 4.471 (2.423-8.25);

[0057] (2) Mean corpuscular hemoglobin

[0058] Regression coefficient α2: -0.0840;

[0059] HR (95% CI): 0.9195 (0.856-0.9876);

[0060] (3) Lymphocyte percentage

[0061] Regression coefficient α3: 0.0016;

[0062] HR (95% CI): 1.002 (1-1.003);

[0063] (4) Intraoperative bleeding

[0064] Regression coefficient α4: 0.000465;

[0065] HR (95% CI): 1 (1-1.001);

[0066] (5) Anti-HBc antibody level

[0067] Regression coefficient α5: 0.0334;

[0068] HR (95% CI): 1.034 (1.009-1.06);

[0069] (6) Albumin / globulin ratio

[0070] Regression coefficient α6: 0.2032;

[0071] HR (95% CI): 1.225 (1.031-1.457);

[0072] (7) Total cholesterol

[0073] Regression coefficient α7: 0.2373;

[0074] HR (95% CI): 1.268 (1.01-1.591).

[0075] S2: Multiply each core predictor by its regression coefficient calculated in LASSO regression and add them together to obtain the LASSO+COX model score. Liver transplant recipients are divided into different risk groups according to the size of the LASSO+COX model score.

[0076] S2.1, construct LASSO+COX model scoring.

[0077] In the embodiment, each core prediction factor is multiplied by its regression coefficient calculated in the LASSO regression and the results are added together to obtain the LASSO+COX model score. The calculation formula is as follows:

[0078] LASSO+COX model score = (1.4975 × diagnosis of nonalcoholic steatohepatitis) + (-0.0840 × mean corpuscular hemoglobin) + (0.0016 × lymphocyte percentage) + (0.000465 × intraoperative blood loss) + (0.0334 × anti-HBc antibody level) + (0.2032 × albumin / globulin ratio) + (0.2373 × total cholesterol).

[0079] S2.2, Risk stratification.

[0080] In the embodiment, according to the LASSO+COX model score, the patients were divided into two groups:

[0081] (1) Low-risk group: LASSO+COX model score <3.5;

[0082] (2) High-risk group: LASSO+COX model score ≥3.5.

[0083] S2.3, Model performance verification.

[0084] In order to verify the clinical predictive ability of the constructed LASSO+COX model score based on peripheral blood indicators, this example carried out a multi-dimensional model performance evaluation, including accuracy, calibration, clinical interpretability and risk stratification effect.

[0085] S2.3.1, Prediction accuracy evaluation: The performance of LASSO+COX was evaluated using time-dependent receiver operating characteristic (ROC) curves:

[0086] 500-day forecast: AUROC = 0.759;

[0087] 1000-day forecast: AUROC = 0.809;

[0088] 1500-day forecast: AUROC = 0.814.

[0089] This model aims to predict the risk of sarcopenia or related adverse outcomes (such as relapse or death) in liver transplant recipients after surgery. Predicted events were defined as a definitive diagnosis of sarcopenia by CT imaging within 500, 1000, or 1500 days, or relapse or death as documented in the medical history. These labeled events were confirmed by independent imaging and follow-up data. Results demonstrated that the model demonstrated good discriminative power in short-, medium-, and long-term predictions.

[0090] S2.3.2, Calibration analysis To verify whether the predicted probability of the model is consistent with the actual postoperative results, the calibration curve of the model at the 500-day, 1000-day, and 1500-day follow-up time points is drawn. Figure 2 As shown in the figure, the predicted probability of the LASSO+COX model at each time point was highly consistent with the actual probability (including CT-confirmed sarcopenia, recurrence or death), indicating that the model has good calibration performance.

[0091] S2.3.3, Correlation analysis between model scores and risk stratification.

