Application technology for enabling screening, early auxiliary diagnosis and postoperative curative effect evaluation of gynecological cancer high-risk group by artificial intelligence algorithm

Abstract

The invention provides an application technology for enabling screening, early auxiliary diagnosis and postoperative curative effect evaluation of gynecological cancer high-risk groups through an artificial intelligence algorithm, and belongs to the field of molecular biology, medical diagnosis and treatment and disease prevention. The method comprises the following steps: collecting samples of human body liquid, exosomes and the like, detecting the DNA methylation level, miRNA expression quantity, protein level and the quantity of molecular markers such as autoantibodies, exosomes, kits and the like of the gynecological cancer by virtue of a modern detection technology, and obtaining individual precision medical treatment by virtue of a model and computing power (including a software algorithm, a large model, intelligent agent application and an artificial intelligence method). The problems of screening of gynecological cancer high-risk groups, early auxiliary diagnosis, pathological typing, clinical staging, metastatic state, postoperative curative effect evaluation and poor accuracy of national approved kits are solved, and the non-destructive detection technology can be used for timely prevention, precise medical treatment and cancer development tendency analysis, and has wide social benefits and popularization and use values. And the traditional Chinese medicine also has a popularization effect on non-gynecological cancers and other cancers.

Description

Technical Field

[0001] The present invention relates to multidisciplinary technical fields such as mathematical models, mathematical application algorithms, molecular bioinformatics, preventive medicine, biological detection technology, medical diagnostics, artificial intelligence applications, cancer therapy, big health, medical imaging, immune system and biotechnology. Background Art

[0002] Gynecological cancer (cervical cancer, endometrial cancer, and ovarian cancer) is a major malignant tumor that affects women worldwide. Early clinical symptoms are often hidden, making diagnosis difficult. Due to the lack of mature early screening methods, many patients are diagnosed in the late stage, when the malignant tumor has already spread and metastasized, easily missing the optimal treatment opportunity. Postoperative cure is poor, and the five-year survival rate is significantly reduced. Gynecological cancer biopsy (a more accurate and commonly used method) and pathological examination are the gold standard for medical diagnosis. These are invasive tests, and preoperative testing is very painful for patients, resulting in new challenges in determining the right patients for biopsy and pathology examination and when. Imaging diagnostic tests have the advantage of detecting tumor lesions and nodules, but they differ significantly from the molecular biological methods of cancer diagnosis (pathology and histology). Imaging diagnostic tests are based on the experience accumulated from cancer morphological changes and are also affected by the subjective experience of the radiologist. Pathology and histology, the gold standard for diagnosis, are based on molecular biology, including gene, cell, and protein diagnosis. To approach the gold standard of pathology and histology, research is being conducted from a molecular perspective. Analyzing molecular biological genes, cells, and proteins at the level of individualized screening is a current hot topic. With the advancement of science and technology, the sensitivity and specificity of cancer-related test kits (or molecular markers) approved for clinical use in China are low. While research on molecular markers related to gynecological cancer, such as DNA methylation levels, protein levels, miRNA expression, and immune markers, has made some progress, selecting and optimizing a panel of these markers remains challenging. With the rapid development of innovative applications of big models, big data, artificial intelligence, digitalization, and intelligent agents, the effective integration of the results of any number of test kits or molecular markers with models, algorithms, and AI methods presents significant challenges. To overcome these shortcomings, the use of non-invasive detection technologies to detect molecular marker data related to gynecological cancer, combined with a series of models and computing power (or large models, algorithms, intelligent agent applications, or AI methods), has been proposed to address the screening, triage, early auxiliary diagnosis, and postoperative treatment monitoring of high-risk gynecological cancer populations. However, such applications in gynecological cancer have yet to be formally reported domestically or internationally. Such research can improve accuracy, prevent over-medicalization, achieve individual precision medicine, and achieve the purpose of screening high-risk groups, early detection, early auxiliary diagnosis and guiding patients' postoperative treatment effects through non-destructive testing. Summary of the Invention

