Ovarian cancer risk assessment system based on tumor marker counting fusion index ALC

Abstract

The invention discloses an ovarian cancer risk assessment system based on a tumor marker counting fusion index ALC, relates to the technical field of medical detection, and aims to solve the problem that existing ovarian cancer diagnosis mainly depends on experience of inspectors, and consequently diagnosis accuracy is low. The method comprises the following steps: acquiring the absolute values of lactic dehydrogenase, CA125 and lymphocyte and the prognosis result of an ovarian cancer patient, combining the absolute values of lactic dehydrogenase, CA125 and lymphocyte to obtain an ALC value, finally training a neural network by using the ALC value and the prognosis result, and performing prognosis prediction by using the trained neural network. The ALC value obtained in the application has very high sensitivity and specificity, so that the prediction accuracy of ovarian cancer prognosis is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical detection, and specifically to an ovarian cancer risk assessment system based on the tumor marker count fusion index ALC. Background Art

[0002] Ovarian cancer is a common malignant tumor in the female reproductive system, ranking fourth in terms of incidence globally and being one of the main causes of cancer-related deaths in women. The early symptoms of ovarian cancer are not obvious. As the disease progresses, symptoms such as abnormal vaginal bleeding and pelvic pain may occur. Early invasive cervical cancer may not have any related abnormal signs, while invasive cervical cancer can be detected by gynecological examination to find a cervical mass. Diagnostic examinations for ovarian cancer include cervical / vaginal cytological smear examination and HPV detection, colposcopy, cervical biopsy, etc.

[0003] In recent years, the research progress of ovarian cancer-related tumor markers has provided new ideas for the diagnosis of ovarian cancer. For example, tumor markers such as SCC-Ag, CA199, and CEA have important guiding significance in the diagnosis, treatment effect evaluation, and prognosis assessment of ovarian cancer. In addition, vascular endothelial growth factor (VEGF), specific growth factor (TSGF), osteopontin (OPN), serum cytokeratin 19 fragment antigen (CYFRA21-1), tumor suppressor gene p53, p16 / Ki-67, cyclooxygenase-2 (COX-2), CD44, matrix metalloproteinase-9 (MMP-9), and microRNA (miRNA) have also been studied as potential tumor markers for ovarian cancer, and they show important value in the early detection of ovarian cancer, prediction of lymph node metastasis, prognosis assessment, etc.

[0004] In the prior art, the diagnosis of ovarian cancer mainly relies on serum tumor markers (such as CA125 and HE4) and imaging examinations (such as transvaginal ultrasound). However, the accuracy of these methods highly depends on the experience of the examiners, and insufficient experience will lead to a low diagnostic accuracy rate. Summary of the Invention

[0005] The object of the present invention is to provide an ovarian cancer risk assessment system based on the tumor marker count fusion index ALC for the problem that the existing diagnosis of ovarian cancer mainly relies on the experience of examiners, which in turn leads to a low diagnostic accuracy rate.

[0006] The technical solution adopted by the present invention to solve the above technical problems is:

[0007] An ovarian cancer risk assessment system based on the tumor marker count fusion index ALC, the system includes an ALC module and a prediction module;

[0008] The ALC module is used to obtain the ALC value of ovarian cancer patients;

[0009] The prediction module is used to input the ALC value of ovarian cancer patients into the trained neural network to obtain the output ovarian cancer prognosis result;

[0010] The trained neural network is obtained through the following steps:

[0011] Step 1-1: Obtain the ALC value of historical ovarian cancer patients and their corresponding ovarian cancer prognosis results;

[0012] Step 1-2: Use the ALC value of historical ovarian cancer patients as the input and their corresponding ovarian cancer prognosis results as the output to train the neural network to obtain the trained neural network;

[0013] The ALC value of the ovarian cancer patient is obtained through the following steps:

[0014] Step 2-1: Obtain the lactate dehydrogenase value in the blood of ovarian cancer patients;

[0015] Step 2-2: Obtain the carbohydrate antigen 125 value in the blood of ovarian cancer patients;

[0016] Step 2-3: Obtain the absolute lymphocyte count in the blood of ovarian cancer patients;

[0017] Step 2-4: According to the lactate dehydrogenase value in the blood of ovarian cancer patients, the carbohydrate antigen 125 value in the blood of ovarian cancer patients, and the absolute lymphocyte count in the blood of ovarian cancer patients, obtain the ALC value, and the ALC value is expressed as:

[0018] ALC = 7×LDH + 23×CA125 + 11×Lymphocyte

[0019] Among them, LDH is the lactate dehydrogenase value in the blood of ovarian cancer patients, CA125 is the carbohydrate antigen 125 value in the blood of ovarian cancer patients, and Lymphocyte is the absolute lymphocyte count in the blood of ovarian cancer patients.