[0092] (1) The LASSO+COX model score was linearly correlated with multiple CT image body composition parameters (such as skeletal muscle index (SMI), visceral adipose tissue (VAT), skeletal muscle attenuation rate (SMRA), etc.). The results are as follows: Figure 3As shown in the figure, BSMI, BVAT, and BSMRA in the vertical axis represent the SMI, VAT, and SMRA indicators before surgery (Before), respectively, and ASMI, AVAT, and ASMRA in the vertical axis represent the SMI, VAT, and SMRA indicators after surgery (After), respectively. It shows that multiple parameters are significantly positively / negatively correlated with the LASSO+COX model score, indicating that the LASSO+COX model score is interpretable in predicting muscle and fat tissue composition.

[0093] (2) This example further explores whether the constructed LASSO+COX model score can replace or indirectly reflect the body composition index measured by CT images. To this end, ROC analysis was used to predict the following CT-derived variables. The results are as follows: Figure 4 As shown:

[0094] SMI: AUC = 0.64;

[0095] VAT: AUC = 0.61;

[0096] SMRA: AUC = 0.63.

[0097] These results indicate that the LASSO+COX model score has a moderate level of predictive ability and can be used as a tool to indirectly reflect changes in body composition when imaging examinations are not possible. It is particularly suitable for primary care or remote monitoring scenarios.

[0098] (3) Risk stratification and prognostic analysis based on LASSO+COX model scores:

[0099] like Figure 5 As shown in Figure 2, the LASSO+COX model score has a good prognostic stratification ability, among which, Figure 5 The left figure is a binary analysis based on the LASSO+COX model score, which divides patients into a high-score group (LASSO+COX model score ≥ 3.5) and a low-score group (LASSO+COX model score < 3.5). The difference in in-hospital recurrence-free survival (RFS) was significant (p = 0.0025), indicating that this score can independently predict postoperative prognostic risk.

[0100] In order to further enhance the ability to identify postoperative prognostic risks, the present invention jointly analyzes the LASSO+COX model score with body composition indicators derived from CT images (such as skeletal muscle index SMI, skeletal muscle radiation attenuation SMR, etc.). This joint strategy is not used to verify whether the model score is valid, but to further enhance the ability to fine-tune the stratification of high-risk patients, especially between high risk and very high risk, low risk and potential risk, to provide a more detailed basis for discrimination. Therefore, the introduction of CT parameters on the basis of LASSO+COX model score does not "negate the independent value of the model", but constructs a hierarchical diagnosis and treatment path based on primary screening + CT enhanced stratification of peripheral blood indicators, achieving a balance between efficiency and accuracy. Specifically, patients are divided into four joint risk level subgroups, namely:

[0101] Low LASSO+COX model score+high CT muscle mass group (double low-risk group), low LASSO+COX model score+low CT muscle mass group, high LASSO+COX model score+high CT muscle mass group, high LASSO+COX model score+low CT muscle mass group (double high-risk group).

[0102] Through Figure 5 The Kaplan-Meier curves shown here compare the postoperative in-hospital recurrence-free survival rates of patients in these four risk subgroups, revealing the following more detailed trends and differences:

[0103] (a) Low LASSO + COX model score + high CT muscle mass group (double low risk group)

[0104] Patients had strong postoperative recovery ability, low risk of complications, the best survival curve performance, the best long-term prognosis, and the recurrence-free survival rate at 6 months and 1 year after surgery was significantly higher than that of the other three groups.

[0105] (b) Low LASSO + COX model score + low CT muscle mass group

[0106] Peripheral biochemical markers and CT imaging results in this group of patients indicated a low risk, and the model scores were consistent with the imaging findings. Clinical prognosis was excellent, second only to that of the double-low-risk group. The Kaplan-Meier curve was slightly lower than that of the double-low-risk group, but still high, indicating that this group of patients had good muscle reserve and systemic compensatory capacity. This suggests that the LASSO+COX model score can provide reliable early risk prediction before surgery and is particularly suitable for preoperative screening in the absence of imaging.

