A survival prediction model for predicting postoperative overall survival of pancreatic cancer patients based on metabolic score, pathological information and adjuvant chemotherapy

Abstract

The application provides a survival prediction model for predicting the overall survival of a pancreatic cancer patient after surgery based on metabolic scores, pathological information and adjuvant chemotherapy, and belongs to the technical field of prediction models. The system is simple to operate, has good discrimination, calibration ability and clinical net benefit, and can be used as an important tool for personalized prediction of the prognosis of pancreatic cancer patients. The survival prediction model considers the nonlinear influence of metabolic syndrome components on pancreatic cancer, uses RCS curve to calculate the metabolic score of metabolic syndrome components affecting the death risk of pancreatic cancer patients, more clearly observes the estimated association between metabolic syndrome components and pancreatic cancer mortality results, and determines any potential threshold effect. The survival prediction model considers the adjuvant radiotherapy and chemotherapy factors of the pancreatic cancer adjuvant therapy decision model, so that the system prediction performance is better, and has a good application prospect in the preparation of a pancreatic cancer prognosis prediction device.

Description

Technical Field

[0001] This invention belongs to the field of predictive model technology, specifically relating to a survival prediction model based on metabolic integral, pathological information and adjuvant chemotherapy to predict the overall survival of pancreatic cancer patients after surgery. Background Technology

[0002] Pancreatic cancer is considered one of the deadliest cancers due to its malignancy, aggressiveness, and low survival rate; its mortality rate is almost equal to its incidence rate. According to GLOBOCAN 2020 data, pancreatic cancer is the 12th most common malignant tumor and the 7th leading cause of cancer death.

[0003] Because cancer is often localized, lacking symptoms or presenting with vague symptoms, patients typically present with advanced-stage disease. Although the 5-year survival rate for patients undergoing surgical resection is 10-25%, surgery remains the only chance to cure the tumor. Advances in adjuvant chemotherapy have improved the long-term prognosis for pancreatic cancer patients. Identifying readily available, prognostic clinical indicators is crucial for improving the survival prognosis and prolonging the lifespan of pancreatic cancer patients.

[0004] 5-10% of pancreatic cancers can be attributed to genetic risk factors, and several familial cancer syndromes associated with an increased risk of developing pancreatic cancer have been identified in current research. Besides genetic factors, a patient's own condition, environment, and diet have a significant impact on the development and progression of pancreatic cancer. Obesity, type 2 diabetes, smoking, alcohol consumption, and pancreatitis are recognized risk factors for pancreatic cancer. In addition, other risk factors include high-fat / high-protein diets and psychological distress.

[0005] Metabolic syndrome (MetS), encompassing factors such as hypertension, hyperglycemia, excess abdominal fat, and abnormal cholesterol or triglyceride levels, is a series of comorbidities that increase the risk of cardiovascular disease (CVD), stroke, and type 2 diabetes (T2D). Some of these metabolic disorders have been proven to be associated with the development of pancreatic cancer. Pancreatic cancer has been reported to be linked to smoking, type 2 diabetes (T2D), and chronic pancreatitis. These factors contributing to the burden of pancreatic cancer may be caused by or induce metabolic syndrome. Because metabolic syndrome is reversible, lifestyle modifications or medical interventions targeting patients with metabolic syndrome may be potential preventative strategies for gastrointestinal cancers, potentially improving the precision and effectiveness of pancreatic interventions. Currently, research on the relationship between metabolic markers and the prognosis of gastrointestinal tumors is relatively limited. There are currently no studies on the personalized prediction of pancreatic cancer patient prognosis based on preoperative metabolic syndrome and its components combined with comprehensive factors such as patient clinical information, pathological indicators, and treatment methods.

