Early warning method for thyroid function state and related equipment

Abstract

The invention provides a thyroid function state early warning method and related equipment, and is applied to the technical field of data processing. The method comprises the following steps: acquiring to-be-monitored user information, a target event set matched with the to-be-monitored user information, a training sample set and a preset multivariate continuous biological reference interval model; the training sample set is processed, a target feature library is generated, and the target feature library comprises discontinuous variable data and continuous variable data; training a preset multivariate continuous biological reference interval model based on the target training set and the target verification set to generate a target multivariate continuous biological reference interval model; processing the to-be-detected user information to generate physiological parameter information of the to-be-detected user; processing the target event set matched with the to-be-monitored user information to generate a detection influence factor; and processing the physiological parameter information and the detection influence factor of the to-be-detected user to generate a detection state of the thyroid function of the target user.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an early warning method for thyroid function status and related equipment. Background Art

[0002] Establishing appropriate biological reference intervals for test items, referred to as reference intervals (commonly known as normal ranges), is crucial to ensuring that clinicians correctly diagnose and treat diseases. Whether the test results are "normal" directly depends on the appropriateness of the reference intervals. Establishing accurate reference intervals is an important basis for clinical judgment of whether various diseases exist and whether the treatment is effective. However, currently clinical laboratories use a unified reference interval, or a reference interval established by age or gender using a zoning method. Most of these reference intervals come from the manufacturer's reagent instructions or current industry standards.

[0003] However, the reference intervals widely used in clinical laboratories currently have limitations, which are mainly reflected in the following aspects: Limitations of unified reference intervals: Many laboratories use unified reference intervals provided by manufacturers or based on industry standards, which do not take into account individual differences. Ignoring individual differences: Test indicators such as thyroid hormones will vary with non-continuous variables such as gender, region, race, and continuous variables such as age, season, and BMI. Unified reference intervals cannot reflect these individual differences. Discontinuity of physiological changes: Reference intervals based on the zoning method may not accurately reflect normal physiological changes near the boundaries of age segments. For example, the upper reference limit of creatinine is significantly different between 59-year-old and 60-year-old men.

[0004] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0005] The purpose of this application is to provide an early warning method and related equipment and system for the status of thyroid function, which at least overcomes the problems existing in the prior art to a certain extent, by obtaining user information, target event set, training sample set and preset model, and generating a target feature library after processing. Then divide the training model of the training set and the validation set to obtain a target model that can predict the hormone reference range. Then process the target event set to obtain historical physiological parameter information, determine the influencing factors and weight information to generate the detection influencing factors. Finally, process the user information to be detected, combine the model and the detection influencing factors to generate the detection status of thyroid function, assist doctors in evaluating thyroid function, integrate multiple factors and formulas, provide support for diagnosis and treatment, and optimize parameters to improve accuracy.

[0006] Other features and advantages of the present application will become apparent from the following detailed description, or may be learned in part by the practice of the present invention.

[0007] According to one aspect of the present application, a thyroid function status early warning method and related equipment are provided, including: obtaining user information to be monitored, a target event set matching the user information to be monitored, a training sample set and a preset multivariate continuous biological reference interval model; processing the training sample set to generate a target feature library, wherein the target feature library includes non-continuous variable data and continuous variable data; processing the training sample set based on the non-continuous variable data to generate an initial training set and an initial verification set; processing the initial training set and the initial verification set based on the continuous variable data to generate a target training set and a target verification set, wherein the target training set includes thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, free thyroxine, and the user's gender of other users. , age and month information of sample collection; based on the target training set and the target verification set, the preset multivariate continuous biological reference interval model is trained to generate a target multivariate continuous biological reference interval model; the user information to be detected is processed to generate physiological parameter information of the user to be detected, wherein the physiological parameter information of the user to be detected includes thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, free thyroxine, and the user's gender, age and month information of sample collection; the target event set matching the user information to be monitored is processed to generate a detection influencing factor; based on the target multivariate continuous biological reference interval model, the physiological parameter information of the user to be detected and the detection influencing factor are processed to generate a detection status of the target user's thyroid function.

[0008] Another aspect of the present application is a thyroid function status early warning device, characterized in that it includes: an acquisition module, used to acquire user information to be monitored, a target event set matching the user information to be monitored, a training sample set and a preset multivariate continuous biological reference interval model; a processing module, used to process the training sample set to generate a target feature library, wherein the target feature library includes non-continuous variable data and continuous variable data; based on the non-continuous variable data, the training sample set is processed to generate an initial training set and an initial verification set; based on the continuous variable data, the initial training set and the initial verification set are processed to generate a target training set and a target verification set, wherein the target training set includes thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, free thyroxine of other users , the user's gender, age and month of specimen collection; based on the target training set and the target verification set, the preset multivariate continuous biological reference interval model is trained to generate a target multivariate continuous biological reference interval model; the user information to be detected is processed to generate physiological parameter information of the user to be detected, wherein the physiological parameter information of the user to be detected includes thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, free thyroxine, and the user's gender, age and month of specimen collection; the target event set matching the user information to be monitored is processed to generate a detection influencing factor; based on the target multivariate continuous biological reference interval model, the physiological parameter information of the user to be detected and the detection influencing factor are processed to generate a detection status of the target user's thyroid function.

[0009] According to another aspect of the present application, an electronic device is provided, characterized in that it comprises: a first processor; and a memory for storing executable instructions of the first processor; wherein the first processor is configured to execute the above-mentioned early warning method for thyroid function status by executing the executable instructions.

[0010] According to another aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a second processor, the above-mentioned early warning method of the thyroid function status is implemented.

[0011] According to another aspect of the present application, a computer program product is provided, including a computer program, characterized in that when the computer program is executed by a third processor, the above-mentioned early warning method of thyroid function status is implemented.

[0012] The present application provides a thyroid function status early warning method and related equipment, in which the server obtains the user information to be monitored, the target event set, the training sample set and the preset model. The training sample set is processed to generate a target feature library and divide the training set and the verification set, and the preset model is trained using it. The user information to be monitored and the target event set are processed separately to obtain the user's physiological parameter information and the detection influencing factors. The target multivariate continuous biological reference interval model is used to process this information to obtain the thyroid stimulating hormone equivalent and the multivariate continuous biological prediction interval, and then judge the user's thyroid function status. The continuous reference interval constructed by this method takes into account the continuous changes in gender, age and sampling month, and is more individually responsive. Compared with the traditional reference interval, it can greatly reduce the possibility of false positive and false negative test results, improve the accuracy of clinical diagnosis, meet the requirements of precision medicine, and can customize the treatment plan according to the characteristics of individual patients, providing a more accurate and personalized basis for thyroid function assessment and clinical decision-making.

[0013] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A flow chart showing a method for early warning of thyroid function status provided by an embodiment of the present application;

[0015] Figure 2 A schematic structural diagram of a thyroid function status early warning device provided in one embodiment of the present application is shown. DETAILED DESCRIPTION

[0016] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0017] Combine the following Figure 1 To describe the early warning method of thyroid function status according to the exemplary embodiment of the present application. It should be noted that the following application scenarios are only shown to facilitate understanding of the spirit and principle of the present application, and the embodiments of the present application are not limited in this regard. On the contrary, the embodiments of the present application can be applied to any applicable scenario.

[0018] In one embodiment, the present application also proposes a thyroid function status early warning method and related equipment. Figure 1 The following schematically shows a flow chart of a method for early warning of thyroid function status according to an embodiment of the present application. Figure 1 As shown, the method is applied to a server, comprising:

[0019] S101, obtaining user information to be monitored, a target event set matching the user information to be monitored, a training sample set, and a preset multivariate continuous biological reference interval model.

[0020] In one implementation, suppose there is a user to be monitored, and we collect the following information: Basic information: Name is Xiao Zhang, age is 32 years old, gender is female. Recent situation: No medication has been taken in the past two weeks, no surgery, blood donation, etc., and the user feels healthy and has no visible or palpable goiter. The target event set can be understood as a series of events or condition sets that meet the specific conditions of the user to be monitored. For Xiao Zhang, the target event set may include: Events of the physical examination population that meet the inclusion requirements: Xiao Zhang's situation meets the inclusion criteria of self-conscious health, no acute or chronic diseases, no recent surgery, no blood donation or transfusion, no long-term medication history, and does not meet any exclusion criteria, so she can be included in the event set of the physical examination population that meets the inclusion requirements.

