Method for constructing breast cancer postoperative upper limb edema prediction model based on ultrasonic scanning axillary arteriovenous flow

Abstract

The invention relates to the field of clinical medicine, and discloses a method for constructing a breast cancer postoperative upper limb edema prediction model based on ultrasonic scanning axillary arteriovenous flow. The method comprises the following steps: S1, collecting data; s2, measuring hemodynamic data; s3, calculating the upper limb venous return proportion PVR: calculating the upper limb venous return proportion PVR according to the ultrasonic measurement result obtained in S2, namely the ratio of axillary vein reflux to axillary artery reflux in unit time, drawing an ROC curve, calculating a Youden index, determining an optimal truncation point of the PVR, and dividing patients into two groups; s4, performing variable screening; s5, constructing a prediction model; and S6, prediction value analysis. The breast cancer postoperative upper limb edema prediction model constructed by the model construction method can effectively predict breast cancer postoperative upper limb edema of a subject, and is high in prediction efficiency, accurate in prediction result, wide in application range and low in cost.

Description

Technical Field

[0001] The present application relates to the field of clinical medicine, and more specifically, to a method for constructing a model for predicting upper limb edema after breast cancer surgery based on ultrasound scanning of axillary artery and vein flow. Background Art

[0002] Breast cancer is the most common malignancy in women, with 2,261,419 new cases worldwide in 2020, accounting for 24.5% of cancer diagnoses in women. The average 5-year and 10-year survival rates for women with non-metastatic invasive breast cancer are 91% and 84%, respectively (American Society of Clinical Oncology, 2020). Although the mortality rate from breast cancer has dropped by about 35% in the past 30 years, the risk of complications has also increased with the increasing survival rate, one of the most serious complications being upper limb lymphedema and dysfunction secondary to cancer treatment.

[0003] The International Society of Lymphology defines lymphedema as “internal and / or external manifestations of lymphatic insufficiency and disrupted lymphatic transport” (International Society of Lymphology, 2016). Breast cancer-related lymphedema (BCRL) is characterized by abnormal accumulation of interstitial fluid due to impaired lymphatic drainage after cancer treatment, resulting in upper extremity swelling. Early lymphedema, defined as subclinical or stage 1, has been shown to be reversible. Once lymphedema progresses to stage 2 or 3, it becomes chronic and can be alleviated but is irreversible.

[0004] Previous studies have shown that risk factors for upper limb lymphedema in breast cancer patients include the number of lymph nodes removed, the use of chemotherapy, radiotherapy, and increased body mass index (BMI). Axillary lymph node dissection (ALND) is used to predict prognosis, improve local control, and assist in staging.

[0005] The normal body fluid volume and tissue volume are relatively constant, and this constancy depends on the body's regulation of the balance of fluid exchange between the body and the outside and between the inside and outside of blood vessels. Under normal circumstances, fluid exchange between interstitial fluid and plasma is constantly taking place, so that the generation and return of tissue fluid maintain a dynamic balance, and this balance is mainly determined by several factors such as effective fluid static pressure, effective colloid osmotic pressure and lymphatic return. When one or more of the above factors are out of balance simultaneously or successively, edema may occur.

[0006] In order to reduce the incidence of upper limb edema and functional disability after breast cancer surgery, it is important to identify the risk factors for BCRL and implement appropriate prevention strategies before it occurs. However, there is currently no postoperative edema-related measurement model that focuses on the physiological and anatomical characteristics of the human body. Summary of the invention

[0007] The purpose of the invention of this application is to provide a method for constructing a prediction model for upper limb edema after breast cancer surgery based on ultrasound scanning of axillary artery and vein flow, and a prediction model, so as to reduce the incidence of upper limb edema and dysfunction after breast cancer surgery, identify the risk factors of BCRL and implement appropriate prevention strategies before occurrence, and provide reliable reference evaluation information for patient prognosis.

