Model for predicting early gastric cancer lymph node metastasis risk and construction method and application thereof

Abstract

The invention relates to a model for predicting an early gastric cancer lymph node metastasis risk and a construction method and application thereof. The input variables of the model comprise gender, tumor position, tumor maximum diameter, pathological differentiation degree, pT staging condition, submitted lymph node number and lymphatic vessel invasion condition, and the output value of the model is lymph node metastasis prediction probability. A model for predicting the lymph node metastasis risk of the early gastric cancer is designed, the lymph node metastasis risk of a patient with the early gastric cancer can be accurately evaluated through the model, a model for predicting the number of the lymph node metastasis is further obtained, a corrected pN stage (pNM) is provided based on the predicted number of the lymph node metastasis, and the establishment of the pNM stage can be used for predicting the lymph node metastasis risk of the patient with the early gastric cancer. The method can be applied to diagnosis of early gastric cancer patients who cannot perform sufficient lymph node cleaning due to various reasons, is helpful for clinicians to evaluate and calculate early gastric cancer patients who do not reach the sufficient number of detected lymph nodes, and provides a new way for accurate clinical evaluation.

Description

Technical Field

[0001] The present invention belongs to the field of biomedical technology and relates to a model for predicting the risk of lymph node metastasis of early gastric cancer and a construction method and application thereof. Background Art

[0002] Gastric cancer is one of the most common malignant tumors. According to the definition of the Japanese Society of Gastrointestinal Endoscopy in 1962, early gastric cancer (EGC) is defined as lesions confined to the mucosa and submucosa, regardless of the status of lymph node metastasis. Although some forms of early gastric cancer can be treated by endoscopic resection, in most cases, surgical treatment remains the main treatment for patients with early gastric cancer.

[0003] There is a certain chance of recurrence after early gastric cancer resection, and the recurrence time ranges from 4 months to more than 10 years. Among them, lymph node metastasis has been shown to be an independent risk factor associated with the prognosis of patients with early gastric cancer. Studies have found that after increasing the number of lymph nodes sent for examination, the 5-year survival rate of gastric cancer patients has significantly increased. The choice of further treatment options for patients with early gastric cancer after surgery depends on the number of metastatic lymph nodes, that is, the pN stage. Therefore, although there is no clear requirement for the presence of lymph node metastasis in the treatment of early gastric cancer, considering the important influence of the number of metastatic lymph nodes on the accurate assessment of the patient's postoperative pN stage and even pTNM stage as well as subsequent treatment and prognosis, patients with early gastric cancer also need to undergo adequate lymph node dissection. At the same time, adequate lymph node dissection will also significantly increase the detection of the number of metastatic lymph nodes.

[0004] So far, relevant studies or guidelines have not clearly stipulated the number of lymph nodes that need to be cleared for patients with early gastric cancer, and various problems have arisen based on this. Due to insufficient lymph node dissection, metastatic lymph nodes may not be fully exposed, so the pN staging obtained by clinicians based on the number of metastatic lymph nodes will be biased, that is, the final pTNM staging may be underestimated. This directly leads to significant deviations in the prognostic assessment of early gastric cancer patients by clinicians, and also directly affects the choice of treatment options for postoperative patients. Therefore, it is one of the urgent problems to formulate a sufficient number of lymph nodes for examination for patients with early gastric cancer and to conduct model evaluation and predict the accurate number of metastatic lymph nodes for patients with insufficient lymph nodes. Summary of the invention

[0005] In view of the deficiencies in the prior art and actual needs, the present invention provides a model for predicting the risk of lymph node metastasis of early gastric cancer and a construction method and application thereof, in order to accurately assess the risk of lymph node metastasis in patients with early gastric cancer and to evaluate and calculate the stages of early gastric cancer patients who do not have sufficient lymph nodes for examination.