[0021] The training sample set can be obtained from examples of the collection methods mentioned above. For example, samples collected by direct sampling method: Assume that the remaining serum samples of the physical examination population that meets the requirements of the group have been collected as required. Among these samples, some are female samples of the same age as Xiao Zhang, and these samples can be used as part of the training sample set. For example, there are multiple 32-year-old women, and their T3 (triiodothyronine), T4 (thyroxine), FT3 (free triiodothyronine), FT4 (free thyroxine), and TSH (thyroid stimulating hormone) test results, as well as other relevant information that meet the inclusion and exclusion criteria, together constitute the training samples for Xiao Zhang's age group and gender. Samples collected by indirect sampling method: The physical examination population data collected from the hospital LIS system within 3 years may also have individual data of the same age and gender as Xiao Zhang after the group screening, which can also be added to the training sample set.

[0022] The key to constructing the preset continuous biological reference interval model is to use surface fitting or regression to establish the 97.5 and 2.5 percentile curves or surfaces of thyroid-related hormone data that change with the dual variables of age and month for people of different genders, so as to characterize the biological reference interval that conforms to normal physiological fluctuations. Each point on the surface corresponds to the upper limit (i.e., 97.5 percentile) or lower limit (i.e., 2.5 percentile) of the reference interval for people of specific gender, age, and month. In addition, in this process, this solution implements correction processing for the corresponding specific reference interval (i.e., 97.5 and 2.5 percentiles), and then compares the user's original test data with the reference interval after correction processing to determine whether there is any abnormality in the test result.

[0023] Take Xiao Zhang for example. This continuous biological reference interval model is established based on the previously collected data and the analysis methods used. For example, when constructing the prediction model for thyroid-related hormones, multiple factors such as gender, age, and month were fully considered. The model can carry out parameter estimation and optimization based on the acquired training sample set so that it can better meet the thyroid hormone level prediction needs of different populations. After obtaining Xiao Zhang's relevant information, her age, gender, current month and other information can be substituted into the model to derive a predicted range for her thyroid hormone levels. Finally, by comparing the actual test results of her thyroid-related hormones with this predicted range, it can be determined whether her test results are within the reference range.

[0024] S102: Process the training sample set to generate a target feature library.

[0025] In one embodiment, a feature extraction operation is performed on the training sample set to construct a user physiological parameter database. The database covers the user's gender, age, month, and information such as thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, and free thyroxine. Assume that we have collected a certain number of training sample sets, which contain specific information of many users. We perform feature extraction processing on these samples to generate a user physiological parameter database. For example, we have the following training samples: User A: male, 33 years old, tested in February, thyroid stimulating hormone value is a specific value 1, triiodothyronine value is a specific value 2, thyroxine value is a specific value 3, free triiodothyronine value is a specific value 4, and free thyroxine value is a specific value 5. User B: female, 48 years old, tested in September, and corresponding hormones also have specific values. User C: male, 51 years old, tested in November, also has various hormone values. After feature extraction, the user physiological parameter database contains the gender, age, month, and various hormone information of these users.

[0026] The user's gender, age and month information are processed separately to generate target feature data. Among them, the target feature data includes various parameters of the smoothing function, and these parameters are used to reflect the rationality of the fit of the multivariate continuous biological reference interval model. For gender, it can be simply coded as 0 (male) and 1 (female), age keeps its specific value, and the month can be represented by numbers 1 to 12. It is assumed that through previous research and data analysis, the degrees of freedom of the smoothing function under different combinations of gender, age and month are determined. For example, for the case where the male age is 30 and the month collected is February, the degrees of freedom of the smoothing function of the location parameter (mu) is df_mu1, and the degrees of freedom of the smoothing function of the skewness parameter (na) is df_na1. For the case where the female age is 45 and the month collected is September, the degrees of freedom of the smoothing function of mu is df_mu2, the degrees of freedom of the smoothing function of na is df_na2, and so on. These smoothing function parameters are used to characterize the rationality of the fit of the multivariate continuous biological reference interval model. If the degrees of freedom are selected appropriately, the model can fit the data better and improve the accuracy of the prediction.

[0027] Based on the target feature data, the user physiological parameter database is processed to construct the target feature library. The target feature library contains non-continuous variable data and continuous variable data. Non-continuous variable data mainly refers to gender, because gender has only two discrete values ​​(male and female). Continuous variable data include age and month. Age is a continuous variable, ranging from 18 to 80 years old. Although month is a discrete value from 1 to 12 in form, it can also be regarded as a continuous variable in some analyses, such as considering the continuous influence of seasons. At the same time, hormone data such as thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, and free thyroxine are also included. These data also show different distribution characteristics under different combinations of gender, age, and month. Based on the target feature data, we process the user physiological parameter database and divide it into non-continuous variable data (gender) and continuous variable data (age, month, and various hormone values) to generate the target feature library. Such a library can better provide data support for the multivariate continuous biological reference interval model for more accurate analysis and prediction.

[0028] S103: Process the training sample set based on the non-continuous variable data to generate an initial training set and an initial validation set.

[0029] In one implementation, it is assumed that our training sample set contains information of 500 users, which covers data such as gender, age, month, and corresponding thyroid hormone test results such as thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, and free thyroxine. The non-continuous variable data here is gender, which is divided into male and female. First, determine the division ratio, such as dividing the training sample set into an initial training set and an initial validation set at a ratio of 7:3. For the data of male users, if there are 200 samples of male users, then select 70% (i.e., 140 samples) of them and put them into the initial training set, and the remaining 60 samples are put into the initial validation set. For the data of female users, if there are 300 samples of female users, also select 70% (i.e., 210 samples) of them and put them into the initial training set, and the remaining 90 samples are put into the initial validation set.

[0030] First, the data of male users and female users are regarded as two layers respectively. For the male user layer, a certain number of samples are randomly selected and put into the initial training set and the initial validation set. For example, 140 samples of 200 male users are randomly selected and put into the initial training set, and 40 samples of the remaining 60 are randomly selected as a backup set (which can be considered for use in subsequent adjustments), and the remaining 20 are put into the initial validation set. For the female user layer, similar operations are performed. 210 samples of 300 female users are randomly selected and put into the initial training set, and 60 samples of the remaining 90 are randomly selected as a backup set, and the remaining 30 are put into the initial validation set. Through such processing based on non-continuous variable data (gender), we obtain the initial training set and the initial validation set. The data in these sets will be used in the training and validation process of the multivariate continuous biological reference interval model in the future to better analyze and predict thyroid hormone levels under different combinations of gender, age and month.

[0031] S104: Process the initial training set and the initial validation set based on the continuous variable data to generate a target training set and a target validation set.

[0032] In one embodiment, it is assumed that we have obtained the initial training set and the initial validation set by dividing based on non-continuous variable data (such as gender). Now, in order to better establish a multivariate continuous biological reference interval model, we further process based on continuous variable data (ie, age and month) to generate a target training set and a target validation set. In terms of age, only specific continuous years are considered, for example, the age range is set to 18 to 79 years old. Months are simply represented by numbers 1 to 12.

[0033] For each sample in the initial training set, we carefully assign it according to its age and month information. For example, if there is a sample user who is 35 years old and the test time is in July, then it is directly labeled according to this specific age and month. For example, another sample user is 50 years old and the test time is in November, and the exact age and month are also labeled. Assuming that we are studying indicators related to thyroid disease, users with different age and month combinations may belong to the same disease type (for example, they are all undergoing thyroid function assessment), but due to differences in age and month, they will show different specific characteristics. For the initial validation set, we also perform similar processing operations. Ensure that the samples in the validation set can also be correctly labeled with age and month information, so that in the subsequent model validation stage, the performance of the model under different age and month combinations can be more accurately evaluated. Through this processing method, we finally generated the target training set and target validation set, thereby providing a more targeted data basis for establishing a multivariate continuous biological reference interval model.

[0034] S105, training the preset multivariate continuous biological reference interval model based on the target training set and the target verification set to generate a target multivariate continuous biological reference interval model.