[0008] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical solutions:

[0009] In a first aspect, the present application provides a method for constructing a prediction model for upper limb edema after breast cancer surgery based on ultrasound scanning of axillary artery and vein flow, comprising:

[0010] S1, Data collection: Acquisition of axillary ultrasound images for model construction;

[0011] S2, hemodynamic data measurement: axillary hemodynamic data were measured based on the axillary ultrasound images obtained in S1;

[0012] S3, calculation of upper limb venous return ratio (PVR): Calculate the upper limb venous return ratio (PVR) based on the ultrasound measurement results obtained in S2, that is, the ratio of the axillary vein return volume to the axillary artery return volume per unit time, draw the ROC curve, calculate the Youden index, determine the optimal cutoff point of PVR, and divide the patients into two groups;

[0013] S4, variable screening: PVR and other clinical variables were included in Lasso regression screening to screen key variables that affect the occurrence of postoperative lymphedema;

[0014] S5, constructing a prediction model: incorporating the selected variables into the ROC curve prediction model for breast cancer postoperative lymphedema, calculating the AUC value, calibration graph and DCA curve of the model, and evaluating the prediction accuracy and discrimination ability of the model;

[0015] S6, Prediction value analysis: According to the ROC curve prediction model analysis, the prediction value of the clinical prediction model combined with PVR for breast cancer postoperative lymphedema was analyzed, and the prediction model effect corresponding to the AUC value range was defined.

[0016] Furthermore, in step S2, an ultrasound device is used to measure axillary hemodynamic data, including axillary artery cross-sectional area S a , axillary vein cross-sectional area S v , axillary artery peak systolic velocity PSV, end diastolic velocity EDV and axillary vein maximum blood flow velocity Vmax.

[0017] Further, in step S3,

[0018]

[0019] Furthermore, in step S3, those below the optimal PVR cutoff point are the dominant lymphatic return group (high-risk group), and those above the optimal PVR cutoff point are the dominant venous return group (low-risk group).

[0020] Furthermore, in step S6, for the ROC curve constructed by the prediction model,

[0021] AUC = 1, indicating that the prediction model is perfectly established;

[0022] AUC = [0.85, 0.95], indicating that the prediction model is well established;

[0023] AUC = [0.7, 0.85], indicating that the prediction model establishment effect is average;

[0024] AUC = [0.5, 0.7], indicating that the prediction model establishment effect is low, but it can be used for prediction;

[0025] AUC ≤ 0.5, indicating that the prediction model has no predictive value.

[0026] In a second aspect, the present application provides a model for predicting upper limb edema after breast cancer surgery constructed by the construction method described in the first aspect.

[0027] In a third aspect, the present application provides a system for evaluating upper limb edema after breast cancer surgery in a subject, comprising:

[0028] Data acquisition module: used to obtain the axillary ultrasound image of the subject;

[0029] Data measurement module: used to measure axillary hemodynamic data;

[0030] Data processing module: used to calculate the upper limb venous return ratio PVR and determine the optimal cutoff point of PVR;

[0031] Variable screening module: used to screen key variables that affect the occurrence of postoperative lymphedema;

[0032] Model evaluation module: used to evaluate the prediction accuracy and discrimination ability of the model;

[0033] Risk assessment module: According to the optimal cutoff point of PVR, the risk level is divided into high risk or low risk.

[0034] In a fourth aspect, the present application provides a product for evaluating upper limb edema after breast cancer surgery in a subject, including a prediction model for upper limb edema after breast cancer surgery constructed by the construction method described in the first aspect.

[0035] In a fifth aspect, the present application provides an electronic device, including:

[0036] a memory for storing instructions executable by a processor;

[0037] A processor is used to execute the instructions to implement the construction method as described in the first aspect.

[0038] In a sixth aspect, the present application provides a computer-readable medium storing a computer program code, wherein the computer program code, when executed by a processor, implements the construction method as described in the first aspect.

