A method for constructing a model to predict the risk of ovarian cancer related to hyperglycemia

Abstract

The present invention relates to the field of medical information processing technology, and specifically to a method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia. The method comprises: performing an isolated analysis of a patient's blood sugar level at each physical examination and combining it with local trend characteristics of the blood sugar level to calculate the patient's blood sugar abnormality at each physical examination; analyzing the difference between the patient's blood sugar level and the overall trend information, and combining the patient's blood sugar abnormality to assess blood sugar abnormality; utilizing the correlation between blood sugar level and tumor marker concentration at different time periods, and combining the assessment results of the patient's blood sugar abnormality at each physical examination, calculating the prediction reliability of the patient's blood sugar level at each physical examination, thereby performing data cleaning on the blood sugar level data to further predict the patient's risk of ovarian cancer associated with hyperglycemia. The present invention effectively improves the accuracy of predicting the risk of ovarian cancer associated with hyperglycemia.

Description

Technical Field

[0001] The present invention relates to the technical field of medical information processing, and in particular to a method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia. Background Art

[0002] Ovarian cancer is a common malignant tumor of the female reproductive system. Due to its lack of obvious early symptoms and lack of effective screening methods, it is often discovered in the late stages, resulting in a high mortality rate. The pathogenesis of ovarian cancer is complex and closely related to multiple factors, including genetics, hormone levels, environmental factors, and lifestyle. Recent studies have shown a correlation between hyperglycemia and the development of ovarian cancer.

[0003] Currently, research on ovarian cancer risk prediction remains limited. Existing detection methods often rely on tumor marker measurements, which often have limitations, such as low sensitivity and a high risk of false-positive or false-negative results. Furthermore, existing studies on the relationship between hyperglycemia and ovarian cancer risk mostly focus on single-factor analysis, lacking comprehensive consideration and modeling of multiple related factors. This results in inaccurate predictions of hyperglycemia-related ovarian cancer risk. Summary of the Invention

[0004] The present invention provides a method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia, so as to solve the existing problems.

[0005] The method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia of the present invention adopts the following technical solutions:

[0006] One embodiment of the present invention provides a method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia, the method comprising the following steps:

[0007] Obtaining patients' blood glucose and tumor marker concentration data;

[0008] Based on the blood glucose data, the patient's blood glucose level at each physical examination is analyzed in isolation, and the degree of blood glucose abnormality of the patient at each physical examination is calculated by combining the local trend characteristics of the blood glucose level at each physical examination; the overall trend information of blood glucose level changes is extracted from the patient's blood glucose data, and the difference between the patient's blood glucose level and the overall trend information is analyzed. Combined with the patient's blood glucose abnormality, the patient's blood glucose abnormality at each physical examination is evaluated;

[0009] The correlation between blood glucose levels and tumor marker concentrations at different time periods was used, and combined with the assessment results of the patient's blood glucose abnormalities at each physical examination, the predicted reliability of the patient's blood glucose level at each physical examination was calculated. The blood glucose level data was then cleaned using the predicted reliability, and the cleaned blood glucose level data was used to predict the patient's risk of high blood glucose-related ovarian cancer.

[0010] Preferably, the blood glucose level of the patient at each physical examination is analyzed in isolation based on the blood glucose level data, and the blood glucose abnormality of the patient at each physical examination is calculated in combination with the local trend characteristics of the blood glucose level at each physical examination, including the specific method of:

[0011] The blood glucose value data is used as the input of the isolation forest algorithm. The isolation coefficient of each blood glucose value in the blood glucose value data is obtained through the isolation forest algorithm. The isolation coefficient of the blood glucose value is used in combination with the local change trend of the blood glucose value data to analyze and calculate the patient's blood glucose abnormality at each physical examination.

[0012] Preferably, the method of analyzing and calculating the abnormal blood sugar level of the patient at each physical examination by utilizing the isolated coefficient of the blood sugar level and combining the local variation trend of the blood sugar level data includes the following specific methods:

[0013] A second-order difference sequence of blood glucose value data is obtained, and the second-order difference value corresponding to each blood glucose value in the second-order difference sequence is obtained. The absolute value of the second-order difference value is used as the local trend parameter of the corresponding blood glucose value. For the patient's blood glucose value at any physical examination, the blood glucose abnormality of the blood glucose value is obtained by combining the isolation coefficient and the local trend parameter of the blood glucose value, wherein the blood glucose abnormality is positively correlated with the isolation coefficient and the local trend parameter.