[0006] To achieve this object, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides a model for predicting the risk of lymph node metastasis in early gastric cancer, wherein the input variables of the model include gender, tumor location, maximum tumor diameter, degree of pathological differentiation, number of lymph nodes submitted for examination, and lymphatic vessel invasion; the output value of the model is the predicted probability of lymph node metastasis, and the calculation formula is shown in formula (1);

[0008]

[0009] Wherein, P is the predicted probability of lymph node metastasis; Logit P = -0.358×a+b×c+0.660×d+0.6388×e+0.933×f+1.719×g-0.898×h-2.609;

[0010] a represents gender, male: a=0, female: a=1;

[0011] b×c represents the location of the tumor, the tumor is located in the lower 1 / 3 of the stomach: c=0; the tumor is located in the middle 1 / 3 of the stomach: b=-0.059, c=1; the tumor is located in the upper 1 / 3 of the stomach: b=-0.616, c=2; the tumor exceeds 1 / 3 of the stomach: b=-0.466, c=3;

[0012] d represents the maximum diameter of the tumor: d = 0 if the maximum diameter of the tumor is ≤ 2 cm; d = 1 if the maximum diameter of the tumor is > 2 cm;

[0013] e represents the degree of pathological differentiation, moderately differentiated or well differentiated: e = 0; poorly differentiated or undifferentiated: e = 1;

[0014] f represents pT stage, pT1a stage: f = 0; pT1b stage: f = 1;

[0015] g represents the status of lymphatic vessel invasion, no lymphatic vessel invasion: g = 0; lymphatic vessel invasion: g = 1;

[0016] h represents the number of lymph nodes submitted for examination. If the number of lymph nodes submitted for examination is ≤20, h=0; if the number of lymph nodes submitted for examination is >20, h=1.

[0017] The present invention designs a model for predicting the risk of lymph node metastasis of early gastric cancer. The model can accurately assess the risk of lymph node metastasis in patients with early gastric cancer and has auxiliary clinical diagnosis value.

[0018] In a second aspect, the present invention provides a method for constructing a model for predicting the risk of lymph node metastasis of early gastric cancer according to the first aspect, the construction method comprising:

[0019] Clinical pathological data of patients with early gastric cancer who underwent surgical treatment were collected as categorical variables, including gender, patient age at surgery, tumor location, maximum tumor diameter, histological type, pathological differentiation degree, Lauren type, pT stage, pN stage, number of metastatic lymph nodes, number of lymph nodes submitted for examination, nerve invasion, and lymphovascular invasion;

[0020] Analyze the correlation between categorical variables and lymph node metastasis in patients with early gastric cancer, and screen categorical variables related to lymph node metastasis in patients with early gastric cancer as a data set;

[0021] The obtained data set was used to train a machine learning algorithm to build a model for predicting the risk of lymph node metastasis in early gastric cancer.

[0022] Preferably, the data set includes gender, tumor location, maximum tumor diameter, degree of pathological differentiation, number of lymph nodes submitted for examination, and lymphatic vessel invasion.

[0023] Preferably, the method for analyzing the correlation between categorical variables and lymph node metastasis in patients with early gastric cancer comprises a chi-square test or a Fisher's exact test.

[0024] Preferably, the machine learning algorithm comprises a logistics regression algorithm.

[0025] In a third aspect, the present invention provides a system for predicting the risk of lymph node metastasis of early gastric cancer, the system comprising a data collection unit, a calculation unit and a determination unit;

[0026] The data collection unit is used to perform the following steps:

[0027] The gender, tumor location, maximum tumor diameter, pathological differentiation degree, number of lymph nodes sent for examination, and lymphatic vessel invasion of patients with early gastric cancer were collected;

[0028] The computing unit is used to perform the following steps:

[0029] Inputting the data collected by the data collecting unit into the model for predicting the risk of lymph node metastasis of early gastric cancer described in the first aspect to calculate the predicted probability of lymph node metastasis;

[0030] The determination unit is used to perform the following steps:

[0031] The prediction probability of lymph node metastasis is calculated by the calculation unit for determination.

[0032] In a fourth aspect, the present invention provides a model for predicting the number of metastatic lymph nodes in patients with early gastric cancer, wherein the calculation formula of the model is:

[0033] Predicted number of metastatic lymph nodes = (number of sufficient lymph nodes for examination - actual number of lymph nodes for examination) × predicted probability of lymph node metastasis + number of metastatic lymph nodes

[0034] Among them, the sufficient number of lymph nodes for examination is 20, and the actual number of lymph nodes and metastatic lymph nodes for examination come from the patient's postoperative pathological results, that is, the pathology indicates how many lymph nodes were removed and sent for examination during the operation (lymph nodes sent for examination), and how many of them are positive lymph nodes (metastatic lymph nodes); the predicted probability of lymph node metastasis is calculated by the model for predicting the risk of lymph node metastasis of early gastric cancer described in the first aspect.