[0006] A study by Yohei Miyashita et al. confirmed the link between metabolic syndrome and pancreatic cancer, demonstrating in a retrospective observational study that even early-stage metabolic syndrome is associated with pancreatic cancer. This study recruited approximately 4.6 million Japanese individuals in 2005 and followed these participants for over 10 years. At enrollment, after obtaining clinical data on pre-prescription medications and checking for the presence or absence of metabolic syndrome (MetS), researchers followed up these participants with or without MetS to assess the incidence of pancreatic cancer. The results showed that pre-MetS was closely associated with the incidence of pancreatic cancer according to the Japanese criteria for metabolic syndrome. (Miyashita Y, Hitsumoto T, Fukuda H, Kim J, Ito S, Kimoto N, Asakura K, Yata Y, Yabumoto M, Washio T, Kitakaze M. Metabolic syndrome is linked to the incidence of pancreatic cancer. EClinicalMedicine. 2023; 67:102353.)

[0007] Raviv NV et al. evaluated health outcomes in hospitalized pancreatic cancer patients. Their study, analyzing data from 47,386 patients hospitalized with a preliminary diagnosis of pancreatic cancer, found that patients with pancreatic cancer and MetS were more likely to undergo pancreatectomy (OR 1.14, 95% CI: 1.04–1.25) compared to those without MetS. Furthermore, compared to patients without MetS, this group had a lower incidence of postoperative complications (OR: 0.90, 95% CI: 0.81–0.99), a lower probability of discharge to a specialized care facility (OR: 0.90, 95% CI: 0.83–0.93), and a lower in-hospital mortality rate (OR: 0.52, 95% CI: 0.44–0.61). Hospitalized pancreatic cancer patients clinically diagnosed with MetS were more likely to undergo pancreatectomy and had lower rates of postoperative complications and in-hospital mortality. (Raviv NV, Sakhuja S, Schlachter M, Akinyemiju T. Metabolic syndrome and in-hospital outcomes among pancreatic cancer patients. DiabetesMetab Syndr. 2017Dec; 11Suppl 2:S643-S650.)

[0008] Current research on the relationship between metabolic syndrome and pancreatic cancer primarily focuses on the correlation between the incidence of metabolic syndrome and pancreatic cancer, or the in-hospital mortality and complication rates in patients with pancreatic cancer and metabolic syndrome. It has not addressed the correlation between metabolic syndrome and its components and the prognosis (overall survival) of pancreatic cancer patients. Furthermore, it has not developed personalized prognostic predictions for poor pancreatic cancer outcomes based on multiple patient indicators. The diagnostic criteria for metabolic syndrome are based on the MetS definition proposed by the Chinese Diabetes Society in 2004, which is more in line with the physical characteristics of Chinese patients. This definition involves multiple continuous variables, which may be non-linear influencing factors. Defining these factors as continuous variables in the diagnostic definition may lead to misinterpretations. For example, body mass index (BMI) is actually a non-linear influencing factor; both excessively high and low BMI are associated with poor prognosis. Treatment methods other than surgery for pancreatic cancer patients can directly affect patient prognosis, but other studies have shown that surgical patients who undergo combined adjuvant chemotherapy have higher survival rates. There is no research on the combination of metabolic syndrome and treatment methods used during patient treatment. Summary of the Invention

[0009] In order to overcome the problems existing in the prior art, the purpose of this invention is to provide a survival prediction model based on metabolic score, pathological information and adjuvant chemotherapy to predict the overall survival of pancreatic cancer patients after surgery.

[0010] This invention provides a survival prediction model for predicting the overall survival after surgery in patients with pancreatic cancer. The survival prediction model includes the following modules:

[0011] I. Data Input Module

[0012] The input is used to determine the patient's characteristic data, which includes age, family history, N stage, metastatic site, histological grade, chemotherapy, and metabolic score; the patient is a pancreatic cancer patient.

[0013] II. Model Building Module

[0014] A nomogram predicting overall survival for pancreatic cancer patients was constructed using feature data from the input module.

[0015] III. Prediction Module

[0016] Input the feature data of the patient to be predicted into the nomogram constructed in step two, and output the prediction results.

[0017] Furthermore, the pancreatic cancer patients referred to are those who underwent pancreatic cancer resection surgery at stage I-IV.