[0014] Preferably, the extracting of the overall trend information of blood glucose level changes from the patient's blood glucose level data includes the following specific methods:

[0015] A two-dimensional rectangular coordinate system is constructed and the patient's blood glucose value data is mapped into the two-dimensional rectangular coordinate system. In the two-dimensional rectangular coordinate system, the horizontal axis represents the patient's physical examination order and the vertical axis represents the patient's blood glucose value under the corresponding physical examination order. The blood glucose value data in the two-dimensional rectangular coordinate system are curve-fitted using the least squares method to obtain the patient's blood glucose value curve. The blood glucose value curve is decomposed using the STL time series decomposition algorithm, and the trend item of the blood glucose value curve is extracted and recorded as the trend curve.

[0016] Preferably, the analysis of the difference between the patient's blood glucose value and the overall trend information, combined with the patient's blood glucose abnormality, thereby evaluating the patient's blood glucose abnormality at each physical examination, includes the following specific methods:

[0017] The numerical value of the corresponding data point in the trend curve during each physical examination of the patient is recorded as the blood glucose trend value for the corresponding number of physical examinations. The blood glucose trend abnormality is obtained based on the difference between the patient's blood glucose value at each physical examination and the blood glucose trend value. The blood glucose abnormality amount of the patient at any physical examination is calculated based on the blood glucose trend abnormality and the blood glucose abnormality degree of the patient. The blood glucose abnormality amount is positively correlated with the blood glucose trend abnormality and the blood glucose abnormality degree.

[0018] Preferably, the specific calculation method of the abnormal blood sugar trend is:

[0019]

[0020] in, For patients Abnormal blood sugar trend during the first physical examination; For patients Blood sugar level at the time of the next physical examination; For patients Blood glucose trend value at the next physical examination; is the absolute value function; is a linear normalization function.

[0021] Preferably, the method of calculating the prediction reliability of the patient's blood glucose level at each physical examination by utilizing the correlation between blood glucose level and tumor marker concentration at different time periods and combining the evaluation results of the patient's blood glucose abnormality at each physical examination includes the following specific methods:

[0022] According to the patient's pathological cycle of hyperglycemia-related ovarian cancer, the collected blood glucose value data and tumor marker concentration data are divided into several periods, and the Pearson correlation coefficient between the blood glucose value data and any tumor marker concentration data in any same period is obtained;

[0023] Based on the Pearson correlation coefficient between blood glucose values ​​and tumor marker concentrations in the same period and the amount of blood glucose abnormality during physical examinations, the prediction reliability of the patient's blood glucose value at each physical examination was obtained.

[0024] Preferably, the step of obtaining the prediction reliability of the patient's blood glucose level at each physical examination further comprises: performing normalization processing on the prediction reliability using a sigmoid function.

[0025] Preferably, the method of using the prediction reliability of the blood glucose value to clean the blood glucose value data and using the cleaned blood glucose value data to predict the patient's risk of high blood glucose-related ovarian cancer includes the following specific methods:

[0026] The blood glucose values ​​whose prediction credibility is lower than a preset credibility threshold are deleted from the blood glucose value data, and the deleted blood glucose values ​​are interpolated and supplemented by linear interpolation to obtain a credible blood glucose sequence;

[0027] The ARIMA model is used to predict the data of the credible blood glucose sequence to obtain a predicted blood glucose sequence. The predicted blood glucose sequence is traversed in reverse order, and a number of high blood glucose corresponding blood glucose values ​​are obtained according to the size of the blood glucose values ​​in the predicted blood glucose sequence. The sequence segment formed by a number of consecutive high blood glucose corresponding blood glucose values ​​in the reverse traversal process is recorded as a high blood glucose sequence segment. The risk rate is obtained according to the number of blood glucose values ​​in the high blood glucose sequence segment, the numerical level of the blood glucose values, and the rate of change.

[0028] Preferably, the specific calculation method of the risk rate is:

[0029]

[0030] in, is the risk rate, is the number of blood glucose values ​​in the high blood glucose sequence segment, is the numerical level of blood glucose value in the high blood glucose sequence segment, is the rate of change of blood glucose value in the high blood glucose sequence segment, is a natural constant, is the sigmoid normalization function.