[0020] Furthermore, the specific steps of Step 2-1 are as follows:

[0021] Step 2-1-1: Obtain the serum sample of ovarian cancer patients;

[0022] Step 2-1-2: After mixing the serum sample with the reaction reagent containing lactic acid and NAD - substrate, put it into a spectrophotometer or biochemical analyzer. The reaction temperature inside the instrument is set to 37°C, and the change in absorbance at a wavelength of 340 nm is detected. After the reaction starts, according to the preset time interval, record the absorbance value every once in a while, and then obtain the data of the change in absorbance over time;

[0023] Step 213: Obtain the absorbance change rate based on the data of the absorbance changing with time;

[0024] Step 214: Establish a standard curve using LDH standard products with different concentrations, and use the standard curve and the absorbance change rate to obtain the activity of LDH in the sample, that is, the lactate dehydrogenase value in the blood of ovarian cancer patients.

[0025] Further, the specific steps of Step 22 are as follows:

[0026] Step 221: Obtain a serum sample of an ovarian cancer patient, and check whether there is precipitation or turbidity in the serum sample. If there is turbidity, centrifuge it at a low speed to make it clear;

[0027] Step 222: Add the serum sample to a reaction tube or microplate that has been coated with a CA125 specific antibody. The sample volume added to each well is 100 - 200 μL, and then place the reaction tube or microplate in an incubator at 37 °C for incubation for 45 - 60 minutes to allow the antigen and antibody to fully bind;

[0028] Step 223: After incubation, wash with a washing solution containing a buffer and a non-ionic surfactant, and suck dry the washing solution after each wash;

[0029] Step 224: After washing, add another CA125 antibody labeled with a chemiluminescent substance, incubate and wash again, then add a chemiluminescent activator, and detect the luminescence intensity using a chemiluminescent immunoassay analyzer;

[0030] Step 225: Establish a standard curve using a strictly calibrated CA125 standard product, and convert the luminescence intensity into the concentration of CA125 using the standard curve.

[0031] Further, the specific steps for obtaining the serum sample of an ovarian cancer patient are as follows:

[0032] Use a blood collection tube to obtain a blood sample of an ovarian cancer patient, then place the blood collection tube in a centrifuge, set the rotation speed to 2000 - 2500 r / min, and centrifuge for 8 - 10 minutes to separate the serum from the blood cells. After centrifugation, take out the blood collection tube from the centrifuge, use a pipette to aspirate the upper-layer serum sample, and transfer the serum sample to another centrifuge tube.

[0033] Further, the specific steps of Step 24 are as follows:

[0034] Use a blood collection tube to obtain a blood sample of an ovarian cancer patient, then place the blood collection tube on the sample rack of a blood cell analyzer, and use the blood cell analyzer to obtain the absolute value of lymphocytes.

[0035] Further, the neural network is a DAWEL network, and the DAWEL network includes a multi-perspective base model, a dynamic attention weight generator, and a self-supervised auxiliary task module;

[0036] The multi-perspective base model includes a LightGBM model, a 1D-CNN model, and a Transformer model.

[0037] The LightGBM model is used to obtain the output of the ovarian cancer prognosis prediction result according to the input ALC value.

[0038] The 1D-CNN model is used to obtain the output of the ovarian cancer prognosis prediction result according to the input ALC value.

[0039] The Transformer model is used to obtain the output of the ovarian cancer prognosis prediction result according to the input ALC value.

[0040] The dynamic attention weight generator is used to obtain the ALC fusion index. The ALC fusion index is expressed as:

[0041]

[0042] α LightGBM + α 1D-CNN + α Transformer = 1

[0043] where α LightGBM 、α 1D-CNN 、α Transformer are dynamic weights;

[0044] The self-supervised auxiliary task module is used to reconstruct the ALC fusion index. for reconstruction.