[0003] The present invention aims to overcome the shortcomings of biotechnology, medical diagnosis and treatment, preventive medicine, and healthcare applications. By non-invasively collecting samples from human fluids or exosomes, the present invention uses existing biological and medical testing technologies to detect DNA methylation levels, miRNA expression, protein levels, various immune indicators, any nationally approved test kits, and any molecular markers, as well as any combination or preferred combination of these. The results are then calculated using mathematical models and applied algorithms (including large models, algorithms, software, intelligent agent applications, or artificial intelligence methods). This approach improves the accuracy of screening, triage, early auxiliary diagnosis, and postoperative treatment monitoring for high-risk gynecological cancer populations. Through early screening, diagnosis, and treatment, the cure rate of gynecological cancer patients can be improved, seizing the optimal opportunity for early screening, early detection, and early treatment, saving gynecological cancer patients from the brink of death.

[0004] The present invention provides a method for comprehensively evaluating any combination of multiple markers in a nationally approved test kit to ultimately determine whether the test kit is negative or positive for cancer. The indicator system for molecular markers is expanded to include imaging markers such as nodules, masses, lesions, and solid blood flow status. Through algorithms enabled by models, computing power, and artificial intelligence, the goal of accurate auxiliary diagnosis of individual gynecological cancers is directly achieved. The invention aims to characterize the histological development and changes of cancer by establishing and constructing linear or nonlinear functions, and to use non-invasive measurement in combination with mathematical models, model algorithms, linear or nonlinear functions, and artificial intelligence to address the histological development and changes, and the classification and identification problems of non-invasive detection of gynecological cancers. Using non-invasive testing methods, pathological classification, cancer metastasis and non-metastasis, invasive versus non-invasive status, and clinical staging are crucial for physicians' preoperative preparation. This approach addresses clinical cancer treatment assessment and guidance. Software is pre-programmed to generate data, models, and algorithms. This software is then installed on a portable device, microcomputer, mobile phone, online platform, or proprietary device. Individual test results are then input into the portable device, microcomputer, mobile phone, online backend, platform, or proprietary device. After calculation, the individual's results and medical conclusions are output. The invention integrates with various key institutions and service providers, leveraging network connectivity to expand its scope of application.

[0005] The subject matter of this invention encompasses not only gynecological cancer treatments but can also be extended to various other cancers within oncology. Similarly, for any other cancer, AI algorithms can be applied to screen high-risk gynecological cancer patients, provide early diagnosis assistance, and assess postoperative efficacy. DETAILED DESCRIPTION

[0006] Gynecological cancer implementation case: Through a large amount of actual clinical data, scientific and reasonable screening was conducted to determine the five related gene molecular biomarkers (B1, B2, B3, B4 and B5) involved in the occurrence, development and evolution of ovarian cancer. Patients with pathological diagnosis of ovarian cancer and tumors were set as the high-risk group (represented by X), and patients with pathological diagnosis of various precancerous lesions and tumors were set as the medium-risk (or low-risk) group (represented by Y). In advance, a clinical population of 453 ovarian cancer patients and a population of 445 precancerous lesions were tested in the hospital using biotechnology and medical testing to obtain the experimental results of the five related gene molecular biomarkers. Through data processing of the experimental results, statistics and probability principles were used to make a prediction. The probability of various events is obtained, and a probability database is established in advance, such as: p(X) = 0.504454, p(Y) = 0.495546, p(B11|X) = 0.132450, p(B12|X) = 0.216336, p(B13|X) = 0.269316, p(B14|X) = 0.381898, etc. are the values ​​of the probability database. Case 1: There is a suspected female patient who has ovarian pain. Her immediate family has no relatives who have died of cancer. She is willing to undergo ovarian risk screening (or is willing to undergo ovarian auxiliary diagnosis in the hospital) and then decide whether to undergo a pathological biopsy. By drawing peripheral blood, through biotechnology and medical testing, the first molecular biological The marker is equal to 46, the second molecular biomarker is equal to 53.43, the third molecular biomarker is equal to 43.58, the fourth molecular biomarker is equal to 63.2, and the fifth molecular biomarker is equal to 54. Assuming that the data of the five molecular markers actually measured are represented by B12, B21, B31, B41, and B52 respectively, and the individual test result of this woman is represented by B, denoted by the symbol B=B12B21B31B41B52, using a calculation method of the Beyes model in the model algorithm application technology involved in the present invention, it is obtained that p(B|X)=p(B12|X)×p(B21|X)×p(B31|X)×p(B41|X)×p(B52|X)=0.21633 6×0.289183×0.337748×0.289183×0.291391=0.001781, and similarly we get p(B|Y)=p(B12|Y)×p(B21|Y)×p(B31|Y)×p(B41|Y)×p(B52|Y)=0.049040, P(B|X)=(p(X)× p(B|X))=0.000898P(B|Y)=(p(Y)×p(B|Y))=0.024302P(B)=P(X)P(B|X)+P (Y)P(B|Y)=0.000898+0.024302=0.025200P(X|B)=0.000898 / 0.025200=0.072000P(Y|B)=0.024302 / 0.025200=0.964365Max{0.072000,0.964365}=0.964365. The calculation results show that this woman is in the medium-risk or low-risk ovarian population, not the high-risk population.