[0031] The beneficial effects of the technical solution of the present invention are as follows: the embodiment of the present invention constructs a multi-level dynamic assessment model by combining isolated analysis of blood glucose values, local trend characteristics, and overall change trends. This model not only focuses on abnormal fluctuations in a single physical examination, but also captures long-term deviation characteristics of blood glucose changes. This overcomes the limitations of traditional methods that rely solely on threshold judgments, significantly improving the sensitivity of abnormal blood glucose detection and the ability to analyze temporal correlations. In addition, by establishing a dynamic correlation model between blood glucose values ​​and tumor marker concentrations and analyzing the coordinated changes of the two at different stages of the disease course, a credibility calculation mechanism based on the joint evolution of metabolism and tumor markers is implemented. By dynamically adjusting the credibility weights of blood glucose value data at different periods, the impact of noise data caused by interference factors such as detection errors and physiological fluctuations is effectively eliminated. Key disease course characteristic information is also retained, making the cleaned data more consistent with the patient's actual metabolic state, effectively improving the accuracy of predicting the risk of ovarian cancer related to hyperglycemia. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0033] Figure 1 The present invention is a flowchart of the steps of a method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia. DETAILED DESCRIPTION

[0034] To further illustrate the technical means and effectiveness of the present invention in achieving its intended objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, details the method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia, including its specific implementation, structure, features, and effectiveness. In the following description, references to different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0035] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0036] The specific scheme of the method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia provided by the present invention is described in detail below with reference to the accompanying drawings.

[0037] See also Figure 1 , which shows a flowchart of a method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia provided by one embodiment of the present invention. The method includes the following steps:

[0038] Step S001: Obtain the patient's blood glucose value data and tumor marker concentration data.

[0039] It's important to note that when predicting a patient's risk of developing hyperglycemia-related ovarian cancer, it's important to fully consider multiple factors related to hyperglycemia (especially diabetes, high insulin levels, and insulin resistance) and ovarian cancer risk. The key evaluation indicators that influence the relationship between hyperglycemia and ovarian cancer risk include clinical indicators and ovarian cancer-related indicators. Clinical indicators primarily include blood glucose and insulin levels, while ovarian cancer-related indicators include the concentrations of the tumor markers CA-125 (carbohydrate antigen 125) and HE4 (human epididymis protein 4). The present invention constructs a model for predicting the risk of developing hyperglycemia-related ovarian cancer. The ultimate goal is to help clinicians identify high-risk individuals earlier, providing a basis for personalized treatment and early intervention, reducing disease incidence, and improving patient quality of life. Hyperglycemia and insulin resistance are hallmarks of diabetes, and diabetes itself is significantly associated with ovarian cancer risk. High insulin levels promote the development of ovarian cancer by activating the insulin-like growth factor (IGF) pathway, increasing cancer cell proliferation and inhibiting apoptosis. Therefore, in the embodiments of the present invention, a predictive model for ovarian cancer risk associated with hyperglycemia is established based on blood glucose levels and corresponding tumor marker concentrations, thereby achieving the goal of ovarian cancer risk management and prevention. Because the process of monitoring a patient's blood glucose levels and tumor marker concentrations to build a model involves long-term monitoring, the patient's blood glucose data is not just a static value but also includes fluctuations. This fluctuation may have different impacts on cancer at different time points. Therefore, peak and trough blood glucose levels have different clinical significance for ovarian cancer risk, and predictions based on these markers are subject to certain errors, resulting in low reliability.

[0040] Specifically, in order to implement the method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia proposed in this embodiment, it is first necessary to collect the patient's blood sugar level and tumor marker concentration data. The specific process is as follows:

[0041] First, the data sequences corresponding to the patient's blood glucose level and tumor marker concentration at each physical examination are obtained to obtain the patient's blood glucose level data and tumor marker concentration data. That is, each data point in the blood glucose level data and tumor marker concentration data corresponds to a physical examination order value and a corresponding blood glucose level or concentration value.

[0042] The tumor marker concentration data is divided into tumor marker CA-125 concentration data and HE4 concentration data.

[0043] Then, the data set was preprocessed, including data cleaning and missing value processing for blood glucose value data and tumor marker concentration data, and standardization was performed.

[0044] It should be noted that the purpose of preprocessing the data set is to screen out outliers through data cleaning to ensure the accuracy of the data. The purpose of standardizing the data is to eliminate the dimensional differences between blood glucose data and tumor marker concentrations and to weaken the impact of dimensional differences between different data.

[0045] At this point, blood sugar level data and tumor marker concentration data are obtained through the above method.

[0046] Step S002: Based on the blood glucose value data, an isolated analysis is performed on the patient's blood glucose value at each physical examination, and combined with the local trend characteristics of the blood glucose value at each physical examination, the patient's blood glucose abnormality at each physical examination is calculated; the overall trend information of the blood glucose value change is extracted from the patient's blood glucose value data, the difference between the patient's blood glucose value and the overall trend information is analyzed, and combined with the patient's blood glucose abnormality, the patient's blood glucose abnormality at each physical examination is evaluated.