[0045] Further, the loss function of the DAWEL network is:

[0046] L total = L LightGBM + L 1D-CNN + L Transformer + L DAW + λL recon

[0047]

[0048] where N is the number of samples, y i is the true label, λ is a hyperparameter, and L LightGBMis the loss function of the LightGBM model, L 1D-CNN is the loss function of the 1D-CNN model, L Transformer is the loss function of the Transformer model, L DAW is the loss function of the dynamic attention weight generator, L recon is the loss function of the self-supervised auxiliary task module, X i is the true survival time of the i-th sample, is the predicted survival time of the i-th sample by the model.

[0049] Further, the low-speed centrifugation in step 221 is specifically:

[0050] Set the rotation speed to 1000 - 1500 revolutions per minute and the centrifugation duration to 3 - 5 minutes for low-speed centrifugation.

[0051] Further, the number of washing times in step 223 is 3 - 4 times.

[0052] Further, the chemiluminescent substance in step 224 is acridinium ester.

[0053] The beneficial effects of the present invention are:

[0054] In this application, by obtaining the lactate dehydrogenase, CA125, and absolute lymphocyte count of ovarian cancer patients, as well as the prognosis results of ovarian cancer patients, and combining the lactate dehydrogenase, CA125, and absolute lymphocyte count, the ALC value is obtained. Finally, the neural network is trained using the ALC value and the prognosis results, and the trained neural network is used for prognosis prediction. The ALC value obtained in this application has high sensitivity and specificity, thereby improving the prediction accuracy of ovarian cancer prognosis. Description of the Drawings

[0055] Figure 1 is the violin plot of CA125 for five-year prognosis;

[0056] Figure 2 is the violin plot of the absolute monocyte count for five-year prognosis;

[0057] Figure 3 is the violin plot of lactate dehydrogenase LDH for five-year prognosis;

[0058] Figure 4 is the violin plot of the absolute lymphocyte count for five-year prognosis;

[0059] Figure 5 is the violin plot of platelets for five-year prognosis;

[0060] Figure 6 is the violin plot of the absolute neutrophil count for five-year prognosis;

[0061] Figure 7 Violin plot of ALC for five-year prognosis;

[0062] Figure 8 Violin plot of CA125 for three-year prognosis;

[0063] Figure 9 Violin plot of absolute monocyte count for three-year prognosis;

[0064] Figure 10 Violin plot of absolute lymphocyte count for three-year prognosis;

[0065] Figure 11 Violin plot of lactate dehydrogenase (LDH) for three-year prognosis;

[0066] Figure 12 Violin plot of absolute neutrophil count for three-year prognosis;

[0067] Figure 13 Violin plot of platelets for three-year prognosis;

[0068] Figure 14 Violin plot of ALC for three-year prognosis;

[0069] Figure 15 Schematic diagram of AUC value for five-year prognosis;

[0070] Figure 16 Schematic diagram of AUC value for three-year prognosis. Detailed implementation manners

[0071] It should be specifically noted that, without conflict, the various implementation manners disclosed in this application can be combined with each other.

[0072] Detailed implementation manner one: The ovarian cancer risk assessment system based on the tumor marker count fusion index ALC described in this implementation manner includes an ALC module and a prediction module;

[0073] The ALC module is used to obtain the ALC value of ovarian cancer patients;

[0074] The prediction module is used to input the ALC value of ovarian cancer patients into a trained neural network to obtain the output ovarian cancer prognosis result;

[0075] The trained neural network is obtained through the following steps:

[0076] Step one: Obtain the ALC value of historical ovarian cancer patients and their corresponding ovarian cancer prognosis results;

[0077] Step two: Use the ALC value of historical ovarian cancer patients as the input and their corresponding ovarian cancer prognosis results as the output to train the neural network to obtain the trained neural network;

[0078] The ALC value of the ovarian cancer patient is obtained through the following steps:

[0079] Step 2-1: Obtain the lactate dehydrogenase value in the blood of the ovarian cancer patient;

[0080] Step 2-2: Obtain the carbohydrate antigen 125 value in the blood of the ovarian cancer patient;

[0081] Step 2-3: Obtain the absolute lymphocyte count in the blood of the ovarian cancer patient;

[0082] Step 2-4: According to the lactate dehydrogenase value in the blood of the ovarian cancer patient, the carbohydrate antigen 125 value in the blood of the ovarian cancer patient, and the absolute lymphocyte count in the blood of the ovarian cancer patient, obtain the ALC value, and the ALC value is expressed as:

[0083] ALC = 7×LDH + 23×CA125 + 11×Lymphocyte

[0084] Wherein, LDH is the lactate dehydrogenase value in the blood of the ovarian cancer patient, CA125 is the carbohydrate antigen 125 value in the blood of the ovarian cancer patient, and Lymphocyte is the absolute lymphocyte count in the blood of the ovarian cancer patient.