[0018] Furthermore, the metabolic integral was calculated as follows: The influence of metabolic syndrome components on the prognosis of pancreatic cancer patients was analyzed using restricted cubic plots to identify nonlinear and linear influencing factors. The metabolic integral value was then calculated using the following R language formula: res.cox<-coxph(Surv(time,status)~TG+HDL+GLU+rcs(BMI,4)+SBP+DBP,data=

[0019] aa);

[0020] The metabolic syndrome components include body mass index, fasting blood glucose, blood pressure, triglycerides, and high-density lipoprotein, and the blood pressure includes systolic and diastolic blood pressure;

[0021] In the R language calculation formula, BMI is body mass index, GLU is fasting blood glucose, SBP is systolic blood pressure, DBP is diastolic blood pressure, TG is triglycerides, and HDL is high-density lipoprotein.

[0022] Furthermore, among the components of metabolic syndrome, BMI is a non-linear influencing factor, while GLU, SBP, DBP, TG, and HDL are linear influencing factors.

[0023] Furthermore, the age is a continuous variable, and the value of the age variable is the patient's age in years;

[0024] The family history is a discrete variable, represented by the value of the family history variable: 0: no family history of cancer, 1: family history of cancer.

[0025] The N stage is a discrete variable, and the value of the N stage variable represents the lymph node metastasis in the tumor area of ​​the patient, 0:N0, 1:N1, 2:N2, 3:N3;

[0026] The metastasis site is a discrete variable, represented by the value of the metastasis site variable: 0: no metastasis, 1: multiple metastases, brain metastasis, 3: bone metastasis, 4: abdominal metastasis, 5: lung metastasis, 6: liver metastasis, 7: metastasis site unknown;

[0027] The histological grading is a discrete variable, represented by the value of the histological grading variable: 1: low differentiation, 2: moderate differentiation, 3: high differentiation;

[0028] The chemotherapy is a discrete variable, represented by the value of the chemotherapy variable: 0: no chemotherapy received, 2: chemotherapy received.

[0029] The metabolic integral is a continuous variable.

[0030] Furthermore, in the model construction module, the method of constructing a nomogram predicting the overall survival of pancreatic cancer patients using the feature data of the input module is to construct and draw the nomogram by performing multivariate Cox regression analysis on the feature data of the input module.

[0031] Furthermore, the total survival is 1 year, 3 years, and 5 years.

[0032] The present invention also provides the use of the above-described survival prediction model in the preparation of a device for predicting the prognosis of pancreatic cancer.

[0033] The present invention also provides a computer-readable storage medium having the above-described survival prediction model stored thereon.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] This invention provides a survival prediction model for pancreatic cancer patients based on a combination of metabolic integral, pathological information, and adjuvant chemotherapy, belonging to the field of predictive model technology. The system is simple to operate, possesses excellent discrimination and calibration capabilities, and offers significant clinical benefits, making it an important tool for personalized prognosis prediction in pancreatic cancer patients. The survival prediction model considers the nonlinear influence of metabolic syndrome components on pancreatic cancer, using RCS curves to calculate the metabolic integral of metabolic syndrome components affecting the risk of death in pancreatic cancer patients. This allows for a clearer observation of the estimated association between metabolic syndrome components and pancreatic cancer mortality outcomes, and identifies any potential threshold effects. Furthermore, this survival prediction model incorporates adjuvant chemoradiotherapy factors into the pancreatic cancer adjuvant therapy decision-making model, further enhancing the system's predictive efficacy and demonstrating promising application prospects in the development of devices for predicting pancreatic cancer prognosis.

[0036] Obviously, based on the above description of the present invention, and according to common technical knowledge and conventional methods in the field, various other modifications, substitutions or alterations can be made without departing from the basic technical concept of the present invention.

[0037] The following detailed embodiments further illustrate the above-described content of the present invention. However, this should not be construed as limiting the scope of the present invention to the following embodiments. All technologies implemented based on the above-described content of the present invention fall within the scope of the present invention. Attached Figure Description

[0038] Figure 1 RCS analysis of metabolic syndrome components and overall survival in pancreatic cancer.