[0107] (c) High LASSO+COX model score+high CT muscle mass group

[0108] This group represents a high-risk group of sarcopenia patients. The in-hospital recurrence-free survival rate was significantly reduced, and a more obvious downward trend appeared in the early postoperative period, suggesting that the reliability of risk assessment is greatest when the LASSO+COX model score is consistent with the CT index. This group requires active intervention before and after surgery.

[0109] (d) High LASSO + COX model score + low CT muscle mass group (double high-risk group)

[0110] Preoperative peripheral markers and imaging findings in this group of patients indicated a high risk. However, postoperative follow-up revealed that while the initial survival curve was relatively stable, the risk gradually accumulated over time, leading to a significant decline in long-term survival, making this group the worst prognostic group among all groups. This phenomenon suggests that relying solely on CT or LASSO+COX model scores may lead to the risk of missed diagnosis or misdiagnosis, especially in the early stages, and may underestimate the actual risk level.

[0111] like Figure 5 As shown in the right figure, by further combining the LASSO score with CT imaging muscle mass indicators (such as SMI) to construct a four-quadrant grouping (high / low score × high / low muscle mass), the results showed a more significant difference in survival rates among the four groups (p = 0.00046), demonstrating the higher stratification accuracy of the combined model. This further supports the clinical value of combining the peripheral blood LASSO + COX model score with CT imaging, suggesting that in actual postoperative management, detailed risk stratification can help accurately identify high-risk patients and thus optimize clinical decision-making and management strategies.

[0112] S3. Sarcopenia is confirmed based on the risk group classification results combined with CT imaging results.

[0113] In the embodiments, liver transplant recipients classified as a low-risk group for sarcopenia are continuously followed up, evaluated, and monitored, and liver transplant recipients classified as a high-risk group for sarcopenia are further subjected to CT examinations and sarcopenia is confirmed based on the CT imaging results.

[0114] In summary, the peripheral blood and CT-based sarcopenia classification method for liver transplant recipients provided by the embodiments of the present invention constructs a LASSO+COX model score and uses preoperative peripheral blood biochemical indicators to replace some CT imaging examinations for muscle status monitoring. It achieves significant clinical and economic benefits while maintaining accuracy, can reduce the frequency of CT examinations, save examination costs, reduce radiation risks, and improve follow-up test compliance. It is suitable for promotion and use in grassroots, telemedicine and the elderly population, and has established a new data-driven paradigm for sarcopenia screening.

[0115] Based on the same inventive concept, Figure 6As shown, an embodiment of the present invention further provides a sarcopenia classification device 600 for liver transplant recipients based on peripheral blood and CT, comprising: a prediction factor screening module 610, a risk classification module 620, and a sarcopenia diagnosis module 630.

[0116] The predictor screening module 610 is used to perform a univariate COX regression primary screening on the collected peripheral blood biochemical indicators of the liver transplant recipient to obtain potential predictors related to sarcopenia, and perform a LASSO regression screening on the potential predictors to obtain core predictors related to sarcopenia;

[0117] The risk classification module 620 is used to multiply each core prediction factor by its regression coefficient calculated in the LASSO regression and add them together to obtain a LASSO+COX model score, and classify the liver transplant recipients into different risk groups according to the LASSO+COX model score.

[0118] The sarcopenia diagnosis module 630 is used to diagnose sarcopenia based on the risk group classification results combined with the CT imaging examination results.

[0119] It should be noted that the sarcopenia classification device for liver transplant recipients based on peripheral blood and CT provided in the above embodiment belongs to the same inventive concept as the sarcopenia classification method for liver transplant recipients based on peripheral blood and CT. The specific implementation process is detailed in the embodiment of the sarcopenia classification method for liver transplant recipients based on peripheral blood and CT, which will not be repeated here.