[0035] In one embodiment, in the target training set and the target validation set, multiple continuous variable data cover the age information of different users, where the age range can range from 18 to 80 years old. It also includes the month information, which is clearly indicated by numbers 1 to 12. For example, user G is 38 years old and the test time is in May; user H is 53 years old and the test time is in September. Non-continuous variable data, that is, the gender information of different users, can be simply encoded as 0 for male and 1 for female. For example, user G is male and user H is female. The user's physiological parameter information includes thyroid stimulating hormone (TSH), triiodothyronine (T3), thyroxine (T4), free triiodothyronine (FT3) and free thyroxine (FT4). Assume that there is information of many users in the target training set and the target validation set, and the specific situations are as follows: in terms of T3, the T3 value of user G is 1.4nmol / L, and the T3 value of user H is 1.5nmol / L; in terms of T4, the T4 value of user G is 100nmol / L, and the T4 value of user H is 98nmol / L; in terms of FT3, the FT3 value of user G is 4.4pmol / L, and the FT3 value of user H is 4.5pmol / L; in terms of FT4, the FT4 value of user G is 14.2pmol / L, and the FT4 value of user H is 14.8pmol / L; in terms of TSH, the TSH value of user G is 2.2mIU / L, and the TSH value of user H is 2.6mIU / L.

[0036] The user's physiological parameter information is processed based on multiple continuous variable data and discontinuous variable data to generate a target biological reference interval. The target biological reference interval is determined based on the gender, age, month, and thyroid stimulating hormone value, triiodothyronine value, thyroxine value, free triiodothyronine value, and free thyroxine value of different users. First, all users are divided into male and female groups according to gender. For the male group, the distribution of T3, T4, FT3, FT4, and TSH values ​​is analyzed based on their specific age and month. For example, a male is 33 years old. In August, through statistical analysis, it is determined that the reference interval of his T3 is a specific continuous range, the reference interval of T4 is also continuous, the reference interval of FT3 is also continuous, the reference interval of FT4 is also presented as a continuous range, and the reference interval of TSH is also a continuous interval. Similar analysis operations are performed for the female group. For example, a woman is 46 years old. In October, after analysis, her T4 reference interval is a continuous range, T3 reference interval is also continuous, FT3 reference interval is a continuous interval, FT4 reference interval is continuous, and TSH reference interval is also continuous. In the end, a series of target biological reference intervals based on the specific ages and months of different individuals can be obtained, such as the T3, T4, FT3, FT4 and TSH reference intervals of a 39-year-old man in June, and the T3, T4, FT3, FT4 and TSH reference intervals of a 47-year-old woman in July. These reference intervals are obtained by constructing continuous reference intervals, not based on segmented specific age groups.

[0037] Generate a target multivariate continuous biological reference interval model based on the target biological reference interval. The key to building this model is to use surface fitting or regression to model the continuous reference interval. For people of different genders, according to the changes in thyroid-related hormone data with the dual variables of age and month, establish 97.5 and 2.5 percentile curves or surfaces that can represent the biological reference interval that conforms to normal physiological fluctuations. On the constructed surface, each point corresponds to the upper limit (97.5 percentile) or lower limit (2.5 percentile) of the reference interval for people of specific gender, age and month. Use a suitable statistical model or machine learning algorithm to build this model, with gender, age, and month as input variables and reference intervals of T3, T4, FT3, FT4 and TSH as output. For example, a multivariate linear regression model can be established, with the input part being gender (represented by coding as 0 or 1), specific age values, and specific month values, and the output being the predicted reference interval of T3. By training the data in the target biological reference interval and continuously adjusting the parameters of the model, the model can accurately predict the corresponding reference intervals of T3, T4, FT3, FT4 and TSH based on the specific gender, age and month information input. The model obtained after a series of training and optimization operations is the target multivariate continuous biological reference interval model we need.

[0038] When new user information (including specific gender, age, and month) is input into this trained target multivariate continuous biological reference interval model, the model can quickly give the user's reference intervals for physiological parameters such as T3, T4, FT3, FT4, and TSH, thereby effectively assisting doctors in diagnosing diseases and assessing the user's health status.

[0039] S106: Process the information of the user to be detected to generate physiological parameter information of the user to be detected.

[0040] In one embodiment, the physiological parameter information of the user to be detected includes thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, free thyroxine, as well as the user's gender, age and current month information. Suppose there is a user to be detected, Xiao Zhao. Xiao Zhao's basic information is determined by questioning and inspection: gender is female, age is 28 years old, and the month of collection is April. Xiao Zhao is subjected to medical examinations such as blood tests, and the following parameter information is obtained: Triiodothyronine (T3): The detection value is 1.5nmol / L. Thyroxine (T4): The detection value is 98nmol / L. Free triiodothyronine (FT3): The detection value is 4.4pmol / L. Free thyroxine (FT4): The detection value is 14.8pmol / L.

[0041] Based on the above information, the physiological parameter information of the user Xiao Zhao to be tested includes triiodothyronine (1.5nmol / L), thyroxine (98nmol / L), free triiodothyronine (4.4pmol / L), free thyroxine (14.8pmol / L), and the user's gender (female), age (28 years old) and current month (April). This information will be used for matching and analysis with the target multivariate continuous biological reference interval model in the future to determine whether Xiao Zhao's thyroid function is within the normal range.

[0042] S107: Process the target event set that matches the user information to be monitored to generate a detection impact factor.

[0043] In one embodiment, the target event set that matches the user information to be monitored is processed to generate the type information of the target event. The type information of the target event is used to characterize the impact factor of the target event on the physiological parameters of the user to be monitored. Assuming that the user to be monitored is Xiao Sun, the target event set covers Xiao Sun's recent long-distance hiking (target event 1) and a period of continuous intense learning (target event 2). The type information of target event 1 can be characterized as "high-intensity exercise", and this type of event may have an impact on Xiao Sun's physiological parameters, such as metabolic rate, thyroid hormone levels, etc. The type information of target event 2 can be characterized as "mental stress", and this type of event may affect the endocrine system, and then have an effect on Xiao Sun's thyroid and other physiological parameters. The type information of the target event is processed to generate the historical physiological parameter information of the user to be monitored. The historical physiological parameter information of the user to be monitored is used to characterize the physiological parameter information of different users when they experience this type of target event.

[0044] For the "high-intensity exercise" event: recall Xiao Sun's physiological parameter information before and after the long-distance hike. Assume that the triiodothyronine (T3) detected at that time was 1.6nmol / L, thyroxine (T4) was 102nmol / L, free triiodothyronine (FT3) was 4.6pmol / L, and free thyroxine (FT4) was 15pmol / L. These data constitute Xiao Sun's historical physiological parameter information when he experienced the "high-intensity exercise" type target event. For the "mental stress" event: recall Xiao Sun's physiological parameters during the intense study period, such as T3 of 1.55nmol / L, T4 of 100nmol / L, FT3 of 4.55pmol / L, and FT4 of 14.8pmol / L. This is Xiao Sun's historical physiological parameter information when he experienced the "mental stress" type target event.

[0045] The historical physiological parameter information of the monitored user is processed to generate physiological parameter influencing factors and weight information corresponding to the physiological parameter influencing factors. For the "high-intensity exercise" event, by comparing Xiao Sun's usual physiological parameters with the parameters during long-distance hiking, it is found that T3, T4, FT3, and FT4 have all changed to a certain extent. It can be determined that "high-intensity exercise" is a physiological parameter influencing factor. After statistical analysis and model calculation, it is assumed that the weight assigned to this influencing factor is 0.25. For the "mental stress" event, the same comparison found that the parameters also changed, and "mental stress" was determined to be another physiological parameter influencing factor. It is assumed that its weight is calculated to be 0.2.

[0046] The physiological parameter influencing factors and the weight information corresponding to the physiological parameter influencing factors are processed to generate the detection influencing factors. Among them, the detection influencing factors are used to adjust the physiological parameter influencing factors of the target user. The physiological parameter influencing factors such as "high-intensity exercise" and "mental stress" and their corresponding weight information are processed to obtain the detection influencing factors. For example, if Xiao Sun is now going to have a thyroid function test, the detection influencing factors can be used to adjust the physiological parameter influencing factors of the target user (Xiao Sun). Assuming that Xiao Sun is now going to have a thyroid function test, according to the detection influencing factors, if the detected T3 original value is 1.7nmol / L, considering the influence of "high-intensity exercise" and "mental stress", after adjustment, a T3 value that more accurately reflects Xiao Sun's current physical state may be obtained, such as 1.68nmol / L. Through the above steps, the impact of the target event on the user's physiological parameters can be better understood, and the test results can be analyzed and adjusted more accurately.