[0035] The present invention can provide a personalized risk assessment tool by constructing a risk model for predicting lymph node metastasis in patients with early gastric cancer. Based on this model, a metastatic lymph node number prediction model is further designed, which helps clinical physicians to evaluate and calculate early gastric cancer patients who do not have enough lymph nodes for examination, and then predict and correct the number of potential metastatic lymph nodes to ultimately obtain an accurate pTNM staging. Accurate staging assessment can bring more accurate treatment plans to patients, including whether to use postoperative adjuvant chemotherapy, which will directly affect the patient's survival prognosis.

[0036] In a fifth aspect, the present invention provides a system for predicting the number of metastatic lymph nodes in patients with early gastric cancer, the system comprising a data collection unit and a calculation unit;

[0037] The data collection unit is used to perform the following steps:

[0038] The gender, tumor location, maximum tumor diameter, pathological differentiation degree, number of lymph nodes sent for examination, and lymphatic vessel invasion of patients with early gastric cancer were collected;

[0039] The computing unit is used to perform the following steps:

[0040] Inputting the data collected by the data collecting unit into the model for predicting the risk of lymph node metastasis of early gastric cancer described in the first aspect to calculate the predicted probability of lymph node metastasis;

[0041] The number of metastatic lymph nodes is predicted by calculating the number of metastatic lymph nodes in patients with early gastric cancer described in the fourth aspect.

[0042] In a sixth aspect, the present invention 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 processor executes the computer program, the system for predicting the risk of lymph node metastasis of early gastric cancer as described in the third aspect or the system for predicting the number of metastatic lymph nodes in patients with early gastric cancer as described in the fifth aspect is implemented.

[0043] In a seventh aspect, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the functions of the system for predicting the risk of lymph node metastasis of early gastric cancer as described in the third aspect or the system for predicting the number of metastatic lymph nodes in patients with early gastric cancer as described in the fifth aspect.

[0044] Compared with the prior art, the present invention has at least the following beneficial effects:

[0045] The present invention designs a model for predicting the risk of lymph node metastasis of early gastric cancer. The model can accurately assess the risk of lymph node metastasis in patients with early gastric cancer, and further designs a model for predicting the number of metastatic lymph nodes, which helps clinical physicians to evaluate and calculate early gastric cancer patients who do not have enough lymph nodes for examination, and then predict and correct the number of potential metastatic lymph nodes to ultimately obtain an accurate pTNM staging, which has auxiliary clinical diagnostic value. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 Schematic diagram of the analysis process for data of patients with early gastric cancer who underwent surgical treatment;

[0047] Figure 2 This is the AUC curve and area under the curve result graph of the model in the training set for patients with early gastric cancer;

[0048] Figure 3 Calibration curve results of the training set model for patients with early gastric cancer;

[0049] Figure 4 The DCA curve results of the training set model for patients with early gastric cancer;

[0050] Figure 5 This is the AUC curve and area under the curve result graph of the validation set model in patients with early gastric cancer;

[0051] Figure 6 Calibration curve results of the centralized model for early gastric cancer patients;

[0052] Figure 7 The DCA curve results of the validation centralized model for patients with early gastric cancer;

[0053] Figure 8 This is a graph showing the correlation analysis results between the number of lymph nodes examined and the number of metastatic lymph nodes in patients with early gastric cancer in a multicenter data set before and after model correction;

[0054] Fig. 9 This is a multi-center data set showing the correlation analysis results between the number of lymph nodes examined and the number of metastatic lymph nodes in patients with early gastric cancer who had less than or equal to 20 lymph nodes before and after model correction;

[0055] Fig.10 This is a graph showing the correlation analysis results between the number of lymph nodes examined and the number of metastatic lymph nodes in patients with pN+ stage early gastric cancer who had less than or equal to 20 lymph nodes examined in a multicenter data set before and after model correction;

[0056] Fig.11 This is a survival analysis result diagram of patients with early gastric cancer in a multicenter data set with different numbers of lymph nodes submitted for examination;

[0057] Fig.12 This is a survival analysis result diagram of patients with early gastric cancer at different pN stages in a multi-center data set;

[0058] Fig.13 Different pN after model correction for multicenter dataset M Survival analysis results of patients with early gastric cancer by stage;

[0059] Fig.14 This is the result of the correlation analysis between the number of lymph nodes sent for examination and the number of metastatic lymph nodes in patients with early gastric cancer in the SEER dataset before and after model correction;