[0085] Specifically, obtaining the lactate dehydrogenase value in the blood of the ovarian cancer patient includes:

[0086] Step 2-1-1: Obtain the serum sample of the ovarian cancer patient;

[0087] Step 2-1-2: After mixing the serum sample with a reaction reagent containing lactic acid and NAD - substrate, place it in a spectrophotometer or biochemical analyzer. The reaction temperature inside the instrument is set to 37°C, and the change in absorbance at a wavelength of 340 nm is detected. After the reaction starts, according to the preset time interval, record the absorbance value every once in a while, and then obtain the data of the change in absorbance over time;

[0088] Step 2-1-3: According to the data of the change in absorbance over time, obtain the absorbance change rate;

[0089] Step 2-1-4: Use LDH standard products with different concentrations to establish a standard curve, and use the standard curve and the absorbance change rate to obtain the activity of LDH in the sample, that is, the lactate dehydrogenase value in the blood of the ovarian cancer patient.

[0090] The test results are recorded in the unit of U / L.

[0091] The specific steps for obtaining the serum sample of the ovarian cancer patient are:

[0092] Blood samples from ovarian cancer patients are obtained using blood collection tubes, which are then placed in a centrifuge with a speed set at 2000-2500 r / min for 8-10 minutes to separate serum from blood cells. After centrifugation, the blood collection tubes are removed from the centrifuge, the upper serum sample is aspirated with a pipette, and the serum sample is transferred to another centrifuge tube.

[0093] Obtaining the carbohydrate antigen 125 value in the blood of ovarian cancer patients specifically includes:

[0094] Step 221: obtaining a serum sample from an ovarian cancer patient and checking whether the serum sample has precipitation or turbidity. If turbidity is present, clarifying the sample by low-speed centrifugation;

[0095] Step 222: Add the serum sample to the reaction tube or microplate coated with CA125-specific antibody, with the sample volume added to each well being 100-200 μL. Then place the reaction tube or microplate in an incubator at 37°C for 45-60 minutes to allow the antigen and antibody to fully bind.

[0096] Step 223: After the incubation, wash with a washing solution containing a buffer and a non-ionic surfactant, and dry the washing solution after each wash;

[0097] Step 224: After washing, add another CA125 antibody labeled with a chemiluminescent substance, incubate and wash again, then add a chemiluminescent stimulator, and use a chemiluminescent immunoassay to detect the luminescence intensity;

[0098] Step 225: Use a strictly calibrated CA125 standard to establish a standard curve, and use the standard curve to convert the luminescence intensity into CA125 concentration. The unit is U / mL

[0099] The low speed centrifugation in step 221 is specifically as follows:

[0100] Set the centrifuge speed to 1000-1500 rpm and centrifuge for 3-5 minutes.

[0101] The number of washing in the steps 223 is 3-4 times.

[0102] In step 224, the chemiluminescent substance is acridinium ester.

[0103] The specific steps of step 24 are:

[0104] Step 241: Place the blood collection tube in the sample rack of the blood cell analyzer. The instrument will automatically draw the sample and then dilute and hemolyze it according to the preset program. For samples from ovarian cancer patients, since their blood components may be abnormal, the instrument may need to appropriately adjust the dilution multiple and hemolysis conditions based on the preliminary test results of the sample (such as indicators such as hematocrit).

[0105] Step 242: The preprocessed samples pass through the detection channels inside the instrument, and lymphocytes are identified and counted using electrical and optical principles. The instrument distinguishes lymphocytes from other blood cells based on characteristics such as the size of lymphocytes and internal structures (through techniques such as laser scattering and fluorescence staining). After the detection is completed, the instrument will automatically calculate the absolute value of lymphocytes, with the unit of ×109 / L.