[0039] In summary, this application has the following beneficial effects:

[0040] The upper limb edema prediction model after breast cancer surgery obtained by the model construction method of the present application can effectively predict the upper limb edema after breast cancer surgery of subjects, with high prediction efficiency, accurate prediction results, wide application range and low cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 : ROC curve diagram in step S3 of the present embodiment;

[0042] Figure 2 : A graph showing the change trend of upper limb circumference between the two groups in step S3 of the embodiment of the present application;

[0043] Figure 3 : DASH score change trend diagram between the two groups in step S3 of the embodiment of the present application;

[0044] Figure 4 : Lasso coefficient distribution in step S4 of the embodiment of the present application;

[0045] Figure 5 : The nomogram for the risk of postoperative lymphedema in step S4 of the embodiment of the present application;

[0046] Figure 6 : ROC curve diagram and calibration diagram in step S5 of the present embodiment;

[0047] Figure 7 : DCA curve diagram of the model in step S5 of the embodiment of the present application. DETAILED DESCRIPTION

[0048] The technical scheme and effects of the present application are further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the invention, rather than to limit the invention.

[0049] Example

[0050] This embodiment discloses a method for constructing a model for predicting upper limb edema after breast cancer surgery based on ultrasound scanning of axillary artery and vein flow, comprising:

[0051] S1, collect preoperative axillary ultrasound images, preoperative data, and postoperative clinical follow-up data of breast cancer patients;

[0052] Inclusion criteria:

[0053] ① Age 18-69 years old; ② No gender restrictions; ③ Breast tumors confirmed by histological pathology, breast cancer, stage II-III; ④ Axillary lymph nodes positive by clinical palpation, or ⑤ Axillary lymph node puncture pathology positive, or ⑥ Multidisciplinary treatment cooperation team (MDT) recommends axillary lymph node dissection; ⑦ Good physical status score (0-2); ⑧ No serious organ complications; ⑨ No congenital or acquired diseases that affect the normal morphology and functional activities of the upper limbs; ⑩ Informed consent, able to understand and comply with the research requirements.

[0054] Exclusion criteria:

[0055] ① Pregnancy or lactation; ② Inflammatory breast cancer; ③ Clinical discovery of metastatic lesions; ④ Negative sentinel lymph node biopsy; ⑤ History of trauma or surgery to the upper limbs, shoulders, chest, or back; ⑥ History of previous local radiotherapy; ⑦ History of other tumors; ⑧ Vascular embolic disease; ⑨ Unable to comply with the requirements of the clinical trial for any reason.

[0056] All patients in the group had their basic BMI, body fat percentage and other basic characteristics recorded before surgery; imaging professionals performed ultrasound examinations of the diameter, velocity and flow of the axillary artery and vein before surgery; measured the arm circumference of both upper limbs and evaluated upper limb function before surgery and one week, one month, six months and one year after surgery; surgical specimens were routinely examined for pathological morphology, immunohistochemistry and FISH examinations; regular follow-up was performed after surgery and adjuvant treatment, and the observation content was routine for breast cancer. In addition, the patient's compliance with postoperative functional exercise and the intensity, frequency and duration of functional exercise will be included in the questionnaire survey and recorded for statistical analysis.

[0057] The incidence of upper limb lymphedema was defined as the proportion of patients with an increase in arm circumference of more than 10% (equivalent to an increase in upper arm volume of more than 1%) among all patients. The incidence of upper limb dysfunction was defined as the proportion of patients with a DASH score of more than 30 among all patients. The degree of disability of the arm, shoulder, and hand will be assessed based on the DASH score.