[0060] Fig.15 This is the result of the correlation analysis between the number of lymph nodes submitted for examination and the number of metastatic lymph nodes in patients with early gastric cancer in the SEER data set with less than or equal to 20 lymph nodes before and after model correction;

[0061] Fig.16 This is the result of the correlation analysis between the number of lymph nodes submitted for examination and the number of metastatic lymph nodes in patients with pN+ early gastric cancer in the SEER data set with less than or equal to 20 lymph nodes before and after model correction;

[0062] Fig.17 This is a graph showing the survival analysis results of patients with early gastric cancer in different groups with different numbers of lymph nodes submitted for examination in the SEER dataset;

[0063] Fig.18 This is the survival analysis result of patients with early gastric cancer in different pN stages of the SEER dataset;

[0064] Fig.19 Different pN after SEER dataset model correction M Survival analysis results of patients with early gastric cancer by stage. DETAILED DESCRIPTION

[0065] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and through specific implementation methods. However, the following examples are only simplified examples of the present invention and do not represent or limit the scope of protection of the present invention. The scope of protection of the present invention shall be subject to the claims.

[0066] If no specific techniques or conditions are specified in the examples, the techniques or conditions described in the literature in the field or the product instructions are used. If no manufacturer is specified for the reagents or instruments used, they are all conventional products that can be purchased through regular channels.

[0067] Example 1

[0068] This example constructs a lymph node metastasis risk assessment model for early gastric cancer patients and a lymph node metastasis number prediction model for early gastric cancer patients.

[0069] A retrospective analysis was performed on the data of patients with early gastric cancer who were admitted to 15 large medical institutions in China, including Tianjin Medical University Cancer Hospital, from 2005 to 2015 and underwent surgical treatment (included in the flowchart as shown in Figure 2). Figure 1 The inclusion criteria were as follows: 1) patients who had undergone radical surgery for gastric cancer; 2) patients whose postoperative pathology confirmed that they had primary early gastric cancer; 3) patients who had no distant organ metastasis or peritoneal implantation (M0); 4) patients who had not received chemotherapy before surgery. The exclusion criteria were as follows: 1) patients with missing important clinical pathological data; 2) patients with other synchronous malignant tumors; 3) patients with incomplete important clinical pathological data; 4) patients with residual gastric cancer; 5) patients who had undergone endoscopic mucosal dissection or excisional biopsy. The clinical pathological data of the patients included gender, age of the patients at the time of surgery, tumor location, maximum diameter of the tumor, histological classification, degree of pathological differentiation, Lauren classification, pT stage, pN stage, number of metastatic lymph nodes, number of lymph nodes submitted for examination, presence of perineural invasion, and presence of lymphovascular invasion. The basic clinical pathological characteristics of 6566 patients with early gastric cancer in the multicenter data set are shown in Table 1.

[0070] Table 1

[0071]

[0072]

[0073] SPSS and R language software were used for statistical analysis. The number of lymph nodes submitted for examination that affected the occurrence of lymph node metastasis in patients with early gastric cancer was confirmed by the cut-off value, which was defined as sufficient lymph nodes submitted for examination. Finally, the prediction efficiency was best when the Youden index was the largest, that is, when 20 lymph nodes were taken as the cut-off value, so 20 lymph nodes were defined as the sufficient number of lymph nodes submitted for examination (Table 2). All patients in the included multicenter data set were grouped into training set and validation set according to the ratio of 7:3. The training set was used to build the model, and the validation set was used to verify the model efficacy (Table 3). All variables were categorical variables. The chi-square test or Fisher's exact test was used to analyze the correlation between potential related variables and lymph node metastasis in patients with early gastric cancer. Variables with P value (P value) <0.05 were included in the multivariate logistic regression analysis, and finally the independent risk factors for lymph node metastasis in patients with early gastric cancer were determined (Table 4).

[0074] Table 2

[0075]

[0076]

[0077] Table 3

[0078]

[0079]

[0080] Table 4

[0081]

[0082]

[0083] Note: Ref. indicates reference variable.

[0084] The logistics formula was constructed based on the β coefficients of each independent risk factor in the multivariate logistic regression analysis results, and the lymph node metastasis probability calculation model (P) for patients with early gastric cancer was constructed. The AUC curve, calibration curve, and DCA curve were calculated and plotted to evaluate the effectiveness of this model. The analysis results are shown in Figure 2. Figure 2-Figure 4 As shown in the figure, the model is evaluated by Roc curve, Calibration curve and DCA curve. The AUC value is 0.789, and the Calibration curve is close to the diagonal line, which proves that it has strong predictive ability. The DCA curve confirms that the model has certain net benefits.