[0007] From the perspective of risk screening, this woman went to a tertiary hospital for further non-invasive vaginal color ultrasound imaging diagnosis to find out whether there was a mass and the degree of blood flow changes in the entity. The auxiliary diagnosis result of vaginal color ultrasound was benign, which was consistent with the result calculated by the model algorithm application technology. This woman did not need to undergo a pathological biopsy for the time being. Further screening was recommended to determine the development trend of the disease and regulate the immune system from a preventive perspective. There are many types of model algorithm application technologies, which can fully realize individual precise triage, risk screening, auxiliary diagnosis support, and accurate prediction of individual disease changes. From a preventive perspective, save money to do big things and solve problems.

Claims

1. An artificial intelligence algorithm enables the application of technology for screening high-risk populations for gynecological cancer, early auxiliary diagnosis, and postoperative efficacy evaluation, characterized by: All existing biological and medical non-invasive detection technologies are used to detect data on any combination of molecular markers in the "basic molecular marker index system" described below; the model algorithms included in the "basic model algorithm" described below are used to perform calculations on the data on any combination of molecular markers, thereby achieving the classification and identification included in the "basic classification and identification problem objectives" described below, thereby achieving the non-invasive individual precision prediction medical goals of early high-risk population screening, triage, early auxiliary diagnosis, and postoperative cancer efficacy evaluation for cervical cancer, ovarian cancer, and endometrial cancer. It should be noted that the term "cancer" in the following claims represents cervical cancer, ovarian cancer, and endometrial cancer, respectively. The basic classification and recognition problem objectives involved in the first paragraph are: basic classification and recognition problem objectives consisting of the following two, three, four, and above classification and recognition problems, of which two categories (category A and category B) classification and recognition: that is, gynecological nodules (or masses or lesions) are positive and negative or malignant and benign; cervical cancer, ovarian cancer, and endometrial cancer are respectively regarded as high risk (the same below) and medium risk (medium risk refers to the three cancers respectively representing their corresponding precancerous lesion states, the same below); the medium risk corresponding to cancer and their corresponding clinical stages 1 and 2; cancer stages 1 and 2 and their corresponding stages 3 and 4; Lymph node metastasis and non-metastasis before cancer surgery; metastasis and non-metastasis before cancer surgery; Cancer invasion and non-cancer invasion; test kit positive and negative; cancer recurrence and non-recurrence after surgery; treatment tendency of patients after cancer surgery getting better or worse; Medium risk and no risk (no risk refers to normal healthy people, or pathological conditions with mild normal diseases and cancer that have not yet begun, the same below); cervical no risk and LSIL, LSIL and HSIL; HSIL and cervical cancer; cancer and other cancers; identification of any two different pathological types in the pathological types corresponding to cancer; no risk and high risk; cervical cancer and endometrial cancer; cervical cancer and ovarian cancer; ovarian cancer and endometrial cancer; and other two types of gynecological identification. Its three categories (A, B, and C) identification: In the aforementioned A and B category identifications, the C category that is arbitrarily expanded as needed is used to distinguish between A and B, and the A, B, and C category identifications will be expanded as needed. For example: cervical cancer, ovarian cancer, endometrial cancer; no risk, medium risk, high risk; identification of any three different pathological types in the pathological types corresponding to cancer; high, medium and low differentiation corresponding to cancer; Class A, Class B and Class C three-category identification includes two cross-identifications of Class A and Class B, Class A and Class C, Class B and Class C; cancer with lymph node metastasis, corresponding to cancer without lymph node metastasis, respectively, corresponding to medium risk; cancer corresponding to medium risk, clinical stage 1 and 2, clinical stage 3 and 4; LSIL, HSIL, cervical cancer; cervical no risk, LSIL, HSIL; LSIL and HSIL, cervical cancer three-category identification; and so on other three types of gynecological identification. The four categories (Class A, Class B, Class C, Class D) and above are: in the above-mentioned Class A, Class B, Class C, Class D can be arbitrarily expanded as needed to distinguish Class A For example, medium risk, ovarian cancer, endometrial cancer, and cervical cancer; medium risk for cancer, any three different pathological types in the corresponding pathological types of cancer; no risk, medium risk for cancer, clinical stage 2 and 1 for cancer, clinical stage 3 and 4 for cancer; stage 1, 2, 3, and 4 for cancer; high, medium, and low differentiation for cancer, medium risk for cancer; normal healthy people, LSIL, HSIL, and cervical cancer; and other four gynecological identification categories. In the above four identification categories of A, B, C, and D, the E category can be expanded as needed to distinguish the differences between A, B, C, and D, and expanded to five categories. In short, in the above classifications, by analogy, such as healthy people, LSIL, HSIL, cervical cancer stages 1 and 2, and cervical cancer stages 3 and 4; For example, any four distinct classes of cancer (high, medium, and no risk (healthy people)), pathological classification, TNM staging, stages 1, 2, 3, and 4, metastasis and non-metastasis, and differentiation level constitute four-class classification. Any four or more distinct classes constitute more than four-class classification, and so on for other gynecological classification and recognition applications. The indicator system of basic molecular markers involved in the first paragraph is