[0039] Figure 2 To screen for prognostic risk factors for pancreatic cancer using Lasso regression analysis.

[0040] Figure 3Nodal plot of the overall survival prediction model for pancreatic cancer constructed by combining metabolic score with other indicators.

[0041] Figure 4 To assess the predictive performance of a pancreatic cancer overall survival prediction model constructed using metabolic integrals in conjunction with other indicators in the (A) internal and (B) external validation cohorts. Detailed Implementation

[0042] The raw materials and equipment used in this invention are all known products, obtained by purchasing commercially available products.

[0043] This invention included 2826 patients with stage I-IV pancreatic cancer who underwent resection surgery. Postoperative survival was followed up, and overall survival (OS) information was collected. The 2826 patients were randomly sampled at a ratio of 7:3 for internal and external validation.

[0044] The following are the steps for screening independent factors that predict overall survival (OS) in patients with pancreatic cancer:

[0045] 1. Calculate the metabolic score of metabolic syndrome components affecting the risk of death in patients with gastrointestinal tumors.

[0046] The specific procedure involves analyzing the impact of metabolic syndrome components (body mass index, fasting blood glucose, blood pressure, triglycerides, and high-density lipoprotein) on the prognosis of pancreatic cancer patients using restricted cube plots (RCS) to identify nonlinear and linear influencing factors. Among the metabolic syndrome components, body mass index (BMI), fasting blood glucose (GLU), blood pressure (including systolic blood pressure (SBP) and diastolic blood pressure (DBP), triglycerides (TG), and high-density lipoprotein (HDL) are used. The RCS model is adjusted for age, sex, race, marital status, occupation, alcohol consumption, smoking, TNM stage, family history, tumor size, and histological grade.

[0047] like Figure 1 As shown, there is a potential non-linear association between BMI and overall survival (OS) in pancreatic cancer (p<0.001), and a J-type association exists. The optimal predictive range for BMI is 18.5–29.9 kg / m². 2 Other components of MetS showed a linear association with overall survival (OS) in pancreatic cancer patients.

[0048] Risk scores for metabolic syndrome components (body mass index, fasting blood glucose, blood pressure, triglycerides, and high-density lipoprotein) were calculated, and nonlinear factors were processed by RCS to obtain metabolic integral values ​​based on metabolic syndrome components.

[0049] The metabolic integral value is calculated using the following R language formula:

[0050] The R language formula for calculating the metabolic integral is: res.cox <- coxph(Surv(time,status) ~ TG + HDL + GLU + rcs(BMI,4) + SBP + DBP, data = aa).

[0051] The meanings of the codes involved in the above R language calculation formulas are well known to those skilled in the art.

[0052] 2. Lasso regression analysis was performed using patient demographic indicators, clinical information, pathological characteristics, and tumor-related treatments received, along with metabolic scores, to preliminarily screen for risk factors for pancreatic cancer patients' prognosis.

[0053] The specific procedure involves using Lasso regression analysis to preliminarily screen for prognostic risk factors in pancreatic cancer patients, including metabolic score, patient sociodemographic information (including age, sex, education, etc.), medical history, lifestyle exposures (including smoking habits, dietary intake, and alcohol consumption), anthropometric measurements (including weight, height, systolic blood pressure, and diastolic blood pressure), pathological features (TNM stage (I, II, III, IV), tumor size, depth of invasion (T1, T2, T3, T4), regional lymph node metastasis (N0, N1, N2, N3), distant metastasis (M0, M1)), and whether the patient has received cancer-related treatment (chemotherapy, immunotherapy, radiotherapy). Figure 2 As shown, the Lasso model uses factors such as age, sex, TNM stage, family history, tumor size, histological grade, metastatic site, chemotherapy, immunotherapy, radiotherapy, race, marital status, occupation, alcohol consumption, smoking, and metabolic score for calculation.

[0054] 3. Univariate and multivariate Cox proportional hazards regression analysis was used to further screen the risk factors identified by Lasso regression, and to select risk factors closely related to the prognosis of pancreatic patients.