[0085]

[0086] Wherein, P is the predicted probability of lymph node metastasis; Logit P = -0.358×a+b×c+0.660×d+0.6388×e+0.933×f+1.719×g-0.898×h-2.609;

[0087] a represents gender, male: a=0, female: a=1;

[0088] b×c represents the location of the tumor, the tumor is located in the lower 1 / 3 of the stomach: c=0; the tumor is located in the middle 1 / 3 of the stomach: b=-0.059, c=1; the tumor is located in the upper 1 / 3 of the stomach: b=-0.616, c=2; the tumor exceeds 1 / 3 of the stomach: b=-0.466, c=3;

[0089] d represents the maximum diameter of the tumor: d = 0 if the maximum diameter of the tumor is ≤ 2 cm; d = 1 if the maximum diameter of the tumor is > 2 cm;

[0090] e represents the degree of pathological differentiation, moderately differentiated or well differentiated: e = 0; poorly differentiated or undifferentiated: e = 1;

[0091] f represents pT stage, pT1a stage: f = 0; pT1b stage: f = 1;

[0092] g represents the status of lymphatic vessel invasion, no lymphatic vessel invasion: g = 0; lymphatic vessel invasion: g = 1;

[0093] h represents the number of lymph nodes examined (ELN), ELN value ≤ 20: h = 0; ELN > 20: h = 1.

[0094] The metastasis probability of lymph nodes was substituted into the formula for early gastric cancer patients with insufficient lymph nodes for examination, and a prediction model for the number of metastatic lymph nodes in early gastric cancer patients with insufficient lymph nodes for examination was constructed.

[0095] Predicted number of metastatic lymph nodes = (number of sufficient lymph nodes for examination - actual number of lymph nodes for examination) × predicted probability of lymph node metastasis + number of metastatic lymph nodes

[0096] Among them, the sufficient number of lymph nodes for examination is 20. The actual number of lymph nodes and metastatic lymph nodes for examination comes from the patient's postoperative pathological results, that is, the pathology indicates how many lymph nodes were removed and sent for examination during the operation (lymph nodes sent for examination), and how many of them are positive lymph nodes (metastatic lymph nodes).

[0097] Example 2

[0098] This example performs a verification test on the model constructed in Example 1.

[0099] Specifically, the prediction model for lymph node metastasis of patients with early gastric cancer based on the training set in Example 1 was constructed, and internal verification was also performed in the validation set. The results are as follows: Figure 5-Figure 7 As shown, the AUC value is 0.775, and the Calibration curve is relatively close to the diagonal line, which proves that it has strong predictive ability. The DCA curve confirms that the model has a certain net benefit, indicating that the model of the present invention also has a high predictive efficiency in the validation set.

[0100] For the prediction model of the number of metastatic lymph nodes, 2094 multicenter early gastric cancer patients with complete survival data were included as the test set, and the lymph node metastasis prediction formula was substituted and calculated (Table 5). It was found that before the model was corrected, the number of lymph nodes submitted for examination was positively correlated with the number of metastatic lymph nodes. This trend existed in early gastric cancer patients, early gastric cancer patients with less than or equal to 20 lymph nodes submitted for examination, and pN+ stage early gastric cancer patients with less than or equal to 20 lymph nodes submitted for examination. After the model of the present invention was corrected, the correlation between the two was close to no correlation, which proves that after the formula proposed by the present invention is substituted, the number of metastatic lymph nodes in early gastric cancer patients tends to be fixed and does not increase with the increase in the number of lymph nodes submitted for examination. This further confirms that after the model is corrected, all potential metastatic lymph nodes can be predicted and discovered ( Figure 8-Figure 10 ). In the survival analysis, the present invention grouped the number of lymph nodes examined into 20 and found that the early gastric cancer patients with less than or equal to 20 lymph nodes had a worse survival rate ( Fig.11 Based on the AJCC guidelines and the number of metastatic lymph nodes calculated by the model of the present invention, a new pN staging was constructed, which we defined as pN M Staging, in different pN and pN M The comparison of the stages revealed that pN M Staging can better differentiate the survival of patients and help better assess the prognosis of patients ( Figure 12-13 ).