[0047] It should be noted that in the process of analyzing the risk of ovarian cancer related to high blood sugar, the size of the blood sugar level plays a relatively critical role. The specific working principle is: the rapid fluctuation of blood sugar in a short period of time will have different effects on insulin sensitivity, thereby affecting the proliferation of cancer cells. For example, short-term high blood sugar fluctuations will lead to high insulin levels and activate the "insulin-like growth factor (IGF)" pathway, which promotes the proliferation and survival of cancer cells. On the contrary, if blood sugar remains low for a long time, it implies low insulin levels or insulin resistance, which in turn affects the risk of cancer. Therefore, if the blood sugar sample values ​​change repeatedly in multiple stages, it is necessary to analyze the abnormal amount of blood sugar dimension caused by the blood sugar value to prevent data anomalies caused by uncertain factors in the collection process, thereby causing deviations in the prediction results.

[0048] It should be further explained that uncertainties in the data collection process include: physical activity: Exercise of varying intensities or daily activities (such as walking and climbing stairs) can affect blood sugar levels, especially changes in blood sugar after exercise; changes in dietary composition: Even if the time windows before and after meals are strictly controlled, dietary types (such as high-sugar, high-fat, and high-protein) can have different effects on blood sugar; sleep quality: Poor sleep or insufficient sleep can affect insulin secretion and glucose metabolism, leading to blood sugar fluctuations; and menstrual cycle: Women's menstrual cycle can cause hormone level fluctuations, which in turn affects blood sugar. Therefore, the present embodiment chooses to first determine the degree of abnormality in blood sugar values ​​at different treatment stages. The more abnormal the stage, the greater the impact on the error in the prediction results when constructing the prediction model, and the more it needs to be optimized and eliminated.

[0049] Specifically, in step S201, based on the blood glucose data, an isolated analysis is performed on the patient's blood glucose value at each physical examination, and combined with the local trend characteristics of the blood glucose value at each physical examination, the blood glucose abnormality of the patient's blood glucose value at each physical examination is calculated.

[0050] As a preferred embodiment of the present invention, a specific method for obtaining the patient's blood sugar abnormality at each physical examination includes: using the blood sugar value data as the input of the isolation forest algorithm, obtaining the isolation coefficient of each blood sugar value in the blood sugar value data through the isolation forest algorithm, using the isolation coefficient of the blood sugar value and combining it with the local change trend of the blood sugar value data to analyze and calculate the patient's blood sugar abnormality at each physical examination, and performing linear normalization on the blood sugar abnormality. When the blood sugar abnormality is greater than or equal to a first threshold value A, the corresponding blood sugar value is abnormal, where the first threshold value A is a preset parameter.

[0051] It should be noted that the first threshold value A is preset to 0.9 based on experience, and the value of the first threshold value can be adjusted according to actual conditions, and is not specifically limited in the embodiment of the present invention.

[0052] As an optional embodiment of the present invention, the method for obtaining the isolation coefficient may further include: using a LOF algorithm to obtain the LOF value of each blood glucose value in the blood glucose value data, and using the LOF value of the blood glucose value as the corresponding isolation coefficient.

[0053] As a preferred embodiment of the present invention, the isolation coefficient of the blood glucose value is used in combination with the local change trend of the blood glucose value data to analyze and calculate the blood glucose abnormality of the patient at each physical examination. The specific method includes: obtaining a second-order difference sequence of the blood glucose value data, obtaining the second-order difference value corresponding to each blood glucose value in the second-order difference sequence, using the absolute value of the second-order difference value as the local trend parameter of the corresponding blood glucose value, and for the blood glucose value of the patient at any physical examination, combining the isolation coefficient and the local trend parameter of the blood glucose value to obtain the blood glucose abnormality of the blood glucose value, wherein the blood glucose abnormality is positively correlated with the isolation coefficient and the local trend parameter.

[0054] As an optional embodiment, the specific method for calculating the abnormal blood sugar level is:

[0055]

[0056] in, Indicates that the patient The degree of abnormal blood sugar level during the first physical examination; For patients The isolation coefficient corresponding to the blood sugar value at the time of the second physical examination; For patients Local trend parameters at the time of the second physical examination.

[0057] It should be noted that the larger the isolation coefficient of a data point in the blood glucose value data, the more abnormal the data point is, indicating that the patient's blood glucose value in the corresponding physical examination order is more likely to be isolated data affected by uncertain factors, such as abnormal detection caused by physical activity, etc., then including this uncertainty factor in the prediction of the future risk of ovarian cancer with high blood glucose will greatly affect the prediction results of some data; by analyzing the difference in blood glucose values ​​in several neighboring stages at this stage, the degree of differential abnormality can be quantitatively analyzed. Ideally, the trend of blood glucose values ​​rising or falling in different stages is fixed, that is, the difference The closer it is to 0, the lower the blood sugar abnormality.