[0055] Risk factors identified by Lasso regression were further screened using univariate and multivariate Cox proportional hazards regression analysis to identify risk factors closely related to the prognosis of pancreatic cancer patients. Cox proportional hazards regression models were used to analyze the impact of metabolic syndrome components (BMI, fasting blood glucose, blood pressure, triglycerides, and high-density lipoprotein) on the prognosis of gastrointestinal tumors. Multivariate Cox survival models were adjusted for age, family history, N stage, metastatic site, histological grade, chemotherapy, and metabolic score. Multivariate Cox regression analysis showed that age, family history, N stage, metastatic site, histological grade, chemotherapy, and metabolic score were independent predictors of overall survival (OS) in pancreatic cancer.

[0056] Among them, age is a continuous variable, and the value of the age variable is the patient's age (in years);

[0057] Family history is a discrete variable, represented by the value of the family history variable: 0: no family history of cancer, 1: family history of cancer.

[0058] N stage is a discrete variable. The value of the N stage variable represents the lymph node metastasis in the tumor area of ​​the patient. 0:N0, 1:N1, 2:N2, 3:N3;

[0059] The metastatic site is a discrete variable, represented by the value of the metastatic site variable: 0: no metastasis, 1: multiple metastases, brain metastasis, 3: bone metastasis, 4: abdominal metastasis, 5: lung metastasis, 6: liver metastasis, 7: metastatic site unknown;

[0060] Histological grading is a discrete variable, represented by the value of the histological grading variable: 1: low differentiation, 2: moderate differentiation, 3: high differentiation;

[0061] Chemotherapy is a discrete variable, represented by the value of the chemotherapy variable: 0: no chemotherapy received, 2: chemotherapy received.

[0062] The metabolic score is a continuous variable, and its value is the patient's metabolic score calculated in step 1.

[0063] The following is an example of a survival prediction model for pancreatic cancer patients, constructed using seven independent prognostic factors: age, family history, N stage, metastatic site, histological grade, chemotherapy, and metabolic score.

[0064] Example 1: Construction Method of a Survival Prediction Model for Pancreatic Cancer Prognosis Based on Metabolic Indices, Pathological Information, and Combined Adjuvant Chemotherapy (referred to as the base+risk system)

[0065] I. Input Module

[0066] Collect the patient's age, family history, N stage, metastatic site, histological grade, chemotherapy, and metabolic score, and enter these indicators into the input module.

[0067] II. Establishment of a prognostic prediction model for pancreatic cancer

[0068] Using seven indicators—patient age, family history, N stage, metastatic site, histological grade, chemotherapy, and metabolic score—a multivariate Cox regression analysis was performed to construct and plot a nomogram to predict the 1-, 3-, and 5-year overall survival prognosis of pancreatic cancer patients.

[0069] Kaplan-Meier (KM) analysis was performed to determine survival outcomes. KM curves were plotted using the median as the critical threshold, and the log-rank test was used to assess their statistical significance.

[0070] III. Predicting Patient Prognosis Using Pancreatic Cancer Prognostic Prediction Models

[0071] Data on age, family history, N stage, metastatic sites, histological grade, chemotherapy, and metabolic scores were collected from patients in the internal and external validation cohorts. The nomogram model constructed in step two was used to predict the 1-, 3-, and 5-year overall survival prognosis of these patients. Figure 3 As shown.

[0072] The following is the method for constructing the control model.

[0073] Compare with Example 1: Methods for building a base system

[0074] Referring to the method in Example 1, the only difference is that the input indicators are modified to include the patient's age, family history, N stage, metastatic site, histological grade, and chemotherapy, to construct a base system for predicting the prognosis of pancreatic cancer.

[0075] Compare with Example 2: Construction method of base+MeTs system

[0076] Referring to the method in Example 1, the only difference is that the input indicators are modified to include the patient's age, family history, N stage, metastatic site, histological grade, chemotherapy, and presence of metabolic syndrome, to construct a base+MeTs system for predicting the prognosis of pancreatic cancer.