[0047] In another embodiment, the historical physiological parameter information of the monitored user is processed to generate the first impact value of the physiological parameter of the monitored user. Assume that the monitored user is Xiao Zhou. Let's review Xiao Zhou's historical physiological parameter information. Collect historical data: Assume that Xiao Zhou has undergone multiple physical examinations in the past period of time, and recorded thyroid-related physiological parameters at different time points, including triiodothyronine (T3), thyroxine (T4), free triiodothyronine (FT3) and free thyroxine (FT4). At the same time, some specific event information at that time was also recorded, such as whether he had experienced long-term overtime work, whether he was in a specific month, etc. For example, in May, Xiao Zhou experienced a long period of overtime work. At that time, the T3 value detected was 1.7nmol / L, the T4 value was 108nmol / L, the FT3 value was 4.6pmol / L, and the FT4 value was 15.2pmol / L. In August, Xiao Zhou's work was relatively easy, and the parameters detected were T3 value of 1.5nmol / L, T4 value of 103nmol / L, FT3 value of 4.4pmol / L, and FT4 value of 14.8pmol / L. By comparing the changes in physiological parameters under different states, it can be determined that the event of long-term overtime work has an impact on Xiao Zhou's thyroid-related physiological parameters. For example, compared with the relaxed state, the values ​​of T3, T4, FT3 and FT4 during the long-term overtime period have increased to a certain extent. After statistical analysis and calculation, the first impact value of the event of long-term overtime work on Xiao Zhou's physiological parameters can be obtained. Assuming that the first impact value of long-term overtime work on Xiao Zhou's T3 is 0.2nmol / L, it means that long-term overtime work has increased Xiao Zhou's T3 value by 0.2nmol / L. Similarly, it can be calculated that the first impact values ​​of T4, FT3 and FT4 are 5nmol / L, 0.2pmol / L and 0.4pmol / L respectively. Correspondingly, in this event, since the value of T3 increased by 0.2nmol / L, the upper and lower limits of the prediction interval of triiodothyronine will also be adjusted accordingly.

[0048] Assume that without the influence of a specific event, we have obtained a user's initial value of triiodothyronine (T3) of 3.0nmol / L, and the standard deviation of the user's triiodothyronine value

[0049] T3 std =0.5nmol / L.

[0050] T3 upper =T3 predicted +1.96T3 std =3.0+1.96×0.98=3.98nmol / L.

[0051] This means that at a given confidence level (here assumed to be 95%), the upper limit of triiodothyronine for this user is expected to be 3.98 nmol / L according to the calculation model.

[0052] T3 lower =T3 predicted -1.96T3 std =3.0-1.96×0.5=2.02nmol / L.

[0053] Therefore, the lower limit of triiodothyronine for this user is expected to be 2.02nmol / L.

[0054] Now suppose that an "extended overtime" event occurs, causing the T3 value to increase by 0.2nmol / L. At this time, the T3 value becomes 1.7nmol / L.

[0055] Then the adjusted T3 interval is:

[0056] T3 upper =T3 predicted +1.96T3 std =3.2+1.96×0.98=4.18nmol / L.

[0057] T3 lower =T3 predicted -1.96T3 std =3.2-1.96×0.5=2.22nmol / L.

[0058] This indicates that the "long overtime" event has affected the predicted interval of triiodothyronine for this user, causing its overall range to shift upward. In practical applications, doctors can re-evaluate the user's thyroid function status based on this adjusted interval. If the actual test value is within the adjusted interval, it may mean that although overtime has a certain impact on thyroid function, it is still within a relatively normal range; if it exceeds the adjusted interval, it is necessary to pay closer attention to whether the thyroid function is abnormal, and further check other related indicators or make a comprehensive judgment based on clinical symptoms.

[0059] For example, when evaluating a patient's thyroid function, the doctor learns that the patient has recently engaged in high-intensity exercise, which is an event that may increase the T3 value. If the patient's T3 value is 1.68nmol / L, the doctor can judge the patient's thyroid function status based on the adjusted T3 prediction interval. If this value is within the adjusted interval, it may mean that the patient's thyroid function is still in a relatively normal state under the influence of the current event; if it is outside the interval, further examination may be required to determine whether there is thyroid disease or other abnormalities.

[0060] Obtain the historical physiological parameter information of other users, process the historical physiological parameter information of other users, and generate the second impact value of the physiological parameters of other users. Assume that we have the historical physiological parameter information and event records of multiple other users. For example, the physical examination data of user Xiao Wu at different time points and the corresponding event information. Xiao Wu participated in a long-distance cycling event in April. At that time, the T3 value detected was 1.8nmol / L, the T4 value was 112nmol / L, the FT3 value was 4.7pmol / L, and the FT4 value was 15.8pmol / L. In July, when there was no special event, the parameters detected were T3 value of 1.6nmol / L, T4 value of 105nmol / L, FT3 value of 4.5pmol / L, and FT4 value of 15pmol / L. For Xiao Wu, the event of participating in a long-distance cycling event had an impact on his thyroid-related physiological parameters. By comparing the parameter changes under different states, the second impact value of the event of participating in a long-distance cycling event on Xiao Wu's physiological parameters can be calculated. Assume that participating in a long-distance cycling activity has a second impact value of 0.2nmol / L on Xiao Wu's T3, a second impact value of 7nmol / L on T4, a second impact value of 0.2pmol / L on FT3, and a second impact value of 0.8pmol / L on FT4.

[0061] The first influencing value of the physiological parameters of the monitored user and the second influencing values ​​of the physiological parameters of other users are processed to generate physiological parameter influencing factors and weight information corresponding to the physiological parameter influencing factors. The first influencing value of Xiao Zhou and the second influencing values ​​of other users such as Xiao Wu are comprehensively analyzed. For example, it is found that both long-term overtime and long-distance cycling activities have a certain degree of positive impact on T3, T4, FT3 and FT4. Similar events such as "long-term overtime" and "participation in strenuous sports activities" can be determined as physiological parameter influencing factors. By performing statistical analysis and model calculation on the data of multiple users, the weight information corresponding to each influencing factor can be determined. Assume that after calculation, the weight of the influencing factor "long-term overtime" on T3 is 0.22, the weight on T4 is 0.18, the weight on FT3 is 0.2, and the weight on FT4 is 0.15. The weight of the influencing factor "participation in strenuous sports activities" on T3 is 0.25, the weight on T4 is 0.22, the weight on FT3 is 0.23, and the weight on FT4 is 0.18. Through the above steps, the physiological parameter influencing factors and the corresponding weight information can be determined, providing a basis for more accurately evaluating the user's physiological state and predicting changes in physiological parameters.

[0062] In another implementation, assuming the user to be monitored is Xiao Zhang, we collect Xiao Zhang's thyroid-related physiological parameter data at different time points to calculate the first impact value.

[0063] Determine the parameter value: After a period of testing, Xiao Zhang's triiodothyronine (T3) values ​​were 1.5nmol / L, 1.6nmol / L, and 1.7nmol / L, so Xiao Zhang's average T3 value is

[0064] =(1.5+1.6+1.7) / 3=1.6nmol / L. Assume that the standard deviation of these three T3 values ​​is T3 std =0.1nmol / L.

[0065] Similarly, assuming that Xiao Zhang's average thyroid hormone (T4) is T4 avg =105nmol / L, standard deviation T4 std =5nmol / L; free triiodothyronine (FT3) average value FT3 avg =4.5pmol / L, standard deviation FT3 std =0.2pmol / L; free thyroxine (FT4) average FT4 avg =15pmol / L, standard deviation FT4 std =0.5pmol / L. Assume that the weight coefficients n1=0.2, n2=0.25, n3=0.15, and n4=0.1. The method also includes a calculation formula for obtaining the first impact value, which is:

[0066]

[0067] Among them, T3 avg 、T4 avg , FT3 avg , FT4 avg are the average values ​​of triiodothyronine, thyroxine, free triiodothyronine and free thyroxine of the monitored users, respectively. std 、T4 std , FT3 std , FT4 std are the standard deviations of the corresponding parameters respectively; n1, n2, n3, n4 are weight coefficients, which can be set according to the importance of each parameter, and n1+n2+n3+n4=1.

[0068] Substituting the above parameters into the calculation formula of the first influence value, I1=0.4+0.25+0.225+0.1=0.975.

[0069] Assume that data of multiple other users are used to calculate the second influence value.

[0070] Determine parameter value: Average T3 values ​​of other users

[0071] T3 other，avg=1.65nmol / L, standard deviation T3 other，std =0.12nmol / L. Thyroxine (T4) average value T4 other，avg =108nmol / L, standard deviation T4 other，std =6nmol / L. Free triiodothyronine (FT3) average FT3 other，avg =4.6pmol / L, standard deviation FT3 other，std =0.25pmol / L. Free thyroxine (FT4) average FT4 other，avg =15.2pmol / L, standard deviation

[0072] FT4 other，std =0.6 pmol / L. Weight coefficients p1 = 0.18, p2 = 0.22, p3 = 0.12, p4 = 0.15.