[0058] Step S202: extracting the overall trend information of the blood glucose level change from the patient's blood glucose level data, analyzing the difference between the patient's blood glucose level and the overall trend information, and combining the patient's blood glucose abnormality to evaluate the patient's blood glucose abnormality during each physical examination.

[0059] It should be noted that, considering the high blood sugar-related ovarian cancer risk, blood sugar levels tend to increase or decrease as the condition evolves, abnormality should be determined not only by the blood sugar level but also by the degree of abnormality in the blood sugar trend.

[0060] As an embodiment, the method for extracting overall trend information of blood glucose level changes from a patient's blood glucose level data includes: constructing a two-dimensional rectangular coordinate system, mapping the patient's blood glucose level data into the two-dimensional rectangular coordinate system, wherein the horizontal axis of the two-dimensional rectangular coordinate system represents the patient's physical examination order, and the vertical axis represents the patient's blood glucose level under the corresponding physical examination order; using the least squares method to perform curve fitting on the blood glucose level data in the two-dimensional rectangular coordinate system to obtain the patient's blood glucose level curve; decomposing the blood glucose level curve using the STL time series decomposition algorithm; extracting the trend item of the blood glucose level curve, and recording it as a trend curve.

[0061] It should be noted that the trend curve reflects the overall trend of changes in all blood sugar levels of the patient after the physical examination, that is, it reflects the overall trend information of the blood sugar level changes in the patient's blood sugar level data.

[0062] As a preferred embodiment, the analysis of the difference between the patient's blood glucose value and the overall trend information is combined with the patient's blood glucose abnormality to evaluate the patient's blood glucose abnormality at each physical examination. The specific method includes: recording the numerical value of the corresponding data point in the trend curve at each physical examination of the patient as the blood glucose trend value under the corresponding number of physical examinations, and obtaining the blood glucose trend abnormality based on the difference between the patient's blood glucose value and the blood glucose trend value at each physical examination; calculating the patient's blood glucose abnormality at any physical examination through the patient's blood glucose trend abnormality and blood glucose abnormality, and the blood glucose abnormality is positively correlated with the blood glucose trend abnormality and the blood glucose abnormality.

[0063] As an embodiment, the specific calculation method of the abnormal blood sugar trend is:

[0064]

[0065] in, For patients Abnormal blood sugar trend during the first physical examination; For patients Blood sugar level at the time of the next physical examination; For patients Blood glucose trend value at the next physical examination; is the absolute value function; is a linear normalization function.

[0066] It should be noted that Reflects the patient's The difference between the blood glucose value and the blood glucose trend value at the time of the physical examination was analyzed. By analyzing the difference between the corresponding blood glucose value and the blood glucose trend value of the patient at each blood glucose test, the abnormality of the blood glucose value obtained in the corresponding physical examination sequence relative to the overall trend of blood glucose value change was quantified. For example, the blood glucose value obtained by the patient at a physical examination was 104 mg / dL, but after trend fitting, it should actually be 130 mg / dL, which proves that the blood glucose value at this physical examination reflected a certain abnormality in the trend of change.

[0067] As an example, the specific method for calculating the abnormal blood sugar level is:

[0068]

[0069] in, For patients Abnormal blood sugar level during the first physical examination; For patients Abnormal blood sugar trend during the first physical examination; Indicates that the patient Abnormal blood sugar level during the second physical examination.

[0070] It should be noted that the abnormal blood sugar level is used to comprehensively describe the abnormalities in the local and overall trends of the patient's blood sugar levels, as well as the degree of deviation of the blood sugar levels from other blood sugar levels. The more abnormal the blood sugar trend and blood sugar level, the higher the corresponding abnormal blood sugar level.

[0071] So far, the abnormal blood sugar level of the patient at each physical examination has been obtained through the above method.

[0072] Step S003: Utilizing the correlation between blood glucose levels and tumor marker concentrations at different time periods, and combining the assessment results of the patient's blood glucose abnormalities at each physical examination, the predicted reliability of the patient's blood glucose levels at each physical examination is calculated, and the blood glucose level data is cleaned using the predicted reliability. The cleaned blood glucose level data is then used to predict the patient's risk of ovarian cancer related to hyperglycemia.