[0101] In addition, in order to further verify the predictive effectiveness of the model, we also selected the SEER database (a public database and research resource of the National Cancer Institute of the United States) as the test set, screened and finally included 1944 early gastric cancer patients for analysis and research (Table 6). The metastatic lymph node prediction model of the present invention was also substituted to obtain the final predicted number of metastatic lymph nodes based on the clinical pathological data. We also conducted correlation and survival analysis studies to confirm its clinical practicality. In the correlation analysis between the number of lymph nodes submitted for examination and the number of metastatic lymph nodes, we found that the results obtained in the SEER database and the multicenter data set were very similar. In the three subgroups of early gastric cancer patients, early gastric cancer patients with less than or equal to 20 lymph nodes submitted for examination, and pN+ stage early gastric cancer patients with less than or equal to 20 lymph nodes submitted for examination, the two were positively correlated before model correction and were almost unrelated after correction. This shows the comprehensiveness of the model of the present invention in the discovery of potential metastatic lymph nodes ( Figure 14-16 ). In the survival analysis, it was also found that patients with early gastric cancer who had less than or equal to 20 lymph nodes had a worse survival rate ( Fig.17 ). For different pN and pN M The comparison of the stages revealed that pN M Staging can also better differentiate the survival of patients and help better assess the prognosis of patients ( Figure 18-19 ).

[0102] Finally, the Akaike information criterion (AIC) and the Bayesian information criterion (BIC) were used to classify the pN stage and the pN corrected based on the model of the present invention. M The predictive efficacy of these two lymph node staging methods was evaluated in both the multicenter test set and the SEER test set. M The staging showed lower AIC and BIC values, which proved its better predictive efficacy (Table 7). Therefore, we have shown through various verifications that the model of the present invention is not only highly feasible in the multi-center database of my country, but also has a strong correction efficacy for the early gastric cancer population in Europe and the United States after testing with the SEER database.

[0103] In general, for EGC patients who cannot undergo adequate lymph node dissection for various reasons, the model of the present invention helps clinicians evaluate and calculate EGC patients who do not have sufficient ELNs, and then predict and correct the number of potential metastatic lymph nodes to finally obtain accurate pTNM staging. Accurate staging assessment brings more accurate treatment plans for patients, including whether to use postoperative adjuvant chemotherapy, which will directly affect the patient's survival prognosis.

[0104] Table 5

[0105]

[0106]

[0107] Table 6

[0108]

[0109]

[0110] Table 7

[0111]

[0112] Note: AIC: Akaike information criterion; BIC: Bayesian information criterion.

[0113] In summary, the present invention provides an individualized risk assessment tool by constructing a lymph node metastasis risk assessment model for early gastric cancer patients. Based on this assessment model, a model for predicting the number of lymph node metastases is further obtained. For early gastric cancer patients who cannot undergo adequate lymph node dissection due to various reasons, it is helpful for clinical physicians to evaluate and calculate early gastric cancer patients who do not have sufficient lymph nodes for examination, and then predict and correct the number of potential metastatic lymph nodes to finally obtain an accurate pTNM staging. Accurate staging assessment can bring more accurate treatment plans to patients, including whether to use postoperative adjuvant chemotherapy, which will directly affect the patient's survival prognosis and provide a new approach for accurate clinical evaluation.

[0114] The applicant declares that the above is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Those skilled in the art should understand that any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention shall fall within the protection scope and disclosure scope of the present invention.

Claims

1. A model for predicting the risk of lymph node metastasis in early gastric cancer, characterized in that: The input variables of the model include gender, tumor location, maximum tumor diameter, pathological differentiation degree, pT stage, number of lymph nodes submitted for examination and lymphatic vessel invasion. The output value of the model is the predicted probability of lymph node metastasis, and the calculation formula is shown in formula (1); Wherein, P is the predicted probability of lymph node metastasis; Logit P = -0.358×a+b×c+0.660×d+0.6388×e+0.933×f+1.719×g-0.898×h-2.609; a represents gender, male: a=0, female: a=1; b×c represents the location of the tumor, the tumor is located in the lower 1 / 3 of the stomach: c=0; the tumor is located in the middle 1 / 3 of the stomach: b=-0.059, c=1; the tumor is located in the upper 1 / 3 of the stomach: b=-0.616, c=2; the tumor exceeds 1 / 3 of the stomach: b=-0.466, c=3; d represents the maximum diameter of the tumor: d = 0 if the maximum diameter of the tumor is ≤ 2 cm; d = 1 if the maximum diameter of the tumor is > 2 cm; e represents the degree of pathological differentiation, moderately differentiated or well differentiated: e = 0; poorly differentiated or undifferentiated: e = 1; f represents pT stage, pT1a stage: f = 0; pT1b stage: f = 1; g represents the status of lymphatic vessel invasion, no lymphatic vessel invasion: g = 0; lymphatic vessel invasion: g = 1; h represents the number of lymph nodes submitted for examination. If the number of lymph nodes submitted for examination is ≤20, h=0; if the number of lymph nodes submitted for examination is >20, h=1.