[0058] Variables that should be collected include:

[0059] Operation time, age (y), height (cm), weight (kg), BMI (18.5-24), body fat percentage calculation | waist circumference (cm), body fat percentage calculation | parameter a, body fat percentage calculation | parameter b, body fat percentage calculation | body fat weight (kg), body fat percentage calculation | body fat percentage (%), HDL-C (mmol / L), LDL-C (mmol / L), illness time (months), blood pressure (mmHg), blood sugar (mmo / L), TC (mmol / L), TG (mmol / L), tumor lesions (1 unilateral; 2 bilateral), axillary lymph node puncture or excisional biopsy (1 negative; 2 positive), clinical axillary lymph node metastasis (1 no; 2 yes), MDT recommendation for axillary dissection (0 controversial; 1 no; 2 yes), concomitant diseases (0 no; 1 diabetes; 2 anemia; 3 malnutrition; 4 others), physical status score PS, axillary artery cross-sectional area (cm 2 )、axillary artery velocity (cm / s)|PSV、axillary artery velocity (cm / s)|EDV

[0060] , arterial perfusion per unit time (cm 3 ), axillary vein cross-sectional area (cm 2 ), axillary vein velocity (cm / s), venous return per unit time (cm 3 ), axillary arteriovenous flow difference per unit time (cm 3), venous return ratio, lymphatic venous return ratio, neoadjuvant targeted therapy (1NO; 2YES), surgery (1 modified radical mastectomy; 2 breast-conserving radical mastectomy; 3 reconstruction; 4 sentinel resection; 5 sentinel preservation), axillary vein branch ligation (1NO; 2YES), intercostal brachial nerve resection (1NO; 2YES), (1Level1; 2Level2; 3Level3), tumor site (0 unknown; 1 medial; 2 lateral; 3 central; 4 whole breast), pathological type (1 invasive ductal carcinoma; 2 invasive lobular carcinoma; 3 other types; 4 unclassified), histological grade (1G1; 2G2; 3G3; 4Unclarified), para-cancer / vascular invasion (0 UNKknown; 1NO; 2YES), lymph node metastasis (1NO; 2YES), pathological stage (1ⅡA; 2ⅡB; 3ⅢA; 4ⅢB; 5; ⅢC; 6Ⅳ), HR (0Unknown; 1N; 2Y), HER2 (0Unknown; 1N; 2Y), tumor size (cm), number of axillary vein branches ligated, total number of lymph nodes, number of metastatic lymph nodes, adjuvant chemotherapy (1No; 2Yes), start time of adjuvant chemotherapy; adjuvant radiotherapy (1No; 2Yes), start time of adjuvant radiotherapy, adjuvant endocrine therapy (1No; 2Yes), start time of adjuvant endocrine therapy, preoperative|affected arm circumference (cm), preoperative|healthy arm circumference (cm) , 1W after surgery|Postoperative wound infection (1NO; 2YES), 1W after surgery|Start functional exercise (1NO; 2YES), 1W after surgery|Inappropriate force contusion (1NO; 2YES), 1W after surgery|Adjuvant chemotherapy (1NO; 2YES), 1W after surgery|Adjuvant radiotherapy (1NO; 2YES), 1W after surgery|Number of exercises per day, 1W after surgery|Daily exercise time (min), 1W after surgery|DASH score, 1W after surgery|Impaired arm circumference (cm), 1W after surgery|Healthy arm circumference (cm), 1M after surgery|Postoperative wound infection (1NO; 2YES), 1M after surgery|Start functional exercise (1NO; 2YES), 1M after surgery|Inappropriate force contusion (1NO; 2YES), 1M after surgery|adjuvant chemotherapy (1NO; 2YES), 1M after surgery|adjuvant radiotherapy (1NO; 2YES), 1M after surgery|number of daily exercises, 1M after surgery|daily exercise time (min), 1M after surgery|DASH score, 1M after surgery|affected arm circumference (cm), 1M after surgery|percentage of arm circumference increase, 1M after surgery|healthy arm circumference (cm), 6M after surgery|postoperative wound infection (1NO; 2YES), 6M after surgery|start functional exercise (1NO; 2YES), 6M after surgery|improper force contusion (1NO; 2YES), 6M after surgery|adjuvant chemotherapy (1NO; 2YES), 6M after surgery|adjuvant radiotherapy (1NO;2YES), 6M after surgery|daily exercise times, 6M after surgery|daily exercise time (min), 6M after surgery|DASH score, 6M after surgery|affected arm circumference (cm), 6M after surgery|increase ratio of arm circumference, 6M after surgery|healthy arm circumference (cm), 1Y after surgery|postoperative wound infection (1NO; 2YES), 1Y after surgery|start functional exercise (1NO; 2YES), 1Y after surgery|improper force contusion (1NO; 2YES), 1Y after surgery|adjuvant chemotherapy (1NO; 2YES), 1Y after surgery|adjuvant radiotherapy (1NO; 2YES), 1Y after surgery|daily exercise times, 1Y after surgery|daily exercise time (min), 1Y after surgery|DASH score, 1Y after surgery|affected arm circumference (cm), 1Y after surgery|increase ratio of arm circumference, 1Y after surgery|healthy arm circumference (cm). ;