that the DNA methylation level corresponding to the gene, the DNA methylation map (methylation block) level of multiple genes, methylation of tfDNA, methylation of ctDNA, miRNA (micro RNA) expression (or expression level), protein level (protein expression level), protein concentration, various indicators in autoantibodies, various indicators in exosomes, any national or provincial approved cancer-related test kit (referring to the test kit on the website of the State Food and Drug Administration, not listed here one by one) is equivalent to a molecular marker, and each of the 12 high-risk HPV-DNA loads (or levels) (HPV16-DNA, HPV18-DNA, not listed here one by one) represents a molecular marker, and its gene DNA methylation levels include: CDO1, BHLHE22 , CELF4, IST1H4F, HIST1H4, PCDHGB7, TERT, C13ORF18, PAX1, FAM19A4, ZNF582, CCNA1, SOX14, LHX2, LHX5, LHX9, ZNF671, SOX17, RXPF3 , ITGA4, ASTN1, LncRNA, SOX21-AS1, SLIT2, HIC1, HPV16-DNA, HPV18-DN, AFBN1, SOX1, SOX9, KLF4, MGMT, SLC5A8, SOX14, HPV-DNA, HPV LCR, HPV E2, CDF1, CADM1, JAM3, RUNX3, SIX6, HAND2, HTR1B, RASSF1, GSTP1, HOXA9, OPCML, SPARC, RASSF1A, DAPK1, P1 6. SFRP1, tfDNA, ctDNA, PAX1, MGMT, DOK1, SLIT2, SOX1, APC, CHDH1, RUNX3, TFP12, SRP5, BRCA1, CDKN2A, SFN TMS1, WT1, AMPD3, NRN1, TBX15, CDH13, ZIC1, MLH1, COL2A1, etc., and miRNA expression levels include: miR-638, miR-1246, miR-1266, miR-203a-3p, miR-21, COX-2, miR-497-5p, miR-378, miR-150, miR-193a-5p, miR-92, miR-93, miR-21, miR-29b, miR-126, miR-30a-5p, miR-200a, miR-200b, miR-200c, miR-141, miR-205, miR-106b, and miR-100; miR-200c-3p, ...150, miR-193a-5p, miR-92, miR-93miR-221-3p,miR-21-5p and miR-484,miR-200c-3p,miR-221-3p,TmiR-3a-0p,miR-3c-0p,miR-373,miR-15b-5 p,miR-17-5p,miR-887-5p,miR-7-5p,miR-24-3p,miR-218-5p,miR-136-5p,miR-377-3p,miR-15a-5p,miRNA 27a,miRNA150-5p,miR-15a-5p,miR-142-3p,miR-151a-5p,miR-143-3p,miR-195-5p,miR-204-5p,m iRNA-30d-5p,let-7d-3p,miR-15a-5p,miR-106b-5p,miR-205,miR-145,miR-30a-5p,miR-92a,miRNA Let-7, miR-7-5p, miR-24-3p, miR-136-5p, miR-20b-5p, miR-204-5p, miR-146a-5p, miR-151a-5p, miR-423-3p, miR-484, etc., protein levels (expression levels) or autoantibody indicators or protein concentrations include: P16, ApoE, CRRABP2, S100A2 , OPN, SCC (SCC-Ag), TK1, TSGF, P185, VEGF-C, CEA, CA153, AFP, RUNX3, CA242, sPLA2, IL-23, COX-2, MMP-9, CA50, CA242, D-dimer, LPA, M-CSF, G-CSF, SPOCK, SLPI, HE4, CA125, CA50, VEGF, B7-H4, TPS, CA153, GFR-1, IL-6, IL-8, VEGF, MSLN, CYFRA21-1, P53, Gal-9, DKK1, TNF-α, NSE, M-CSF, G-CSF, SPOC, SMRP, VCAME, GFR-1, VEGF, MSLN, CRP, sVEGFR1, NLR, CEA, LDH, CRP, NLR, YKL-40, CA199, SAA, MSLN, IGF-1, sVEGFR1, SPAG9, HSP70, PKD1, CA724, LP4, etc. It is clearly stated that molecular markers also include protein levels, protein concentrations, the amount of autoantibody indicators, the amount of exosome indicators, multiple gene DNA methylation maps (methylation blocks), gene DNA methylation levels, miRNA expression levels, which are not listed above but will be discovered in the future.The test kits and so on that the country will approve in the future are all regarded as molecular markers. All the molecular markers described above constitute the index system of basic molecular markers. This index system of basic molecular markers also includes taking any one or any two or more molecular markers in the index system of basic molecular markers, sampling in human fluids (plasma, serum, peripheral blood, blood-free tfDNA, blood-free ctDNA, urine, exosomes, saliva, ascites, circulating free DNA, circulating tumor DNA, cervical exfoliated cells, cervical scrapings, vaginal fluid), processing the samples with sample technology and using fluorescent quantitative PCR, quantitative methylation-specific PCR, real-time fluorescent quantitative PCR (qPCR), methylation chips, digital PCR, ELISA, immunofluorescence Chemiluminescence, electrochemiluminescence, next-generation sequencing, hybrid capture technology, miRNA expression profiling, nationally approved and unapproved, as well as foreign test kit detection technologies, domestic and foreign Abbott and Roche measurement instrument detection methods, or future developments, etc., any one, two, or more molecular markers in the basic molecular marker indicator system are tested to obtain corresponding test data (or data from other sources). It should be noted that the positive or negative test results of the test kit, or the results of the test kit that can be quantified, are all included in the data here. The basic model algorithms involved in the first paragraph are: random forest algorithm classification, AdaBoost classification algorithm, decision tree classification algorithm, logistic regression classification (including sparse canonical correlation analysis and logistic regression methods, etc.), Bayesian classification algorithm, support vector machine classification algorithm, neural network classification algorithm, machine learning classification algorithm (including hybrid deep learning model classification, etc.), deep learning classification algorithm, cluster analysis, discriminant analysis algorithm, multi-task association analysis method (HDS-MTAA), artificial intelligence classification algorithm, generative artificial intelligence classification algorithm, K-nearest neighbor algorithm, association rule classification algorithm, kernel-based algorithm, K-meas algorithm, multi-layer perceptron (MLP) classification, extreme gradient boosting (XGBoost) classification, hierarchical clustering algorithm, model Fuzzy classification recognition algorithm, principal component analysis, large model classification algorithm, deep self-reconstruction fusion similarity hashing method (DS-FSH), other classification algorithm recognition algorithms, etc., or a comprehensive evaluation algorithm based on one of their classification recognition algorithms or a combination of any two or more classification recognition algorithms, or for a specific basic classification recognition problem target selected in advance, arbitrarily combine a group of specific molecular markers in the indicator system of basic molecular markers, and the data after detection (or any data obtained from other aspects, or detection data, or two data sources), perform any algorithm after optimization or any determined algorithm, or any algorithm among the above classification algorithms. All of the above algorithms and any combination of them constitute the basic model algorithm.