[0073] It should be noted that when predicting the risk of hyperglycemia-related ovarian cancer, changes in blood glucose levels and abnormalities are only considered as a single-dimensional assessment factor and cannot effectively assess the risk of ovarian cancer. Therefore, embodiments of the present invention analyze abnormal blood glucose fluctuations while also incorporating changes in tumor marker concentrations to ensure that the blood glucose analysis results are correlated with important physiological factors related to ovarian cancer, thereby improving the accuracy of subsequent analysis results in predicting the risk of hyperglycemia-related ovarian cancer. For example, some patients' blood glucose levels may fluctuate due to non-cancer-related reasons (such as medication use and metabolic changes). In such cases, relying solely on blood glucose data to determine data unreliability may lead to misjudgment. However, changes in tumor marker concentrations (such as CA-125 and HE4) can effectively provide direct information on tumor activity. In particular, in the detection of specific cancers such as ovarian cancer, fluctuations in the corresponding tumor marker concentration data are directly correlated with disease progression. Therefore, ignoring this tumor marker concentration information and relying solely on blood glucose fluctuations can lead to overlooking signals related to cancer progression. However, due to the interaction of high blood sugar, the marker concentration values ​​will show synchronous changes. Therefore, when there is no corresponding synchronous change between the marker concentration and blood sugar, for example, the marker concentration changes steadily but the blood sugar fluctuates violently, the blood sugar value is considered to be unreliable, but the marker concentration can still be used as a predictive indicator.

[0074] As a preferred embodiment of the present invention, a method for obtaining the predicted reliability of a patient's blood glucose level at each physical examination is as follows: first, according to the patient's pathological cycle of hyperglycemia-related ovarian cancer, the collected blood glucose level data and tumor marker concentration data are divided into several periods, and the Pearson correlation coefficient between the blood glucose level data and any tumor marker concentration data at any given period is obtained; then, based on the Pearson correlation coefficient between the blood glucose level and tumor marker concentration at the same period and the abnormal blood glucose level at the physical examination, the predicted reliability of the patient's blood glucose level at each physical examination is obtained.

[0075] As an embodiment, the collected blood glucose value data and tumor marker concentration data are divided into periods to obtain several periods, wherein the several periods at least include: an observation period, a treatment period, and a recovery period.

[0076] It should be noted that there are multiple physical examinations in a period of time. When abnormal blood sugar levels are high in a period of time, the correlation between blood sugar and tumor markers in the same period should be analyzed. When the correlation is weak, there is no correlation between blood sugar changes and tumor marker concentrations. The prediction of blood sugar values ​​at this time is less reliable and needs to be optimized and screened out.

[0077] As an embodiment, a specific method for calculating the prediction reliability is:

[0078]

[0079] in, represents the prediction reliability of the patient at the uth physical examination, For patients The Pearson correlation coefficient between the blood sugar value data and the tumor marker CA-125 concentration data during the physical examination period, For patients The Pearson correlation coefficient between the blood sugar value data and the tumor marker HE4 concentration data during the physical examination period, For patients Abnormal blood sugar levels during the first physical examination.

[0080] It should be noted that the prediction credibility reflects the credibility of the prediction results when the blood glucose data is subsequently used to predict the risk of hyperglycemia-related ovarian cancer. Each physical examination corresponds to a prediction credibility. The larger the value, the lower the degree of abnormality of the patient's blood glucose level at this time, and the higher the correlation between the blood glucose level and the tumor marker concentration, then the patient is more likely to be at risk of hyperglycemia-related ovarian cancer when hyperglycemia occurs.

[0081] Finally, the sigmoid function is used to normalize the prediction credibility.

[0082] As an embodiment, a patient's risk of developing ovarian cancer related to hyperglycemia is predicted, including a specific method: first, blood glucose values ​​in the blood glucose value data whose prediction credibility (i.e., prediction credibility after normalization) is lower than a preset credibility threshold are deleted, and the deleted blood glucose values ​​are interpolated and supplemented by linear interpolation to obtain a credible blood glucose sequence; then, the credible blood glucose sequence is predicted using an ARIMA model to obtain a predicted blood glucose sequence, the predicted blood glucose sequence is traversed in reverse order, and a number of blood glucose values ​​corresponding to high blood glucose are obtained based on the size of the blood glucose values ​​in the predicted blood glucose sequence; a sequence segment formed by a number of consecutive blood glucose values ​​corresponding to high blood glucose during the reverse traversal process is recorded as a high blood glucose sequence segment, and a risk rate is obtained based on the number of blood glucose values ​​in the high blood glucose sequence segment, the numerical level of the blood glucose values, and the change rate.

[0083] It should be noted that the embodiment of the present invention presets the credibility threshold as 0.9 based on experience, and can be adjusted according to actual conditions, and the embodiment of the present invention does not make any specific limitation.