[0077] Compare with Example 3: The construction method of the stage system

[0078] Referring to the method in Example 1, the only difference is that the input indicators are modified to the patient's T stage, N stage, and metastatic site to construct a pathological staging system for predicting the prognosis of pancreatic cancer, also known as the stage system.

[0079] The following experimental examples demonstrate the beneficial effects of the survival prediction model for pancreatic cancer prognosis based on the combination of metabolic scores, pathological information, and adjuvant chemotherapy.

[0080] Experiment Example 1: Model Validation

[0081] 1. To verify the accuracy of the survival prediction model for pancreatic cancer prognosis based on the combined use of metabolic integral, pathological information, and adjuvant chemotherapy in Example 1 of this invention, a bootstrap calibration curve was used, and the results were quantified as a C-index to predict the outcome probabilities of the internal and external validation cohorts. The results are as follows: Figure 4 As shown, the actual probabilities of OS for 1 year, 3 years, and 5 years are in good agreement with the predicted probabilities in the nomogram in both the internal and external validation queues.

[0082] 2. Analysis of C-index, NRI, and IDI

[0083] The Harrells C-Index (C-index), also known as the consistency index, is mainly used to calculate the discriminant between the Cox model predictions and the actual values ​​in survival analysis. It is an indicator used in survival analysis to evaluate the accuracy of the prediction model.

[0084] Net Reclassification Index (NRI) analysis method: If NRI>0, it is a positive improvement, indicating that the new model has improved the predictive ability of the old model; if NRI<0, it is a negative improvement, indicating that the predictive ability of the new model has decreased; if NRI=0, it is considered that the new model has not improved.

[0085] Integrated discrimination improvement (IDI) analysis method: The larger the IDI, the better the new model predicts compared to the old model.

[0086] Table 1. Comparison of predictive capabilities between metabolic score combined with other indicator systems and single indicator systems.

[0087]

[0088] Comparative analysis of different models in the embodiments and control examples using Harrell's C-statistic, NRI, and IDI (Table 1) confirmed that, compared with the stage system, base system, and base+MeTs system, the system constructed in Embodiment 1 of this invention (base+risk system) is significantly superior in predicting pancreatic cancer prognosis and assessing adjuvant therapy selection. The sensitivity of both results is higher than that of the stage system, base system, and base+MeTs system, and the specificity is higher than that of the stage system (assessment of pancreatic cancer prognosis and adjuvant therapy selection), significantly improving the prognostic predictive ability for pancreatic cancer patients.

[0089] Experimental Example 2: Clinical Benefit Evaluation

[0090] The model was validated using Decision Curve Analysis (DCA) and Calibration Curve, and ROC curves and C-index were calculated for the prediction model.

[0091] DCA (Discretionary Clinical Assessment) is a method for evaluating the net clinical benefit of a predictive model. DCA reflects a positive net benefit with a broad range of clinically reasonable risk threshold probabilities. To evaluate the clinical application of the metabolic score, decision curve analysis was employed. Compared to the Base system, the Base+MeTs system, or the stage system, the system constructed in Example 1 of this invention showed consistently positive results and greater net benefits across a broad range of risk thresholds. Based on the DCA results, a clinical impact curve was further plotted to assess the model's clinical application value. The clinical impact curve represents the acceptable potential clinical effect of the predicted metabolic score.

[0092] ROC analysis showed that, compared with other systems (Base system, Base+MeTs system, or stage system), the system constructed in Example 1 of this invention performed better in predicting overall survival in pancreatic cancer (Table 1). AUC over time showed that the system constructed in Example 1 of this invention outperformed the control system at every time point (Table 1).