[0073] The method also includes a calculation formula for obtaining a second impact value, the calculation formula being:

[0074]

[0075] Among them, T3 other，avg 、T4 other，avg , FT3 other，avg , FT4 other，avg are the average values ​​of other users on triiodothyronine, thyroxine, free triiodothyronine and free thyroxine; T3 other，std 、T4 other，std , FT3 other，std , FT4 other，std are the standard deviations of the corresponding parameters respectively; p1, p2, p3, p4 are weight coefficients, which can be set according to the importance of each parameter, and p1+p2+p3+p4=1.

[0076] Substitute the above parameters into the calculation formula of the second impact value, I2 = 0.225 + 0.0733 +

[0077] 0.096+0.075=0.4693.

[0078] S108, processing the physiological parameter information of the user to be detected and the detection influencing factors based on the target multivariate continuous biological reference interval model to generate a detection status of the target user's thyroid function.

[0079] In one embodiment, it is assumed that for a 30-year-old woman in April, the reference interval of T3 determined by the model is 1.2nmol / L-1.8nmol / L (here it is assumed that the interval is a range determined by the 97.5 and 2.5 percentiles calculated based on a large amount of data). The same is true for other hormone indicators, for example, the reference interval of T4 is 90nmol / L-120nmol / L, the reference interval of FT3 is 4pmol / L-6pmol / L, and the reference interval of FT4 is 12pmol / L-22pmol / L.

[0080] Assume that Xiao Li is a 42-year-old female who was tested in July. Xiao Li's thyroid stimulating hormone value, triiodothyronine value, thyroxine value, free triiodothyronine value and free thyroxine value are 2.3mIU / L, 1.7nmol / L, 105nmol / L, 4.5pmol / L and 15pmol / L respectively. Without considering the influencing factors of the test, the target multivariate continuous biological reference interval model is used to calculate that the thyroid stimulating hormone prediction interval of a 42-year-old female in July is 2.0mIU / L-3.0mIU / L, the T3 prediction interval is 1.5nmol / L-1.7nmol / L, the T4 prediction interval is 100nmol / L-110nmol / L, the FT3 prediction interval is 4.2pmol / L-4.8pmol / L, and the FT4 prediction interval is 14pmol / L-16pmol / L. Assume that Xiao Li's test influencing factors include recent changes in living environment (newly arrived in a high altitude area) and changes in eating habits (increased intake of iodine-rich seafood). The high altitude environment causes the upper limit of the TSH prediction range of a 42-year-old female in July to increase by 0.2mIU / L and the lower limit to increase by 0.1mIU / L, becoming

[0081] The upper limit of the T3 prediction interval increased by 0.05nmol / L and the lower limit increased by 0.03nmol / L, becoming 1.53nmol / L-1.75nmol / L; the upper limit of the FT3 prediction interval increased by 0.02pmol / L and the lower limit increased by 0.01pmol / L, becoming 4.21pmol / L-4.82pmol / L. The increase in iodine intake increased the upper limit of the T4 prediction interval by 1nmol / L and the lower limit by 0.5nmol / L, becoming 100.5nmol / L-111nmol / L; the upper limit of the FT4 prediction interval increased by 0.3pmol / L and the lower limit increased by 0.2pmol / L, becoming 14.2pmol / L-16.3pmol / L. The target thyroid status prediction interval was generated for a 42-year-old woman in July with recent changes in living environment (newly arrived at a high altitude area) and dietary habits (increased intake of iodine-rich seafood). By comparing Xiao Li's thyroid stimulating hormone value, triiodothyronine value, thyroxine value, free triiodothyronine value and free thyroxine value with the target thyroid status prediction interval, it was found that Xiao Li's various indicators were within the normal range, so Xiao Li's thyroid function test status was "normal test status".

[0082] Assume that Xiao Zhang is a 35-year-old male and was tested in September. Xiao Zhang's thyroid stimulating hormone value, triiodothyronine value, thyroxine value, free triiodothyronine value and free thyroxine value were 4.2mIU / L, 2.05nmol / L, 124nmol / L, 6.1pmol / L and 21.5pmol / L respectively.

[0083] After model calculation and consideration of test influencing factors, for 35-year-old men in September, the preset thyroid status prediction interval is TSH normal range is 0.4mIU / L-4.1mIU / L, T3 normal range is 1.3nmol / L-1.8nmol / L, T4 normal range is 96nmol / L-116nmol / L, FT3 normal range is 4.2pmol / L-5.2pmol / L, FT4 normal range is 13pmol / L-17pmol / L. By comparison, it was found that Xiao Zhang's TSH, T4, FT3, and FT4 values ​​were all beyond the normal range. According to the degree of excess, a thyroid function test status of "obviously abnormal, need to see a doctor as soon as possible" can be generated.

[0084] In another embodiment, the target multivariate continuous biological reference interval model also includes a calculation formula for obtaining a prediction interval of triiodothyronine, and the calculation formula is:

[0085] T3 predicted =a i +b i(xt i )+c i (xt i ) 2 +d i (xt i ) 3 +e1×age+f1×month

[0086] +g1×gender;

[0087] T3 upper =T3 predicted +1.96T3 std ; T3 lower =T3 predicted -1.96T3 std ;

[0088] Among them, T3 predicted stands for triiodothyronine, T3 upper represents the upper limit of the prediction interval of triiodothyronine, T3 lower represents the lower limit of the prediction interval of triiodothyronine, T3 std represents the standard deviation of triiodothyronine values, a i is a reference value, b i 、c i ,d i Determine the linear, quadratic and cubic trends of triiodothyronine values ​​with the change of variable x; t i is a specific parameter or reference value related to triiodothyronine; e1, f1, and g1 represent the influence coefficients of age, month, and gender on the triiodothyronine value, respectively.

[0089] Suppose we want to predict the prediction interval of triiodothyronine (T3) for a user whose age is 30, the current month is July, and the gender is male.

[0090] Determine the formula parameters: Assume a i =1.5, as a benchmark value. Assume b i =0.1, c i =-0.01, d i =0.001, respectively, determine the trend of T3 prediction value with variable x (assumed to be a variable related to thyroid function but not clearly defined). Assuming t i =10, as a specific reference value related to triiodothyronine. Assuming e1 = 0.05, it indicates the influence coefficient of age on T3 prediction value; f1 = 0.03, it indicates the influence coefficient of month on T3 prediction value; g1 = 0.08, it indicates the influence coefficient of gender on T3 prediction value, T3std =0.23.

[0091] Calculation process: Assume variable x=12, and male is coded as 0.

[0092] First calculate the linear, quadratic, and cubic terms: b i (xt i )=0.1×(12-10)=0.2.

[0093] c i (xt i ) 2 =-0.01×(12-10) 2 =-0.04.d i (xt i ) 3 =0.001×(12-10) 3 =0.008.

[0094] The effects of age, month, and sex are then calculated:

[0095] e1×age=0.05×30=1.5;f1×month=0.03×7=0.21;

[0096] g1×gender=0.08×0=0 (because he is a male).

[0097] Then calculate the predicted T3 value:

[0098] T3 predicted =1.5+0.2-0.04+0.008+1.5+0.21+0=3.378.

[0099] Finally, calculate the prediction interval for T3:

[0100] T3 upper =T3 predicted +1.96T3 std =3.378+0.4508=3.3288;

[0101] T3 lower =T3 predicted -1.96T3 std =3.378-0.4508=2.9272.

[0102] The target multivariate continuous biological reference interval model also includes a calculation formula for obtaining the prediction interval of thyroxine, which is:

[0103]

[0104] Among them, t iis a specific parameter or reference value related to thyroid hormone; k is a fitting parameter that affects the predicted value of thyroid hormone; e2, f2, g2 represent the influence coefficients of age, month and gender on the predicted value of thyroid hormone respectively; B i,k-1 (x) and B i+1,k-1 (x) jointly determine a component of the thyroid hormone prediction value, reflecting the complex relationship between other variables and thyroid hormone. Their values ​​will change with the input variable x and the parameters i and k, showing nonlinear, cyclical or other complex change trends. For example, under different age, gender, month combinations, or for different physiological states, B i,k-1 (x) and B i+1,k-1 The value of (x) will vary.