[0084] As an optional embodiment, the specific calculation method of the risk rate is:

[0085]

[0086] in, is the risk rate, is the number of blood glucose values ​​in the high blood glucose sequence segment, is the numerical level of blood glucose value in the high blood glucose sequence segment, is the rate of change of blood glucose value in the high blood glucose sequence segment, is a natural constant, is the sigmoid normalization function.

[0087] The numerical level of the blood glucose value in the hyperglycemia sequence segment is the average value of the blood glucose value in the hyperglycemia sequence segment, and the change rate of the blood glucose value in the hyperglycemia sequence segment is the slope value of the fitting result after the blood glucose value in the hyperglycemia sequence segment is fitted linearly.

[0088] It should be noted that because the risk of developing hyperglycemia-related ovarian cancer is highly correlated with the patient's blood glucose level, embodiments of the present invention improve the credibility and accuracy of subsequent prediction results after processing the patient's blood glucose value sequence to obtain a credible blood glucose sequence. When using the credible blood glucose sequence to predict the patient's subsequent blood glucose changes, the patient's blood glucose changes, i.e., the number of consecutive blood glucose values ​​corresponding to high blood glucose, reflects the duration of the patient's blood glucose level. The numerical level of the blood glucose value reflects the patient's degree of hyperglycemia, and the rate of change reflects the trend of changes in the patient's hyperglycemia. The longer the patient's high blood glucose level lasts, the higher the degree of hyperglycemia, and the more severe the hyperglycemia, the higher the risk of developing the disease. Therefore, based on the above concept, a risk rate is calculated, which is used to describe the patient's risk of developing hyperglycemia-related ovarian cancer. The higher the risk rate, the higher the patient's risk of developing the disease.

[0089] It should be noted that the embodiments of the present invention process and analyze blood glucose data to predict the risk of ovarian cancer related to hyperglycemia based on the patient's blood glucose data. The prediction results eliminate the effects of prediction result deviations caused by environmental variables, and achieve a highly accurate prediction effect of the risk of ovarian cancer related to hyperglycemia.

[0090] At this point, this embodiment is completed.

[0091] It should be noted that the The model is only used to represent negative correlation and constrain the output of the model to be in In the specific implementation, it can be replaced by other models with the same purpose. This embodiment is only based on The model is described as an example without any specific limitation. is the input to the model.

[0092] In a specific embodiment of the present invention, a method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia is provided. In step S001, blood glucose level data and tumor marker concentration data of a patient are obtained. Examples of specific patient blood glucose level data and tumor marker concentration data obtained are shown in Table 1:

[0093] Table 1: Example table of patient blood glucose value data and tumor marker concentration data.

[0094]

[0095] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for constructing a model for predicting the risk of ovarian cancer related to hyperglycemia, characterized in that: The method comprises the following steps: Obtaining patients' blood glucose and tumor marker concentration data; Based on the blood glucose data, the patient's blood glucose level at each physical examination is analyzed in isolation, and the degree of blood glucose abnormality of the patient at each physical examination is calculated by combining the local trend characteristics of the blood glucose level at each physical examination; the overall trend information of blood glucose level changes is extracted from the patient's blood glucose data, and the difference between the patient's blood glucose level and the overall trend information is analyzed. Combined with the patient's blood glucose abnormality, the patient's blood glucose abnormality at each physical examination is evaluated; The correlation between blood glucose levels and tumor marker concentrations at different time periods, combined with the assessment results of blood glucose abnormalities during each physical examination, was used to calculate the predicted reliability of the patient's blood glucose level at each physical examination. The blood glucose value data was then cleaned using the predicted reliability, and the cleaned blood glucose value data was used to predict the patient's risk of ovarian cancer related to high blood glucose. The method of calculating the prediction reliability of the patient's blood glucose level at each physical examination by utilizing the correlation between blood glucose levels and tumor marker concentrations at different time periods and combining the evaluation results of the patient's blood glucose abnormality at each physical examination includes the following specific methods: According to the patient's pathological cycle of hyperglycemia-related ovarian cancer, the collected blood glucose value data and tumor marker concentration data are divided into several periods, and the Pearson correlation coefficient between the blood glucose value data and any tumor marker concentration data in any same period is obtained; Based on the Pearson correlation coefficient between blood glucose values ​​and tumor marker concentrations in the same period and the amount of blood glucose abnormality during physical examinations, the prediction reliability of the patient's blood glucose value at each physical examination was obtained; The method of using the prediction reliability of the blood glucose value to clean the blood glucose value data and using the cleaned blood glucose value data to predict the patient's high blood glucose-related ovarian cancer risk includes the following specific methods: The blood glucose values ​​whose prediction credibility is lower than a preset credibility threshold are deleted from the blood glucose value data, and the deleted blood glucose values ​​are interpolated and supplemented by linear interpolation to obtain a credible blood glucose sequence; The ARIMA model is used to predict the credible blood glucose sequence to obtain a predicted blood glucose sequence. The predicted blood glucose sequence is traversed in reverse order. Several high blood glucose corresponding blood glucose values ​​are obtained based on the blood glucose values ​​in the predicted blood glucose sequence. The sequence segment formed by several consecutive high blood glucose corresponding blood glucose values ​​during the reverse traversal process is recorded as a high blood glucose sequence segment. The risk rate is obtained based on the number of blood glucose values ​​in the high blood glucose sequence segment, the numerical level of the blood glucose values, and the rate of change. The specific calculation method of the risk rate is: ; in, φ is the risk rate, a is the number of blood glucose values ​​in the high blood glucose sequence segment, l is the numerical level of blood glucose value in the high blood glucose sequence segment, k is the rate of change of blood glucose value in the high blood glucose sequence segment, e is a natural constant, sigmoid ( ) is the sigmoid normalization function.