[0093] In summary, this invention provides a survival prediction model for pancreatic cancer patients based on a combination of metabolic integral, pathological information, and adjuvant chemotherapy to predict postoperative overall survival. The system is simple to operate, possesses excellent discrimination and calibration capabilities, and demonstrates significant clinical benefit, making it an important tool for personalized prognosis prediction in pancreatic cancer patients. This survival prediction model considers the nonlinear influence of metabolic syndrome components on pancreatic cancer, using RCS curves to calculate the metabolic integral of metabolic syndrome components affecting the risk of death in pancreatic cancer patients. This allows for a clearer observation of the estimated association between metabolic syndrome components and pancreatic cancer mortality outcomes and identifies any potential threshold effects. Furthermore, this survival prediction model incorporates adjuvant chemoradiotherapy factors into its pancreatic cancer adjuvant therapy decision-making model, resulting in improved predictive efficacy and promising application prospects in the development of devices for predicting pancreatic cancer prognosis.

Claims

1. A survival prediction model for predicting overall postoperative survival in pancreatic cancer patients, characterized in that, The survival prediction model includes the following modules: I. Data Input Module The input is used to determine the patient's characteristic data, which includes age, family history, N stage, metastatic site, histological grade, chemotherapy, and metabolic score; the patient is a pancreatic cancer patient. II. Model Building Module A nomogram predicting overall survival for pancreatic cancer patients was constructed using feature data from the input module. III. Prediction Module Input the feature data of the patient to be predicted into the nomogram constructed in step two, and output the prediction results.

2. The survival prediction model according to claim 1, characterized in that, The pancreatic cancer patients referred to are those who underwent pancreatic cancer resection surgery in stages I-IV.

3. The survival prediction model according to claim 1, characterized in that, The metabolic integral was calculated as follows: the influence of metabolic syndrome components on the prognosis of pancreatic cancer patients was analyzed using restricted cubic plots to identify nonlinear and linear influencing factors, and the metabolic integral value was calculated according to the following R language calculation formula: res.cox<-coxph(Surv(time,status)~TG+HDL+GLU+rcs(BMI,4)+SBP+DBP,data=aa); The metabolic syndrome components include body mass index, fasting blood glucose, blood pressure, triglycerides, and high-density lipoprotein, and the blood pressure includes systolic and diastolic blood pressure; In the R language calculation formula, BMI is body mass index, GLU is fasting blood glucose, SBP is systolic blood pressure, DBP is diastolic blood pressure, TG is triglycerides, and HDL is high-density lipoprotein.

4. The survival prediction model according to claim 3, characterized in that, Among the components of metabolic syndrome, BMI is a non-linear influencing factor, while GLU, SBP, DBP, TG, and HDL are linear influencing factors.

5. The survival prediction model according to claim 1, characterized in that, The age is a continuous variable, and the value of the age variable is the patient's age in years; The family history is a discrete variable, represented by the value of the family history variable: 0: no family history of cancer, 1: family history of cancer. The N stage is a discrete variable, and the value of the N stage variable represents the lymph node metastasis in the tumor area of ​​the patient, 0:N0, 1:N1, 2:N2, 3:N3; The metastasis site is a discrete variable, represented by the value of the metastasis site variable: 0: no metastasis, 1: multiple metastases, brain metastasis, 3: bone metastasis, 4: abdominal metastasis, 5: lung metastasis, 6: liver metastasis, 7: metastasis site unknown; The histological grading is a discrete variable, represented by the value of the histological grading variable: 1: low differentiation, 2: moderate differentiation, 3: high differentiation; The chemotherapy is a discrete variable, represented by the value of the chemotherapy variable: 0: no chemotherapy received, 2: chemotherapy received. The metabolic integral is a continuous variable.

6. The survival prediction model according to any one of claims 1-5, characterized in that, In the model construction module, the method of constructing a nomogram predicting the overall survival of pancreatic cancer patients using the feature data of the input module is to perform multivariate Cox regression analysis on the feature data of the input module and then draw the nomogram.

7. The survival prediction model according to any one of claims 1-5, characterized in that, The total survival is defined as 1-year, 3-year, and 5-year total survival.

8. Use of the survival prediction model according to any one of claims 1-7 in the preparation of a device for predicting the prognosis of pancreatic cancer.

9. A computer-readable storage medium having stored thereon a survival prediction model as described in any one of claims 1-7.

Images