2. An artificial intelligence algorithm enables the screening of high-risk groups for gynecological cancer, early auxiliary diagnosis, and postoperative efficacy evaluation, characterized by: After accurately screening the benign or malignant ovarian tissue according to the above method 1, the benign and malignant early auxiliary diagnosis of the ovarian tissue is performed in combination with imaging. Take ovarian cancer as an example (the same ideas and methods are used for the diagnosis of benign and malignant cervix or endometrium, which are also protected by this right): (1) Use a similar method as the above method 1 to obtain the results of high risk (or positive) and medium risk (or negative) ovarian cancer. If the results are consistent with any of the four transvaginal ultrasound images or CT or MRI or PET-CT examination results, the consistent results are determined; if they are inconsistent, whether the examinee has direct relatives with a genetic history of cancer, if so, the risk of ovarian cancer tends to be high (positive), if not, it is determined in combination with other clinical examination results. (2) Among the four transvaginal ultrasound imaging, CT, MRI, or PET-CT tests, taking transvaginal ultrasound imaging as an example (CT, MRI, or PET-CT use the same ideas and methods, and are all protected by the same rights): According to the above-mentioned approach 1, after the individual accurately screens for ovarian malignancy, plus the transvaginal ultrasound test showing any of the following conditions, it is judged to be positive, which is: (a) a punctate blood flow signal is visible in the entire solid part of the mass or lesion; (b) a solid mass or a partially solid lesion is seen in the appendage, and a punctate blood flow signal is visible in it. , (c) there is infiltration in and around the lesion or mass, (d) there is metastasis to the tissues and organs involved in the lesion or mass, or there are signs of lymph node metastasis; according to the above method 1, after the individual is accurately screened for benign ovarian disease, and the vaginal ultrasound examination shows any of the following situations, it is judged to be negative, which are: (a) there is no lesion or mass in the ovarian tissue, (b) there is no blood flow signal inside the whole or part of the solid mass in the adnexal area, (c) there is no blood flow signal around and inside the lesion or mass, (d) there is a dot-shaped or line-shaped blood flow signal around the mass and no blood flow signal inside.