[0106] Calculate the ALC value of the sample to be tested according to the following formula;

[0107] ALC = 7×LDH + 23×CA125 + 11×Lymphocyte

[0108] Put the calculated ALC into the fusion algorithm framework of Dynamic Attention-Weighted Ensemble Learning (DAWEL). This algorithm combines the dynamic attention mechanism with ensemble learning to achieve adaptive adjustment of the weights of the base models for the purpose of diagnostic discrimination.

[0109] Among them, DAWEL consists of the following three parts:

[0110] 1. Multi-view base model construction: Simultaneously train 3 heterogeneous base models to capture different characteristics of the data. Train the LightGBM model to process structured features; train the 1D-CNN model to extract local sequence patterns; train the Transformer model to capture long-range dependencies.

[0111] (1) LightGBM model

[0112] Input: ALC fusion metric

[0113] Output: Predicted survival time

[0114] Loss function: Adopt the mean squared error (MSE) loss function, and the formula is as follows:

[0115]

[0116] Among them, N is the number of samples, y i is the true label, is the model prediction value

[0117] (2) 1D-CNN model

[0118] Input: ALC fusion metric

[0119] Convolutional layer: Use multiple one-dimensional convolutional layers to extract local features, with the convolutional kernel size of 3 and the stride of 1

[0120] Pooling layer: The maximum pooling layer is used to reduce the feature dimension.

[0121] Fully connected layer: The feature map output by the convolutional layer is flattened and then input into the fully connected layer to obtain the predicted time output of the model.

[0122] Loss function: The cross-entropy loss function is adopted, and the formula is as follows:

[0123]

[0124] (3) Transformer model

[0125] Input: ALC fusion metrics

[0126] Encoder: It contains a multi-head self-attention mechanism and a feed-forward neural network, and extracts long-range dependence features through multiple layers of encoders.

[0127] Features

[0128] Output: Predicted survival time

[0129] Loss function: The cross-entropy loss function is adopted, and the formula is as follows:

[0130]

[0131] 2. Dynamic attention weight generator: Design a lightweight neural network (including a gating mechanism + residual connection), which inputs the original features and the intermediate layer output of the base model, and generates the dynamic weights of each base model in real time, rather than the static weights of the traditional model. The specific network structure is as follows:

[0132] Input: Original features ALC and the intermediate layer output H of the base model LightGBM , H 1D-CNN , H Transformer

[0133] Network structure: It contains multiple fully connected layers and a gating mechanism, and enhances the training stability of the model through residual connections. Output: Dynamic weights α LightGBM , α 1D-CNN , α Transformer , satisfying:

[0134] α LightGBM + α 1D-CNN + α Transformer = 1

[0135] Loss function: The mean squared error loss function is adopted, and the formula is as follows:

[0136]

[0137] Among them,

[0138] 3. Self-supervised auxiliary task: During the training phase, a feature reconstruction task is added. The input data is reconstructed through the decoder, forcing the model to learn more robust feature representations and alleviating the problem of overfitting with few samples. The specific network structure is as follows:

[0139] Input: The fusion metric ALC

[0140] Decoder: A multi-layer fully connected layer network is adopted to reconstruct the fused features into the input features

[0141] Loss function: The mean squared error loss function is adopted, and the formula is as follows:

[0142]

[0143] 4. Overall loss function

[0144] The loss functions of the above parts are combined to obtain the overall loss function:

[0145] L total = L LightGBM + L 1D-CNN + L Transformer + L DAW + λL recon

[0146] where λ is a hyperparameter used to balance the weight of the self-supervised auxiliary task.

[0147] After passing through the DAWEL model, the patient's prognostic survival time is finally output.

[0148] 1. By drawing a violin plot, as Figures 1 to 14 shown, the distribution and probability density of the data are presented. Grouped by survival and death, compared with the existing metrics, the ALC metric has obvious discrimination in the distribution for 5-year survival and 3-year survival.