[0105] The fitting parameter k determines the complexity of the function. When the k value is large, the function may be more complex and can fit the complex relationship between thyroid hormone values ​​and other variables more finely. When the k value is small, the function is relatively simple and may be more suitable for situations where the data relationship is relatively simple. The input variable x represents a physiological indicator, environmental factor or other influencing factor related to thyroid hormone. For example, it can be the level of other hormones, specific lifestyle indicators (such as exercise volume, the content of certain key ingredients in the diet structure, etc.), the quantitative value of seasonal factors, etc. B i,k-1 (x) and B i+1,k-1 (x) together constitute an important part of the thyroid hormone value. They reflect the complex relationship between other variables and thyroid hormone under specific conditions. Specifically, these two parameters participate in the calculation of the predicted value of thyroid hormone through combined operations with input variables and other parameters (such as age, month, gender, etc., influence coefficients e2, f2, g2), and further determine the upper and lower limits of the thyroid hormone prediction interval. The specific values ​​and variation patterns of these two parameters are obtained through the analysis and fitting of a large amount of thyroid hormone-related data. The purpose is to more accurately predict the range of thyroid hormone values ​​under different conditions and provide more valuable reference for clinical diagnosis and treatment.

[0106] Let's take the predicted thyroid hormone (T4) range of a user who is 30 years old, the current month is July, and the gender is male as an example. Determine the formula parameters:

[0107] Assume t i=8, as a specific reference value related to thyroxine. Assume k = 3, as a fitting parameter affecting the predicted value of thyroxine. Assume e2 = 0.06, indicating the influence coefficient of age on the predicted value of T4; f2 = 0.04, indicating the influence coefficient of month on the predicted value of T4; g2 = 0.07, indicating the influence coefficient of gender on the predicted value of T4. std =0.2.

[0108] Assume x = 5, and assume that B i,k-1 (x) and B i+1,k-1 The values ​​of (x) are 2.5 and 3.2 respectively (here we assume that these two function values ​​are specific values ​​obtained through other complex calculations).

[0109] Begin by calculating the first two terms in the formula:

[0110]

[0111] Then calculate the effects of age, month and gender: e2×age=0.06×30=1.8. f2×month=0.04×7=0.28. g2×gender=0.07×0=0 (because it is a male).

[0112] Then calculate the predicted value of T4: T4 predicted =5.83-6.4+1.8+0.28+0=1.51.

[0113] Finally, calculate the T4 prediction interval:

[0114] T4 upper =T4 predicted +1.96T4 std =1.51+0.392=1.902;

[0115] T4 lower =T4 predicted -1.96T4 std =1.51-0.392=1.118.

[0116] The target multivariate continuous biological reference interval model also includes a calculation formula for obtaining the prediction interval of free triiodothyronine, which is:

[0117]

[0118] FT3 upper =FT3 predicted +1.96FT3 std ;

[0119] FT3 lower =FT3 predicted-1.96FT3 std ;

[0120] Among them, FT3 predicted FT3 stands for free triiodothyronine upper represents the upper limit of the prediction interval for free triiodothyronine, FT3 lower Represents the lower limit of the prediction interval for free triiodothyronine, FT3 std represents the standard deviation of triiodothyronine values, and P0 is a constant term, which represents a baseline free triiodothyronine predicted value without considering factors such as age, month, and gender;

[0121] P1, P2, P3, and P4 are coefficients that determine the degree to which different factors affect the predicted value of free triiodothyronine;

[0122] The introduction of makes the effect of age on free triiodothyronine present a periodic or fluctuating trend. For example, with the change of age, the level of free triiodothyronine may fluctuate periodically. Assuming that the user is 30 years old, this value will change within a certain range according to the characteristics of the trigonometric function, thus affecting the predicted value.

[0123] The introduction of makes the effect of months on free triiodothyronine present another periodic change trend. Assuming that the current month is July, then this value will also change according to the characteristics of the cosine function, reflecting the impact of different months on the level of free triiodothyronine.

[0124] Gender: Assume that male is coded as 0 and female is coded as 1. If the user is male, the value of this part is 0; if the user is female, assuming P3 = 0.1, the value of this part is 0.1.

[0125] month×gender: Assuming the user is female and the current month is July, the value of this part is P4×7×1. If P4=0.08, the value of this part is 0.56.

[0126] Suppose we want to predict the free triiodothyronine range for a 30-year-old female user whose current month is July.

[0127] Calculate the values:

[0128] Assume this value is 0.14. Assume this value is 0.65.

[0129] Because she is a female, gender=1.

[0130] Then calculate the predicted value of free triiodothyronine:

[0131]

[0132] FT3 predicted =4.5+0.2×0.14+0.15×0.65+0.1×1+0.08×7×1=

[0133] 5.2844pmol / L.

[0134] Finally, the prediction interval for free triiodothyronine is calculated:

[0135] FT3 upper =FT3 predicted +1.96FT3 std =5.2844+0.392=5.6764;

[0136] FT3 lower =FT3 predicted -1.96FT3 std =5.2844-0.392=4.8924.

[0137] The target multivariate continuous biological reference interval model also includes a calculation formula for obtaining the prediction interval of free thyroxine, which is:

[0138]

[0139] FT4 upper =FT4 predicted +1.96FT4 std ;

[0140] FT4 lower =FT4 predicted -1.96FT4 std ;

[0141] Among them, FT4 predicted Stands for free thyroxine value, FT4 upper Represents the upper limit of the prediction interval of free thyroxine, FT4 lower Represents the lower limit of the prediction interval of free thyroxine, FT4 std represents the standard deviation of free thyroxine values, q0 is a constant term, which represents a baseline free thyroxine predicted value without considering factors such as age, month and gender;

[0142] q1, q2, q3, and q4 are coefficients that determine the degree to which different factors affect the predicted value of free thyroxine;

[0143] The square root of age is used to smooth out the effect of age or to present a certain trend. The cyclical variation factor of the month is introduced, and the q4 coefficient is multiplied by this part, which reflects the degree of influence of age and month in this specific combination on the predicted value of free thyroxine.

[0144]

[0145] FT4 predicted : Represents the predicted value of free thyroxine. The estimated value of free thyroxine level is calculated by this formula.

[0146] Assume q0 = 15, which is a constant term. If age, month, and gender are not considered, the baseline free thyroxine prediction value is 15 pmol / L. Assume q1 = 0.05, q2 = 0.1, q3 = 0.08, q4 = 0.12, and q5 = 0.06. These coefficients determine the degree of influence of different factors on the free thyroxine prediction value. Age-related part: q1×age: Assuming the user is 40 years old, the value of this part is q1×40=0.05×40=2.

[0147] Month related part: q2×month: Assuming the current month is August, the value of this part is q2×8=0.1×8=0.8.

[0148] Gender-related part: q3×gender: Assume that male is coded as 0 and female is coded as 1. If the user is male, the value of this part is q3×0=0; if the user is female, assuming q3=0.08, the value of this part is q3×1=0.08.

[0149] Age and month specific combination: The square root of age makes the effect of age smoother or presents a specific trend, while introducing the periodic change factor of the month. Assuming the user is 40 years old and the current month is August, Assuming sin(8) is 0.99, the value of this part is

[0150] Age, month, and gender interaction part: q5×age×month×gender: Assuming the user is a female, 40 years old, and the current month is August, the value of this part is q5×40×8×1=0.06×40×8=19.2.

[0151] Suppose we want to predict the free thyroxine interval for a 40-year-old female user whose current month is August. The user is. q1×age=0.05×40=2. q2×month=0.1×8=0.8. Because she is a female, gender=1, q3×gender=0.08×1=0.08. sin(8) is assumed to be 0.99, q5×age×month×gender=0.06×40×8×1=19.2.

[0152] Next, the predicted free thyroxine value is calculated:

[0153] FT4 predicted ==15+2+0.8+0.08+0.74+19.2=37.82.

[0154] Finally, the prediction interval of free thyroxine is calculated:

[0155] FT4 upper =FT4 predicted +1.96 FT4 std =38.82+9.6=48.42;

[0156] FT4 lower =FT4 predicted -1.96 FT4 std =38.82-9.6=29.22.