3. An artificial intelligence algorithm enables the screening of high-risk groups for gynecological cancer, early auxiliary diagnosis, and postoperative efficacy evaluation, characterized by: Use any one or any two or more molecular markers in the indicator system of basic molecular markers to characterize the molecular variation law of the process from the development of healthy people to the formation of cancer (or the molecular evolution process law of cancer or the risk law of the molecular evolution process of cancer) in cancer tissue. On the other hand, use non-destructive measurement methods, according to the above method 1, or construct linear functions or nonlinear functions to establish a molecular variation law (or the molecular evolution process law of cancer or the risk law of the molecular evolution process of cancer) that reflects the process from the development of healthy people to the formation of cancer in cancer tissue. Through the mode of non-destructive measurement, with the help of the algorithm or intelligent model of the above method 1, the molecular evolution process law of cancer in the tissue is replaced, so as to achieve the purpose of screening high-risk populations for cancer or clinical auxiliary diagnosis.

4. An artificial intelligence algorithm enables the screening of high-risk groups for gynecological cancer, early auxiliary diagnosis, and postoperative efficacy evaluation, characterized by: Research on the pathological classification of cancer, and identification of the presence or absence of invasion and metastasis. Using a similar method to the above 1, different pathological types of ovarian cancer are used as classification categories, and any two, three, or four of these categories are identified. The presence or absence of metastasis or invasion of ovarian cancer is identified as a binary classification. Research on the pathological classification of cervical cancer or endometrial cancer, identification of the presence or absence of invasion, and metastasis is similar to the above ovarian cancer research and will not be repeated, but is also protected by this right. Similar methods and approaches can also be used in combination with imaging results to improve accurate individual identification, which is also protected by this right.