[0149] 2. Perform a p-value test. When the p-value is less than 0.05, we consider there to be a significant difference. The null hypothesis is that there is no significant difference between the ALC metric and the original metrics of absolute lymphocyte count, absolute neutrophil count, absolute monocyte count, platelet count, CA125, and lactate dehydrogenase LDH for 5-year survival. According to the p-value calculation, the p-values of the absolute monocyte count, platelet count, CA125, lactate dehydrogenase LDH, and ALC metric are much smaller than the significance level, indicating that 5-year survival has a significant impact on the ALC metric, and the p-value of the ALC metric is the smallest, with the most significant prognostic effect for 5 years. In the p-value test for 3-year prognosis, the p-value of the ALC metric is less than 0.05, achieving a significant effect and meeting the requirements.

[0150]

[0151]

[0152]

[0153] Example:

[0154] For the five-year prognosis, as Figure 15 shown, the AUC value of the ALC index is 0.71, the AUC value of CA125 is 0.67, the AUC value of the absolute lymphocyte count is 0.58, the AUC value of the absolute neutrophil count is 0.56, the AUC value of lactate dehydrogenase LDH is 0.70, the AUC value of the absolute monocyte count is 0.61, and the AUC value of the platelet count is 0.65. In comparison, the ALC index has a better ability to distinguish survival at five years.

[0155] For the three-year prognosis as Figure 16 shown, the AUC value of the ALC index is 0.57, the AUC value of CA125 is 0.51, the AUC value of the absolute lymphocyte count is 0.62, the AUC value of the absolute neutrophil count is 0.46, the AUC value of lactate dehydrogenase LDH is 0.54, the AUC value of the absolute monocyte count is 0.55, and the AUC value of the platelet count is 0.57. In comparison, the ALC index still has a better ability to distinguish survival at three years.

[0156] Finally, the ALC values of patients with a survival time within 1 year, 1 - 3 years, 3 - 5 years, and over 5 years were statistically analyzed respectively, and the normal range of ALC was obtained as [11000, 21000).

[0157] It should be noted that the specific implementation manners are only explanations and illustrations of the technical solutions of the present invention, and the scope of the claimed rights cannot be limited thereby. Those that are only partial changes made according to the claims and the description of the present invention should still fall within the protection scope of the present invention.

Claims

1. An ovarian cancer risk assessment system based on tumor marker count fusion index ALC, characterized by The system includes an ALC module and a prediction module; The ALC module is used to obtain the ALC value of ovarian cancer patients; The prediction module is used to input the ALC value of the ovarian cancer patient into the trained neural network to obtain the output ovarian cancer prognosis result; The trained neural network is obtained by the following steps: Step 1: Obtain the ALC values ​​of historical ovarian cancer patients and their corresponding ovarian cancer prognosis results; Step 1 and 2: Take the ALC values ​​of historical ovarian cancer patients as input and the corresponding ovarian cancer prognosis results as output, train the neural network, and obtain a trained neural network; The ALC value of the ovarian cancer patient is obtained by the following steps: Step 21: Obtaining lactate dehydrogenase values ​​in the blood of ovarian cancer patients; Step 22: Obtaining the carbohydrate antigen 125 value in the blood of ovarian cancer patients; Step 23: Obtain the absolute value of lymphocytes in the blood of ovarian cancer patients; Step 24: According to the lactate dehydrogenase value in the blood of ovarian cancer patients, the carbohydrate antigen 125 value in the blood of ovarian cancer patients and the absolute value of lymphocytes in the blood of ovarian cancer patients, the ALC value is obtained, and the ALC value is expressed as: ALC=7×LDH+23×CA125+11×Lymphocyte Among them, LDH is the lactate dehydrogenase value in the blood of ovarian cancer patients, CA125 is the carbohydrate antigen 125 value in the blood of ovarian cancer patients, and Lymphocyte is the absolute value of lymphocytes in the blood of ovarian cancer patients.

2. The ovarian cancer risk assessment system based on tumor marker count fusion index ALC according to claim 1, characterized in that The specific steps of step 21 are: Step 2-1: Obtain serum samples from ovarian cancer patients; Step 212: Mix serum samples with lactate and NAD - After the reaction reagents of the substrate are mixed, they are placed in a spectrophotometer or biochemical analyzer. The reaction temperature inside the instrument is set to 37°C, and the change of absorbance at a wavelength of 340nm is detected. After the reaction starts, the absorbance value is recorded once every certain period of time according to the preset time interval, thereby obtaining the data of absorbance change over time; Step 213: Obtain the absorbance change rate according to the data of absorbance change over time; Step 214: Use LDH standards of different concentrations to establish a standard curve, and use the standard curve and the absorbance change rate to obtain the activity of LDH in the sample, that is, the lactate dehydrogenase value in the blood of ovarian cancer patients.