[0157] The following is a practical application example of the free thyroxine prediction interval calculation formula: In the endocrinology department of a large general hospital, doctors are committed to finding a more accurate way to evaluate users' thyroid function, especially free thyroxine (FT4) levels. They decided to conduct the evaluation based on the established reference interval model adjusted according to various variables (age, gender, month). Doctors use this model, combined with the user's specific information, to provide a more accurate basis for diagnosis and treatment. Taking Xiao Li as an example, the model derives a free thyroxine prediction interval based on Xiao Li's age, gender, test month and other information. After that, the doctor compares this prediction interval with Xiao Li's free thyroxine level obtained by the actual test. If the actual test value is significantly different from the prediction interval, the doctor will conduct further detailed examinations to determine whether there is a thyroid disease or other factors that affect thyroid function. For example, if the actual test value is significantly lower than the prediction interval, this may indicate hypothyroidism; if the actual test value is significantly higher than the prediction interval, it may mean hyperthyroidism.

[0158] In addition, doctors will also comprehensively consider other thyroid-related indicators (such as triiodothyronine, thyroid-stimulating hormone, etc.) and the user's clinical symptoms (such as fatigue, weight changes, abnormal heart rate, etc.) to comprehensively evaluate the user's thyroid health status and develop a personalized treatment plan. Moreover, by applying this model to multiple users, the parameters in the model can be continuously optimized, thereby improving the accuracy and reliability of the prediction, and providing stronger support for the diagnosis and treatment of thyroid diseases.

[0159] In this application, the server obtains the user information to be monitored, the target event set, the training sample set and the preset model. The training sample set is processed to generate a target feature library and divide the training set and the verification set, and the preset model is trained using it. The user information to be monitored and the target event set are processed separately to obtain the user's physiological parameter information and the detection influencing factors. This information is processed using the target multivariate continuous biological reference interval model to obtain the thyroid stimulating hormone equivalent and the multivariate continuous biological prediction interval, and then the user's thyroid function status is judged. The continuous reference interval constructed by this method takes into account the continuous changes in gender, age and sampling month, and is more individually responsive. Compared with the traditional reference interval, it can greatly reduce the possibility of false positives and false negatives in the test results, improve the accuracy of clinical diagnosis, meet the requirements of precision medicine, and can customize the treatment plan according to the characteristics of individual patients, providing a more accurate and personalized basis for thyroid function assessment and clinical decision-making.

[0160] In one embodiment, if Figure 2 As shown, the present application also provides an early warning device for thyroid function status, comprising:

[0161] An acquisition module 201 is used to acquire user information to be monitored, a target event set matching the user information to be monitored, a training sample set, and a preset multivariate continuous biological reference interval model;

[0162] Processing module 202 is used to process the training sample set to generate a target feature library, wherein the target feature library includes non-continuous variable data and continuous variable data; based on the non-continuous variable data, the training sample set is processed to generate an initial training set and an initial verification set; based on the continuous variable data, the initial training set and the initial verification set are processed to generate a target training set and a target verification set, wherein the target training set includes a plurality of biological reference intervals, and different biological reference intervals correspond to different user information of the same disease type; based on the target training set and the target verification set, the preset multivariate continuous biological reference interval model is trained. Practice to generate a target multivariate continuous biological reference interval model; process the user information to be detected to generate physiological parameter information of the user to be detected, wherein the physiological parameter information of the user to be detected includes thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, free thyroxine, and the user's gender, age and current month information; process the target event set that matches the user information to be monitored to generate a detection influencing factor; based on the target multivariate continuous biological reference interval model, process the physiological parameter information of the user to be detected and the detection influencing factor to generate a detection status of the target user's thyroid function.

[0163] The computer-readable storage medium provided in the above-mentioned embodiments of the present application and the early warning method for thyroid function status provided in the embodiments of the present application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0164] An embodiment of the present application provides a computer program product, including a computer program, wherein the computer program is executed by a third processor to implement the method described above.

[0165] The computer program product provided in the above-mentioned embodiments of the present application and the early warning method for thyroid function status provided in the embodiments of the present application are based on the same inventive concept and have the same beneficial effects as the method adopted, run or implemented by the application program stored therein.

Claims

1. A method for early warning of thyroid function status, characterized in that: include: Obtaining user information to be monitored, a target event set matching the user information to be monitored, a training sample set, and a preset multivariate continuous biological reference interval model; Processing the training sample set to generate a target feature library, wherein the target feature library includes non-continuous variable data and continuous variable data; Processing the training sample set based on the non-continuous variable data to generate an initial training set and an initial validation set; The initial training set and the initial validation set are processed based on the continuous variable data to generate a target training set and a target validation set, wherein the target training set includes thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, free thyroxine of other users, the user's gender, age and month information of sample collection; Training the preset multivariate continuous biological reference interval model based on the target training set and the target verification set to generate a target multivariate continuous biological reference interval model; Processing the user information to be detected to generate physiological parameter information of the user to be detected, wherein the physiological parameter information of the user to be detected includes thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, free thyroxine, and the user's gender, age and month information of sample collection; Process the target event set that matches the user information to be monitored to generate a detection impact factor; The physiological parameter information of the user to be detected and the detection influencing factors are processed based on the target multivariate continuous biological reference interval model to generate a detection status of the target user's thyroid function.

2. The method according to claim 1, characterized in that Processing the training sample set to generate a target feature library includes: Performing feature extraction processing on the training sample set to generate a user physiological parameter database, wherein the user physiological parameter database includes the user's gender, age, month information and the test results of five thyroid function items, wherein the test results of the five thyroid function items include the test results of thyroid stimulating hormone, the test results of triiodothyronine, the test results of thyroxine, the test results of free triiodothyronine, and the test results of free thyroxine; The user's gender, age, month information and the test results of the five thyroid function items are processed respectively to generate target feature data, wherein the target feature data includes various parameters of the smoothing function, wherein the various parameters of the smoothing function are used to characterize the accuracy of the multivariate continuous biological reference interval model; The user physiological parameter database is processed based on the target feature data to generate a target feature library.

3. The method according to claim 1, characterized in that The preset multivariate continuous biological reference interval model is trained based on the target training set and the target verification set to generate a target multivariate continuous biological reference interval model, including: Processing the target training set and the target validation set to generate multiple continuous variable data, discontinuous variable data, and physiological parameter information of users, wherein the multiple continuous variable data include age information of different users and month information, the discontinuous variable data include gender information of different users, and the physiological parameter information of users includes thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, free thyroxine, gender, age of users, and month information of sample collection; Processing the physiological parameter information of the user based on the multiple continuous variable data and the non-continuous variable data to generate a target biological reference interval, wherein the target biological reference interval is generated based on thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, free thyroxine of different users, the user's gender, age and month information of specimen collection; A target multivariate continuous biological reference interval model is generated based on the target biological reference interval.

4. The method according to claim 1, characterized in that The target event set matching the user information to be monitored is processed to generate detection influencing factors, including: Processing a target event set that matches the user information to be monitored to generate type information of the target event, wherein the type information of the target event is used to characterize an influence factor of the target event on the physiological parameters of the user to be monitored; Processing the type information of the target event to generate historical physiological parameter information of the user to be monitored, wherein the historical physiological parameter information of the user to be monitored is used to characterize physiological parameter information when different users experience the target event of the same type; Processing the historical physiological parameter information of the user to be monitored to generate physiological parameter influencing factors and weight information corresponding to the physiological parameter influencing factors; The physiological parameter influencing factor and the weight information corresponding to the physiological parameter influencing factor are processed to generate a detection influencing factor, wherein the detection influencing factor is used to adjust the physiological parameter influencing factor of the target user.

5. The method according to claim 4, characterized in that The historical physiological parameter information of the user to be monitored is processed to generate physiological parameter influencing factors and weight information corresponding to the physiological parameter influencing factors, including: Processing the historical physiological parameter information of the user to be monitored to generate a first impact value of the physiological parameter of the user to be monitored; Obtain historical physiological parameter information of other users; Processing historical physiological parameter information of other users to generate second impact values ​​of the physiological parameters of other users; Processing the first influence value of the physiological parameter of the user to be monitored and the second influence value of the physiological parameter of the other user to generate a physiological parameter influence factor and weight information corresponding to the physiological parameter influence factor; The method further includes a calculation formula for obtaining a first impact value, wherein the calculation formula is: Among them, T3 avg 、T4 avg , FT3 avg , FT4 avg They are the average values ​​of triiodothyronine, thyroxine, free triiodothyronine and free thyroxine of the monitored users respectively; T3 std 、T4 std , FT3 std , FT4 std are the standard deviations of the corresponding parameters; n1, n2, n3, n4 are weight coefficients; The method further includes a calculation formula for obtaining a second impact value, wherein the calculation formula is: Among them, T3 other，avg 、T4 other，avg , FT3 other，avg , FT4 other，avg are the average values ​​of other users on triiodothyronine, thyroxine, free triiodothyronine, and free thyroxine, respectively; T3 other，std 、T4 other，std , FT3 other，std , FT4 other，std are the standard deviations of the corresponding parameters; p1, p2, p3, p4 are weight coefficients.