5. An artificial intelligence algorithm enables the application of technology for screening high-risk populations for gynecological cancer, early auxiliary diagnosis, and postoperative efficacy evaluation, characterized by: Cancer clinical staging studies for phases 1, 2, 3, and 4. A positive (or high-risk) screening or diagnosis result has been determined in 1 or 2 above. In the basic molecular marker index system, any molecular marker with staging characteristics is selected. Using the test data, a linear function of one type of linear function or a nonlinear function of another type of nonlinear function is mathematically constructed. The zero root of this linear function or nonlinear function is calculated using mathematical methods and data. This zero root is the first clinical stage. Within the basic molecular marker index system, randomly select another molecular marker with staging characteristics and similarly derive a second clinical stage. Repeat this process n times (n is arbitrarily determined), and the arithmetic average of these n clinical stages is the specific clinical stage value for the positive (or high-risk) patient. This falls under the protection of this right. The TNM method for clinical staging, referring to the N, T, and M clinical stages, also falls under the protection of this right. For any specific pathological type of cancer, similar research can be used to derive the clinical stage value for that specific pathological type, also falling under the protection of this right. Cancer here is not limited to ovarian, cervical, and endometrial cancer, but refers to any type of cancer, such as liver, lung, and gastric cancer. Calculating the clinical stage value using the above method falls within the scope of this provision of this right. For any specific cancer, using the two similar methods and similar research approaches described above to perform clinical staging, improving accurate individual identification, also falls under the protection of this right.

6. An artificial intelligence algorithm enables the application of technology for screening high-risk populations for gynecological cancer, early auxiliary diagnosis, and postoperative efficacy evaluation, characterized by: For any specific type of cancer (such as liver cancer, cervical cancer, gastric cancer, ovarian cancer, breast cancer, etc.), many corresponding preferred molecular biomarkers can be found. Any test kit approved by the state or provincial level is equivalent to a molecular marker. Among the molecular markers corresponding to any of the above specific cancers, any two or more molecular markers can be randomly selected, and each molecular marker has a method for determining negative and positive. All the methods for determining the final positive and negative of any two or more molecular markers taken are carried out according to the following methods one, two, three, four and five, respectively, which are all within the scope of the requirements of this right and this clause. Method 1: If the sum of the number of positives is greater than the sum of the number of negatives, the final diagnosis is positive; if the sum of the number of positives is less than the sum of the number of negatives, the final diagnosis is negative; Method 2: For each molecular marker that has been determined to be positive or negative, the size of each molecular marker's contribution to the entirety is compared to determine the weight coefficient of each molecular marker (or each molecular marker has an equal weight coefficient, i.e., each molecular marker has an equal probability value), each molecular marker takes a value of 1 for actual positive and 0 for actual negative (or different values ​​are assigned to the two), and the value of each different molecular marker is multiplied by their corresponding weight coefficients, and the resulting sum is greater than or equal to a fixed number or less than a fixed number to determine positive and negative; Method three: expert scoring method, or hierarchical analysis method to set the weight coefficient, and then use similar methods of method two for calculation; Method four: entropy weight method, or principal component analysis method, or coefficient of variation method to respectively calculate the weight coefficient of each different molecular marker, and then use similar methods of method two for calculation; Method five: The National Drug Administration website will approve any test kit (or any diagnostic kit) related to any cancer, including any multiple molecular biomarkers, in the development, and the methods of determining negative and positive according to the above-mentioned methods one, two, three and four are all within the scope of the requirements of this right and this clause.

7. An artificial intelligence algorithm enables the screening of high-risk groups for gynecological cancer, early auxiliary diagnosis, and postoperative efficacy evaluation, characterized by: The index system of the basic molecular markers in the above 1 is expanded to include molecular markers related to cancer that are newly discovered in the future, as well as molecular markers equivalent to cancer-related test kits that are newly approved by the country in the future. These newly discovered molecular markers are put together with the index system of the basic molecular markers in the above 1, and are processed using similar methods as the above 1, 2, 3, 4 and 5 to achieve the classification and identification purpose of the above 1, which falls within the scope of the requirements of this clause of this right. Its characteristic is that the research on the above 1, 2, 3, 4, 5 and 6 will be combined with large models, intelligent agent applications, artificial intelligence AI algorithms, and generative AI technologies respectively, or the research and application of large models, intelligent agent applications, artificial intelligence AI algorithms, generative AI technologies and new mathematical algorithms to solve the problems of 1, 2, 3, 4, 5 and 6 will fall within the scope of the requirements of this right and this clause.