3. The ovarian cancer risk assessment system based on tumor marker count fusion index ALC according to claim 1, characterized in that The specific steps of step 22 are: Step 221: obtaining a serum sample from an ovarian cancer patient and checking whether the serum sample has precipitation or turbidity. If turbidity is present, clarifying the sample by low-speed centrifugation; Step 222: Add the serum sample to the reaction tube or microplate coated with CA125-specific antibody, with the sample volume added to each well being 100-200 μL. Then place the reaction tube or microplate in an incubator at 37°C for 45-60 minutes to allow the antigen and antibody to fully bind. Step 223: After the incubation, wash with a washing solution containing a buffer and a non-ionic surfactant, and dry the washing solution after each wash; Step 224: After washing, add another CA125 antibody labeled with a chemiluminescent substance, incubate and wash again, then add a chemiluminescent stimulator, and use a chemiluminescent immunoassay to detect the luminescence intensity; Step 225: Use a strictly quantified CA125 standard to establish a standard curve, and use the standard curve to convert the luminescence intensity into the concentration of CA125.

4. The ovarian cancer risk assessment system based on tumor marker count fusion index ALC according to claim 2 or 3, characterized in that The specific steps of obtaining serum samples from ovarian cancer patients are: Blood samples from ovarian cancer patients are obtained using blood collection tubes, which are then placed in a centrifuge with a speed set at 2000-2500 r / min for 8-10 minutes to separate serum from blood cells. After centrifugation, the blood collection tubes are removed from the centrifuge, the upper serum sample is aspirated with a pipette, and the serum sample is transferred to another centrifuge tube.

5. The ovarian cancer risk assessment system based on tumor marker count fusion index ALC according to claim 4 is characterized in that The specific steps of step 24 are: A blood sample from an ovarian cancer patient is obtained using a blood collection tube, which is then placed on a sample rack of a blood cell analyzer to obtain an absolute value of lymphocytes using the blood cell analyzer.

6. The ovarian cancer risk assessment system based on tumor marker count fusion index ALC according to claim 1, characterized in that The neural network is a DAWEL network, which includes a multi-view base model, a dynamic attention weight generator, and a self-supervised auxiliary task module; The multi-view base model includes the LightGBM model, the 1D-CNN model and the Transformer model. The LightGBM model is used to obtain the output ovarian cancer prognosis prediction result according to the input ALC value The 1D-CNN model is used to obtain the output ovarian cancer prognosis prediction result according to the input ALC value. The Transformer model is used to obtain the output ovarian cancer prognosis prediction result according to the input ALC value The dynamic attention weight generator is used to obtain the ALC fusion index The ALC fusion index It is expressed as: α LightGBM +α 1D-CNN +α Transformer ＝1 Among them, α LightGBM , α 1D-CNN , α Transformer is the dynamic weight; The self-supervised auxiliary task module is used to integrate the ALC fusion index Refactoring.

7. The ovarian cancer risk assessment system based on tumor marker count fusion index ALC according to claim 6, characterized in that The loss function of the DAWEL network is: THE total =L LightGBM +L 1D-CNN +L Transformer +L DAW +λL recon Where N is the number of samples, y i is the true label, λ is a hyperparameter, L LightGBM is the loss function of the LightGBM model, L 1D-CNN is the loss function of the 1D-CNN model, L Transformer is the loss function of the Transformer model, L DAW is the loss function of the dynamic attention weight generator, L recon is the loss function of the self-supervised auxiliary task module, X i is the actual survival time of the i-th sample, Predict the survival time for the i-th sample model.

8. The ovarian cancer risk assessment system based on tumor marker count fusion index ALC according to claim 7, characterized in that The low speed centrifugation in step 221 is specifically as follows: Set the centrifuge speed to 1000-1500 rpm and centrifuge for 3-5 minutes.

9. The ovarian cancer risk assessment system based on tumor marker count fusion index ALC according to claim 8, characterized in that The number of washing in the steps 223 is 3-4 times.

10. The ovarian cancer risk assessment system based on tumor marker count fusion index ALC according to claim 9, characterized in that In step 224, the chemiluminescent substance is acridinium ester.

Images