[0061] Table 1 Disabilities of the upper limb, shoulder, and hand (DASH) scoring table

[0062]

[0063]

[0064] DASH value = [(sum of A and B) - 30] ÷ 1.20, or [(A + B) / 30 - 1] × 25. A DASH value of 0 indicates that the upper limb function is completely normal, and a DASH value of 100 indicates that the upper limb function is extremely limited.

[0065] S2, hemodynamic data measurement: axillary hemodynamic data were measured based on the axillary ultrasound images obtained in S1;

[0066] Test equipment:

[0067] Instrument model: Mindray resonar7;

[0068] Probe type: L14-5 linear array probe;

[0069] Parameter setting: frequency 7.5MHz;

[0070] Calculation method:

[0071] Grayscale ultrasound was used to scan the axillary artery and the accompanying axillary vein. The presence of plaque in the artery and thrombosis in the vein were evaluated. The cross-sectional area was recorded and measured by making a transverse cut at the proximal end of the main artery and vein at the axillary level (where there were no obvious branches): S a is the cross-sectional area of ​​the axillary artery, S vis the cross-sectional area of ​​the axillary vein. Adjust the color Doppler blood flow scale so that there is no color aliasing in the lumen. Adjust the Doppler angle to less than or equal to 60 degrees, avoid the stenosis of the lumen, and measure the axillary artery systolic peak velocity PSV (peaksystolic velocity), end diastolic velocity EDV (end diastolic velocity) and the maximum blood flow velocity Vmax of the axillary vein.

[0072] S3, PVR (proportions of upper limb venous reflux) is the proportion of upper limb venous reflux. PVR is calculated based on ultrasound measurement results, that is, the ratio of axillary venous reflux to axillary artery reflux per unit time. Based on the presence or absence of upper limb edema in S1 as the classification basis, the receiver operating characteristic (ROC) curve was drawn, the Youden index was calculated, and the optimal cutoff point of PVR was determined. Then, the entire cohort was divided into two groups: the group with dominant lymphatic return (lower PVR or lower axillary vein / lymphatic return ratio) and the group with dominant venous return (higher PVR or higher axillary vein / lymphatic return ratio). The average PVR of the upper limb was 25.94%±18.87%. The optimal cutoff point of PVR was determined to be 15.223% (such as Figure 1 shown).

[0073] The PVR cutoff value can significantly distinguish the high-risk group (dominant lymphatic return group) and the low-risk group (dominant venous return group) of axillary lymphedema. The following is the data verification:

[0074] Fisher's exact test showed that the incidence of lymphedema between the two groups of patients was significantly different at 6 months (1 / 110 vs. 10 / 51, ARR 18.70%, P = 0.000, OR 26.585, 95% CI [3.299, 214.244]) and 12 months (1 / 75 vs. 11 / 35, ARR 30.10%, P = 0.000, OR 33.917, 95% CI [4.161, 276.485]) after surgery. The incidence of upper limb edema in the dominant venous return group (RR at 6 months: 0.125 [0.019, 0.813], RR at 12 months: 0.110 [0.017, 0.723]) was significantly lower than that in the dominant lymphatic return group (RR at 6 months: 3.326 [2.413, 4.585], RR at 12 months: 3.743 [2.541, 5.513]). There was a significant difference in the change of upper limb circumference between the two groups, and they showed different trends (such as Figure 2 shown).