6. The method according to claim 4, characterized in that The physiological parameter information of the user to be detected and the detection influencing factors are processed based on the target multivariate continuous biological reference interval model to generate a detection status of the target user's thyroid function, including: Based on the target multivariate continuous biological reference interval model, the physiological parameter information of the user to be detected and the detection influencing factors are processed to generate the thyroid stimulating hormone value, triiodothyronine value, thyroxine value, free triiodothyronine value, free thyroxine value of the user to be monitored and the multivariate continuous biological prediction interval matching the user to be monitored, wherein the multivariate continuous biological prediction interval matching the user to be monitored includes the thyroid stimulating hormone prediction interval, the triiodothyronine prediction interval, the thyroxine prediction interval, the free triiodothyronine prediction interval and the free thyroxine prediction interval; Based on the multivariate continuous biological prediction intervals that match the user to be monitored, the thyroid-stimulating hormone value, triiodothyronine value, thyroxine value, free triiodothyronine value and free thyroxine value of the monitored user are compared, evaluated and processed to generate the detection status of the target user's thyroid function.

7. The method according to claim 6, characterized in that The target multivariate continuous biological reference interval model is used to compare and process the physiological parameter information of the user to be detected and the detection influencing factors to generate the detection status of the thyroid function of the target user, further comprising: The target multivariate continuous biological reference interval model also includes a calculation formula for obtaining a prediction interval of triiodothyronine, and the calculation formula is: T3 predicted =a i +b i (x-t i )+c i (x-t i ) 2 +d i (x-t i ) 3 +e1×age+f1×month+g1×gender; <h2 style=";text-align:left;direction:ltr">T3<h2 style=";text-align:left;direction:ltr"> upper <h2 style=";text-align:left;direction:ltr"> =T3<h2 style=";text-align:left;direction:ltr"> predicted <h2 style=";text-align:left;direction:ltr"> +1.96T3<h2 style=";text-align:left;direction:ltr"> std <h2 style=";text-align:left;direction:ltr"> ; <h2 style=";text-align:left;direction:ltr">T3<h2 style=";text-align:left;direction:ltr"> lower <h2 style=";text-align:left;direction:ltr"> =T3<h2 style=";text-align:left;direction:ltr"> predicted <h2 style=";text-align:left;direction:ltr"> -1.96T3<h2 style=";text-align:left;direction:ltr"> std <h2 style=";text-align:left;direction:ltr"> ; Among them, T3 predicted stands for triiodothyronine, T3 upper represents the upper limit of the prediction interval of triiodothyronine, T3 lower represents the lower limit of the prediction interval of triiodothyronine, T3 std represents the standard deviation of triiodothyronine values, a i is a reference value, b i 、c i ,d i Determine the linear, quadratic and cubic trends of triiodothyronine values ​​with the change of variable x; t i is a specific parameter or reference value related to triiodothyronine; e1, f1, and g1 represent the influence coefficients of age, month, and gender on the triiodothyronine value, respectively; The target multivariate continuous biological reference interval model also includes a calculation formula for obtaining a thyroxine prediction interval, which is: T4 upper =T4 predicted +1.96T4 std ; T4 lower =T4 predicted -1.96T4 std ; Among them, T4 predicted Represents thyroxine value, T4 upper Represents the upper limit of the prediction interval of thyroxine, T4 lower Represents the lower limit of the prediction interval of thyroxine, T4 std represents the standard deviation of thyroxine values, t i is a specific parameter or reference value related to thyroxine; k is a fitting parameter that affects the prediction interval of thyroxine; i is an index parameter, which is related to the grouping of data or the division of specific intervals, and is used to distinguish different situations or data subsets; k is a fitting parameter that affects the thyroxine value, which determines the complexity of the formula fitting the data; x is an input variable, which is a physiological indicator, environmental factor or other influencing factor related to thyroxine; B i,k-1 (x) and B i+1,k-1 (x) jointly determines a component of the thyroid hormone value, reflecting the complex relationship between other variables and thyroid hormone; e2, f2, and g2 represent the coefficients of influence of age, month, and gender on thyroid hormone values, respectively; The target multivariate continuous biological reference interval model also includes a calculation formula for obtaining a prediction interval of free triiodothyronine, which is: <h2 style=";text-align:left;direction:ltr">FT3<h2 style=";text-align:left;direction:ltr"> upper <h2 style=";text-align:left;direction:ltr"> =FT3<h2 style=";text-align:left;direction:ltr"> predicted <h2 style=";text-align:left;direction:ltr"> +1.96FT3<h2 style=";text-align:left;direction:ltr"> std <h2 style=";text-align:left;direction:ltr"> ; <h2 style=";text-align:left;direction:ltr">FT3<h2 style=";text-align:left;direction:ltr"> lower <h2 style=";text-align:left;direction:ltr"> =FT3<h2 style=";text-align:left;direction:ltr"> predicted <h2 style=";text-align:left;direction:ltr"> -1.96FT3<h2 style=";text-align:left;direction:ltr"> std <h2 style=";text-align:left;direction:ltr"> ; Among them, FT3 predicted FT3 stands for free triiodothyronine upper represents the upper limit of the prediction interval for free triiodothyronine, FT3 lower Represents the lower limit of the prediction interval for free triiodothyronine, FT3 std represents the standard deviation of the triiodothyronine value, P0 is a constant term, which represents a baseline free triiodothyronine value without considering factors such as age, month and gender; P1, P2, P3, P4 are coefficients, which respectively determine the degree of influence of different factors on the free triiodothyronine value; The introduction of makes the effect of age on free triiodothyronine present a cyclical or fluctuating trend; The introduction of makes the influence of month on free triiodothyronine present another cyclical change trend; The target multivariate continuous biological reference interval model also includes a calculation formula for obtaining a prediction interval of free thyroxine, which is: FT4 upper =FT4 predicted +1.96FT4 std ; FT4 lower =FT4 predicted -1.96FT4 std ; Among them, FT4 predicted Stands for free thyroxine value, FT4 upper Represents the upper limit of the prediction interval of free thyroxine, FT4 lower Represents the lower limit of the prediction interval of free thyroxine, FT4 std represents the standard deviation of free thyroxine values, q0 is a constant term, which represents a baseline free thyroxine value when factors such as age, month and gender are not considered; q1, q2, q3, q4 are coefficients, which respectively determine the degree of influence of different factors on the free thyroxine value; The square root of age is used to smooth out the effect of age or to present a certain trend. The monthly cyclical variation factor is introduced, and the q4 coefficient is multiplied by this part, reflecting the degree of influence of age and month on the free thyroxine value in this specific combination.

8. A thyroid function status early warning device, characterized in that: The device comprises: An acquisition module, used to acquire user information to be monitored, a target event set matching the user information to be monitored, a training sample set, and a preset multivariate continuous biological reference interval model; A processing module is used to process the training sample set to generate a target feature library, wherein the target feature library includes non-continuous variable data and continuous variable data; based on the non-continuous variable data, the training sample set is processed to generate an initial training set and an initial verification set; based on the continuous variable data, the initial training set and the initial verification set are processed to generate a target training set and a target verification set, wherein the target training set includes thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, free thyroxine of other users, the user's gender, age and month of sample collection; based on the target training set and the target verification set, the preset multivariate continuous The target biological reference interval model is trained to generate a target multivariate continuous biological reference interval model; the user information to be detected is processed to generate physiological parameter information of the user to be detected, wherein the physiological parameter information of the user to be detected includes thyroid stimulating hormone, triiodothyronine, thyroxine, free triiodothyronine, free thyroxine, and the user's gender, age and month information of sample collection; the target event set matching the user information to be monitored is processed to generate a detection influencing factor; based on the target multivariate continuous biological reference interval model, the physiological parameter information of the user to be detected and the detection influencing factor are processed to generate a detection status of the target user's thyroid function.

9. An electronic device, characterized in that: include: a first processor; and a memory for storing executable instructions of the first processor; Wherein, the first processor is configured to execute the early warning method for thyroid function status as described in any one of claims 1 to 7 by executing the executable instructions.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the second processor, the method for early warning of thyroid function status described in any one of claims 1 to 7 is implemented.

Images