8. An artificial intelligence algorithm enables the screening of high-risk groups for gynecological cancer, early auxiliary diagnosis, and postoperative efficacy evaluation, characterized by: Use similar methods as mentioned in 1, 2, 3, 4, 5 and 7 above respectively, for any postoperative cancer patient (for any subject), adopt non-destructive detection technology, take any number of molecular markers, perform one or two or more measurements, and perform one or two or more experimental treatments to obtain corresponding data, dynamically observe the development trend of the efficacy of postoperative cancer patients (dynamically observe the development trend of the risk level of the subjects), and then use quantitative methods to analyze and compare, help doctors guide treatment, and evaluate the treatment effect (and then use quantitative methods to analyze and compare, help subjects prevent, guide them to go to the hospital for treatment, and guide the immune system to regulate balance).

9. An artificial intelligence algorithm enables the screening of high-risk groups for gynecological cancer, early auxiliary diagnosis, and postoperative efficacy evaluation, characterized by: Use similar methods as mentioned above 1, 2, 3, 4, 5, 6, 7 and 8 to conduct research respectively, and take any (or preferably take) combination of multiple molecular markers, and calculate them into various indexes or comprehensive indexes after a large amount of non-destructive detection data processing. For any person, the data of multiple corresponding molecular markers are tested experimentally, and the data are substituted into various indexes or comprehensive indexes to calculate the negative and positive of that person's cancer accordingly. The invention is further characterized by processing a large amount of non-invasive test data using any (or preferably any) combination of multiple molecular markers, and then using similar methods as described in 1, 2, 3, 4, 5, 6, 7, and 8 above. Software codes are compiled using different similar methods and pre-installed on a portable device (or microcomputer, mobile phone, online platform, proprietary device, etc.) for calculation and storage. For any individual, the data of multiple molecular markers detected through non-invasive experiments is inserted into the portable device (or microcomputer, mobile phone, online platform, proprietary device, etc.) for individual calculations, thereby achieving the various research objectives covered by this invention. Its characteristic is that for any specific cancer (such as liver cancer, gastric cancer, breast cancer, etc.), a corresponding indicator system of basic molecular markers similar to the above 1 can be found. In any specific cancer, any one or any two or more molecular markers can be taken, and similar studies can be conducted in the above 1, 2, 3, 4, 5, 6, 7, 8, and 9. Then, any key module paths corresponding to these key module paths (AI+model and algorithm+medical testing institution: AI+model and algorithm+medical equipment; AI+model and algorithm+mobile phone; AI+model and algorithm+laptop computer (desktop computer); AI+model and algorithm+portable instrument; AI+model and algorithm+portable instrument+detection and collection bookstore; AI+model and algorithm+Internet system+service customer (hospital); AI+model and algorithm+hospital cooperation (cooperation between hospital or local hospital or nursing home)) are combined to provide intelligent application services, which fall within the scope of this claim.

10. An artificial intelligence algorithm enables the screening of high-risk groups for gynecological cancer, early auxiliary diagnosis, and postoperative efficacy evaluation, characterized by: For any molecular marker associated with any other specific cancer (such as ovarian cancer, breast cancer, colorectal cancer, pancreatic cancer, gastric cancer, etc.) (including any test kit that is equivalent to a molecular marker), among them, any combination of molecular markers, the above-mentioned 1, 2, 3, 4, 5, 6, 7, 8, 9 high-risk cancer screening, early auxiliary diagnosis and postoperative treatment efficacy research of cancer patients can be respectively extended to any other specific cancer, and any other specific cancer can use similar ideas, models and algorithm ideas of the above-mentioned 1, 2, 3, 4, 5, 6, 7, 8, 9 to respectively solve the high-risk population screening, early auxiliary diagnosis, pathological classification, clinical staging and postoperative treatment efficacy evaluation technology of any other specific cancer, which all fall within the scope of the requirements of this clause of this right.