[0075] By general linear model analysis, there were significant differences between the two groups in terms of reported upper limb dysfunction and investigator-rated DASH scores (the DASH scores of the dominant venous return group were 8.77±3.657, 4.13±2.615, and 1.76±1.778 at 1, 6, and 12 months, respectively, while the DASH scores of the dominant lymphatic return group were 11.46±3.148, 7.25±3.794, and 3.98±2.651, respectively). Over time, the DASH scores of both groups decreased (Greenhouse Geisser test, F=324.130, P=0.000), and the downward trend was similar (Greenhouse Geisser test, F=1.492, P=0.225); the mean DASH score of the dominant venous return group was significantly lower than that of the dominant lymphatic return group (F=25.839, P=0.000) (e.g. Figure 3 shown).

[0076] S4. PVR and other clinical variables (surgical group (1 retain axillary vein; 2 do not retain axillary vein), hospital (1A hospital; 2B hospital), age, BMI, body fat percentage, diabetes (1 no; 2 diabetes), lymph node metastasis (1 NO; 2 YES), paracancerous / vascular invasion (0 UNK known; 1 NO; 2 YES), histological grade (1 G1; 2 G2; 3 G3; 4 Unclarified), stage (1 Ⅱ A; 2 Ⅱ B; 3 Ⅲ A; 4 Ⅲ B; 5 Ⅲ C; 6 Ⅳ), Tumor size (cm), PVR, intercostobrachial nerve resection (1NO; 2YES), number of axillary vein tributaries ligated, lymph node dissection range (0Level1-2; 1Level3), total number of lymph nodes, number of metastatic lymph nodes, chemotherapy (1no; 2yes), radiotherapy (1no; 2yes), postoperative wound infection (0no; 1yes), postoperative improper force contusion (0no; 1yes), lymphatic venous return group (1dominant lymphatic return group; 2dominant venous return group)) were included in the lasso regression screening, and finally the effective variables were screened out. When performing lasso regression analysis, choosing the best penalty coefficient λ is a key step, which determines which variables will be retained in the model. The lasso coefficient distribution of 22 features was evaluated by 20-fold cross validation (such as Figure 4 The model finally determined that the dominant lymphatic drainage group, axillary lymph node dissection (ALND) without preserving axillary vein branches, level 3 lymph node dissection, and postoperative wound infection were the key variables affecting the results, thus evaluating the risk of postoperative lymphedema through the nomogram (such as Figure 5 shown);

[0077] S5, the variables were included to construct the ROC curve prediction model for lymphedema after breast cancer surgery.

[0078] The model we developed for predicting postoperative upper limb edema achieved an AUC value of 0.919 (95% CI: 0.778-1.000) in the cohort. The calibration plot showed that the intercept was 0 and the slope was 1, indicating that the model was well calibrated, the predicted probability matched the observed probability, and the C index was 0.92 (95% CI: 0.52-0.99), indicating good discrimination ability (such as Figure 6 As shown in Figure 2 ). From the DCA curve, the net benefit of the prediction model for the internal validation set is significantly higher than the two extreme cases, indicating that the nomogram model has higher net benefit and prediction accuracy. The AUC value of the optimized model obtained through internal validation is 0.910 (as shown in Figure 2 ). Figure 7 shown).

[0079] S6, Predictive value analysis: According to the ROC curve prediction model analysis, the predictive value of the clinical prediction model combined with PVR for breast cancer surgery lymphedema was analyzed, and the prediction model effect corresponding to the AUC value range was defined.

[0080] For the ROC curve constructed by the prediction model,

[0081] AUC = 1, indicating that the prediction model is perfectly established;

[0082] AUC = [0.85, 0.95], indicating that the prediction model is well established;

[0083] AUC = [0.7, 0.85], indicating that the prediction model establishment effect is average;

[0084] AUC = [0.5, 0.7], indicating that the prediction model establishment effect is low, but it can be used for prediction;

[0085] AUC ≤ 0.5, indicating that the prediction model has no predictive value.