2. The method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia according to claim 1, characterized in that: The blood glucose level of the patient at each physical examination is analyzed in isolation based on the blood glucose level data, and the blood glucose abnormality of the patient at each physical examination is calculated in combination with the local trend characteristics of the blood glucose level at each physical examination. The specific method includes: The blood glucose value data is used as the input of the isolation forest algorithm. The isolation coefficient of each blood glucose value in the blood glucose value data is obtained through the isolation forest algorithm. The isolation coefficient of the blood glucose value is used in combination with the local change trend of the blood glucose value data to analyze and calculate the patient's blood glucose abnormality at each physical examination.

3. The method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia according to claim 2, characterized in that: The method of analyzing and calculating the abnormal blood sugar level of the patient at each physical examination by utilizing the isolated coefficient of the blood sugar level and combining the local variation trend of the blood sugar level data includes the following specific methods: A second-order difference sequence of blood glucose value data is obtained, and the second-order difference value corresponding to each blood glucose value in the second-order difference sequence is obtained. The absolute value of the second-order difference value is used as the local trend parameter of the corresponding blood glucose value. For the patient's blood glucose value at any physical examination, the blood glucose abnormality of the blood glucose value is obtained by combining the isolation coefficient and the local trend parameter of the blood glucose value, wherein the blood glucose abnormality is positively correlated with the isolation coefficient and the local trend parameter.

4. The method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia according to claim 1, characterized in that: The specific method of extracting the overall trend information of blood sugar level changes from the patient's blood sugar level data includes: A two-dimensional rectangular coordinate system is constructed and the patient's blood glucose value data is mapped into the two-dimensional rectangular coordinate system. In the two-dimensional rectangular coordinate system, the horizontal axis represents the patient's physical examination order and the vertical axis represents the patient's blood glucose value under the corresponding physical examination order. The blood glucose value data in the two-dimensional rectangular coordinate system are curve-fitted using the least squares method to obtain the patient's blood glucose value curve. The blood glucose value curve is decomposed using the STL time series decomposition algorithm, and the trend item of the blood glucose value curve is extracted and recorded as the trend curve.

5. The method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia according to claim 1, characterized in that: The analysis of the difference between the patient's blood sugar value and the overall trend information, combined with the patient's blood sugar abnormality, thereby evaluating the patient's blood sugar abnormality at each physical examination, includes the following specific methods: The numerical value of the corresponding data point in the trend curve during each physical examination of the patient is recorded as the blood glucose trend value for the corresponding number of physical examinations. The blood glucose trend abnormality is obtained based on the difference between the patient's blood glucose value at each physical examination and the blood glucose trend value. The blood glucose abnormality amount of the patient at any physical examination is calculated based on the blood glucose trend abnormality and the blood glucose abnormality degree of the patient. The blood glucose abnormality amount is positively correlated with the blood glucose trend abnormality and the blood glucose abnormality degree.

6. The method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia according to claim 5, characterized in that: The specific calculation method of the abnormal blood sugar trend is: ; in, k u For patients u Abnormal blood sugar trend during the first physical examination; b u For patients u Blood sugar level at the time of the next physical examination; B u For patients u Blood glucose trend value at the next physical examination; is the absolute value function; Norm [ ] is the linear normalization function.

7. The method for constructing a model for predicting the risk of ovarian cancer associated with hyperglycemia according to claim 6, characterized in that: The method of obtaining the prediction reliability of the patient's blood glucose level at each physical examination further includes: performing normalization processing on the prediction reliability using a sigmoid function.

Images