2. The method for constructing a model for predicting the risk of lymph node metastasis of early gastric cancer according to claim 1, characterized in that: The construction method comprises: Clinical pathological data of patients with early gastric cancer who underwent surgical treatment were collected as categorical variables, including gender, patient age at surgery, tumor location, maximum tumor diameter, histological type, pathological differentiation degree, Lauren type, pT stage, pN stage, number of metastatic lymph nodes, number of lymph nodes submitted for examination, nerve invasion, and lymphovascular invasion; Analyze the correlation between categorical variables and lymph node metastasis in patients with early gastric cancer, and screen categorical variables related to lymph node metastasis in patients with early gastric cancer as a data set; The obtained data set was used to train a machine learning algorithm to build a model for predicting the risk of lymph node metastasis in early gastric cancer.

3. The method for constructing a model for predicting the risk of lymph node metastasis of early gastric cancer according to claim 2, characterized in that: The data set includes gender, tumor location, maximum tumor diameter, pathological differentiation degree, number of lymph nodes submitted for examination, and lymphatic vessel invasion.

4. The method for constructing a model for predicting the risk of lymph node metastasis of early gastric cancer according to claim 2 or 3, characterized in that: The method for analyzing the correlation between categorical variables and lymph node metastasis in patients with early gastric cancer includes chi-square test or Fisher's exact test.

5. The method for constructing a model for predicting the risk of lymph node metastasis of early gastric cancer according to any one of claims 2 to 4, characterized in that: The machine learning algorithm includes a logistics regression algorithm.

6. A system for predicting the risk of lymph node metastasis in early gastric cancer, characterized in that: The system comprises a data collection unit, a calculation unit and a determination unit; The data collection unit is used to perform the following steps: The gender, tumor location, maximum tumor diameter, pathological differentiation degree, number of lymph nodes sent for examination, and lymphatic vessel invasion of patients with early gastric cancer were collected; The computing unit is used to perform the following steps: Inputting the data collected by the data collecting unit into the model for predicting the risk of lymph node metastasis of early gastric cancer as claimed in claim 1 to calculate the predicted probability of lymph node metastasis; The determination unit is used to perform the following steps: The prediction probability of lymph node metastasis is calculated by the calculation unit for determination.

7. A model for predicting the number of metastatic lymph nodes in patients with early gastric cancer, characterized in that: The calculation formula of the model is: Predicted number of metastatic lymph nodes = (number of sufficient lymph nodes for examination - actual number of lymph nodes for examination) × predicted probability of lymph node metastasis + number of metastatic lymph nodes The predicted probability of lymph node metastasis described in the formula is calculated by the model for predicting the risk of lymph node metastasis of early gastric cancer described in claim 1.

8. A system for predicting the number of metastatic lymph nodes in patients with early gastric cancer, characterized in that: The system comprises a data collection unit and a calculation unit; The data collection unit is used to perform the following steps: The gender, tumor location, maximum tumor diameter, pathological differentiation degree, number of lymph nodes sent for examination, and lymphatic vessel invasion of patients with early gastric cancer were collected; The computing unit is used to perform the following steps: Inputting the data collected by the data collecting unit into the model for predicting the risk of lymph node metastasis of early gastric cancer as claimed in claim 1 to calculate the predicted probability of lymph node metastasis; The number of metastatic lymph nodes is predicted by using the model for predicting the number of metastatic lymph nodes in patients with early gastric cancer as described in claim 7.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the function of the system for predicting the risk of lymph node metastasis of early gastric cancer as claimed in claim 6 or the system for predicting the number of metastatic lymph nodes in patients with early gastric cancer as claimed in claim 8 is realized.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, it realizes the functions of the system for predicting the risk of lymph node metastasis of early gastric cancer as claimed in claim 6 or the system for predicting the number of metastatic lymph nodes in patients with early gastric cancer as claimed in claim 8.

Images