[0086] This specific embodiment is merely an explanation of the present application and is not a limitation of the present application. After reading this specification, those skilled in the art may make modifications to the present embodiment without any creative contribution as needed, but such modifications are protected by the patent law as long as they are within the scope of the claims of the present application.

Claims

1. A method for constructing a prediction model for upper limb edema after breast cancer surgery based on ultrasound scanning of axillary artery and vein flow, characterized in that: include: S1, Data collection: Acquisition of axillary ultrasound images for model construction; S2, hemodynamic data measurement: axillary hemodynamic data were measured based on the axillary ultrasound images obtained in S1; S3, calculation of upper limb venous return ratio (PVR): Calculate the upper limb venous return ratio (PVR) based on the ultrasound measurement results obtained in S2, that is, the ratio of the axillary vein return volume to the axillary artery return volume per unit time, draw the ROC curve, calculate the Youden index, determine the optimal cutoff point of PVR, and divide the patients into two groups; S4, variable screening: PVR and other clinical variables were included in Lasso regression screening to screen the key variables that affect the occurrence of postoperative lymphedema; S5, constructing a prediction model: incorporating the selected variables into the ROC curve prediction model for breast cancer postoperative lymphedema, calculating the AUC value, calibration graph and DCA curve of the model, and evaluating the prediction accuracy and discrimination ability of the model; S6, Predictive value analysis: According to the ROC curve prediction model analysis, the predictive value of the clinical prediction model combined with PVR for breast cancer surgery lymphedema was analyzed, and the prediction model effect corresponding to the AUC value range was defined.

2. The construction method according to claim 1, characterized in that: In step S2, an ultrasound device is used to measure axillary hemodynamic data, including the cross-sectional area S of the axillary artery. a , axillary vein cross-sectional area S v , axillary artery peak systolic velocity PSV, end diastolic velocity EDV and axillary vein maximum blood flow velocity Vmax.

3. The construction method according to claim 2, characterized in that: In step S3, 4. The construction method according to claim 1, characterized in that: In step S3, those below the optimal PVR cutoff point are the dominant lymphatic return group (high-risk group), and those above the optimal PVR cutoff point are the dominant venous return group (low-risk group).

5. The construction method according to claim 1, characterized in that: In step S6, for the ROC curve constructed by the prediction model, AUC = 1, indicating that the prediction model is perfectly established; AUC = [0.85, 0.95], indicating that the prediction model is well established; AUC = [0.7, 0.85], indicating that the prediction model establishment effect is average; AUC = [0.5, 0.7], indicating that the prediction model establishment effect is low, but it can be used for prediction; AUC ≤ 0.5, indicating that the prediction model has no predictive value.

6. A prediction model for upper limb edema after breast cancer surgery constructed by the construction method described in any one of claims 1-5.

7. A system for evaluating upper limb edema after breast cancer surgery in a subject, characterized in that: include: Data acquisition module: used to obtain the axillary ultrasound image of the subject; Data measurement module: used to measure axillary hemodynamic data; Data processing module: used to calculate the upper limb venous return ratio PVR and determine the optimal cutoff point of PVR; Variable screening module: used to screen key variables that affect the occurrence of postoperative lymphedema; Model evaluation module: used to evaluate the prediction accuracy and discrimination ability of the model; Risk assessment module: According to the optimal cutoff point of PVR, the risk level is divided into high risk or low risk.

8. A product for evaluating upper limb edema after breast cancer surgery in a subject, characterized in that: It comprises a model for predicting upper limb edema after breast cancer surgery constructed by the construction method according to any one of claims 1 to 5.

9. An electronic device, characterized in that: include: a memory for storing instructions executable by a processor; A processor, configured to execute the instructions to implement the construction method according to any one of claims 1 to 5.

10. A computer readable medium storing computer program code, characterized in that: When the computer program code is executed by a processor, the construction method according to any one of claims 1 to 5 is implemented.

Images