[0120] The specific implementation methods described above provide a detailed description of the technical solutions and beneficial effects of the present invention. It should be understood that the above is only the most preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, supplements and equivalent substitutions made within the scope of the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT, characterized in that: The following steps are involved: The peripheral blood biochemical indicators collected from liver transplant recipients were screened by univariate COX regression to obtain potential predictors related to sarcopenia. LASSO regression was then performed on the potential predictors to obtain core predictors related to sarcopenia. Each core predictor was multiplied by its regression coefficient calculated in LASSO regression and the results were added together to obtain the LASSO+COX model score. Liver transplant recipients were divided into different risk groups according to the LASSO+COX model score. Sarcopenia was diagnosed based on the risk group division results combined with CT imaging results.

2. The method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT according to claim 1, characterized in that: In the univariate COX regression, the significance level P<0.1 was set as the initial screening standard.

3. The method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT according to claim 1 or 2, characterized in that: A total of 14 potential predictive factors were screened, including recipient age, alpha-fetoprotein level, tumor number, anti-HBc antibody level, microvascular invasion, albumin / globulin ratio, international normalized ratio, intraoperative blood loss, platelet count, mean corpuscular hemoglobin, lymphocyte percentage, triglycerides, total cholesterol, and diagnosis of nonalcoholic steatohepatitis.

4. The method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT according to claim 1, wherein: In LASSO regression screening, the optimal performance is achieved when log(λ)=-2.57, where λ represents the penalty parameter.

5. The method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT according to claim 1 or 4, characterized in that: A total of 7 core predictive factors were screened, including: diagnosis of non-alcoholic steatohepatitis, mean corpuscular hemoglobin, lymphocyte percentage, intraoperative blood loss, anti-HBc antibody level, albumin / globulin ratio, and total cholesterol.

6. The method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT according to claim 5, characterized in that: The LASSO+COX model score is obtained by multiplying each core predictor by its regression coefficient calculated in the LASSO regression and adding them together. The calculation formula is as follows: LASSO+COX model score = (α1×diagnosis of nonalcoholic steatohepatitis) + (α2×mean corpuscular hemoglobin) + (α3×lymphocyte percentage) + (α4×intraoperative blood loss) + (α5×anti-HBc antibody level) + (α6×albumin / globulin ratio) + (α7×total cholesterol), where α1, α2, α3, α4, α5, α6, and α7 represent the regression coefficients corresponding to each core predictor, respectively.

7. The method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT according to claim 1, wherein: Liver transplant recipients are divided into different risk groups according to the LASSO+COX model score, including: Liver transplant recipients whose LASSO+COX model scores were less than the risk threshold were divided into the low-risk group for sarcopenia; Liver transplant recipients with LASSO+COX model scores greater than or equal to the risk threshold were divided into a high-risk group for sarcopenia.

8. The method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT according to claim 7, characterized in that: The risk threshold is set at 3.

5.

9. The method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT according to claim 7, characterized in that: The diagnosis of sarcopenia based on the risk group classification results combined with CT imaging results includes: Liver transplant recipients classified as low-risk for sarcopenia will undergo continuous follow-up evaluation and monitoring, while liver transplant recipients classified as high-risk for sarcopenia will undergo further CT examinations and sarcopenia will be confirmed based on the CT imaging results.

10. A device for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT, implemented using the method for classifying sarcopenia in liver transplant recipients based on peripheral blood and CT according to any one of claims 1 to 9, characterized in that: include: Predictor screening module, risk classification module, and sarcopenia diagnosis module; The predictor screening module is used to perform a single factor COX regression primary screening on the peripheral blood biochemical indicators collected from liver transplant recipients to obtain potential predictors related to sarcopenia, and perform LASSO regression screening on the potential predictors to obtain core predictors related to sarcopenia; The risk classification module is used to multiply each core prediction factor by its regression coefficient calculated in LASSO regression and cumulatively calculate the LASSO+COX model score, and divide the liver transplant recipients into different risk groups according to the size of the LASSO+COX model score; The sarcopenia diagnosis module is used to diagnose sarcopenia based on the risk group division results combined with CT imaging examination results.

Images