Construction method of nomogram model for predicting efficacy of thyroid-associated ophthalmopathy treated with tocilizumab

Abstract

The application discloses a nomogram model construction method for predicting the curative effect of tocilizumab on thyroid eye disease, relates to the technical field of intelligent medical treatment, and comprises the following steps: acquiring multi-modal diagnosis and treatment data of thyroid eye disease; constructing a training set and a verification set based on the multi-modal diagnosis and treatment data; in the training set, a single factor Cox regression analysis, LASSO ten-fold cross-validation and multi-factor Cox regression analysis are adopted to screen a prediction variable combination; and a nomogram model is constructed based on the prediction variable combination; by acquiring the multi-modal diagnosis and treatment data, screening the prediction variable combination via the single factor Cox regression analysis, the LASSO ten-fold cross-validation and the multi-factor Cox regression analysis, constructing the nomogram model from the screened variables, and completing multi-dimensional testing by using the verification set, the prediction accuracy of the risk probability of non-response after the tocilizumab treatment of thyroid eye disease can be moderately improved, the prediction model suitable for the tocilizumab medication scene is formed, and a reference basis is provided for clinical diagnosis and treatment.

Description

Technical Field

[0001] This invention relates to the field of smart medical technology, and in particular to a method for constructing a nomogram model to predict the efficacy of tocilizumab in treating thyroid eye disease. Background Technology

[0002] Thyroid ophthalmopathy is an autoimmune orbital disease associated with thyroid dysfunction, primarily affecting the orbit and periorbital soft tissues. Symptoms include eyelid retraction, proptosis, diplopia, and even compressive optic neuropathy, severely impacting visual function and quality of life. For moderate to severe active thyroid ophthalmopathy, intravenous corticosteroids are currently the recommended first-line treatment. However, some patients do not respond to intravenous corticosteroids and require second-line biologic therapy. Tocilizumab, a recombinant humanized anti-interleukin-6 receptor monoclonal antibody, has been proven effective and safe for corticosteroid-refractory thyroid ophthalmopathy. However, tocilizumab treatment exhibits significant individual heterogeneity, and not all patients with thyroid ophthalmopathy benefit from it. Therefore, accurately predicting patient response to tocilizumab before treatment is of significant clinical importance in avoiding ineffective treatment, reducing medical costs, and facilitating timely switching to alternative therapies.

[0003] Currently, clinical assessment of inflammatory activity in thyroid eye disease and prediction of drug efficacy primarily relies on the Clinical Activity Score (COSS) system, with some medical institutions combining it with magnetic resonance imaging (MRI) to detect orbital tissue inflammation. The COSS only collects subjective inflammatory manifestations of the anterior segment, such as the eyelids and conjunctiva, while MRI can obtain high-resolution anatomical images of deep orbital tissues. Both methods provide a basic reference for predicting clinical efficacy, and multimodal index-based joint modeling is increasingly being applied to research related to predicting the efficacy of ophthalmic treatments.

[0004] However, current clinical activity scoring systems only cover superficial inflammation indicators of the anterior segment and cannot quantify the degree of inflammatory infiltration in deep orbital tissues such as extraocular muscles and lacrimal glands, resulting in limited assessment dimensions. Magnetic resonance imaging (MRI) suffers from limitations such as high examination costs, long scanning times, and inability to detect in patients with implanted metal structures. Most existing efficacy prediction models are built around glucocorticoid drugs. Due to the significant difference in the mechanism of action of tocilizumab compared to glucocorticoids, existing models are difficult to adapt to the needs of predicting tocilizumab efficacy, cannot accurately reflect the true level of inflammation in deep orbital tissues, and cannot accurately identify the therapeutic effect of tocilizumab before administration. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a nomogram model construction method for predicting the efficacy of tocilizumab in treating thyroid eye diseases, which solves the technical problems of insufficient accuracy in efficacy prediction and lack of a prediction model suitable for tocilizumab in existing technologies.

[0006] The technical means employed in this invention are as follows: In a first aspect, the present invention provides a method for constructing a nomogram model to predict the efficacy of tocilizumab in treating thyroid ophthalmopathy, comprising: Acquire multimodal diagnostic and treatment data for thyroid eye diseases; Based on the multimodal diagnosis and treatment data, a training set and a validation set were constructed. In the training set, univariate Cox regression analysis, LASSO 10-fold cross-validation, and multivariate Cox regression analysis were used to select combinations of predictive variables. A nomogram model is constructed based on the combination of the predictor variables. The nomogram model is used to predict the probability of no response risk at multiple preset time periods after tocilizumab treatment. The nomogram model is validated in multiple dimensions based on the validation set to obtain the validated nomogram model.

[0007] Furthermore, the multimodal diagnostic and treatment data includes at least one of demographic characteristic data, thyroid disease history data, ocular symptoms and signs data, serological index data, SPECT / CT functional imaging parameters, and efficacy evaluation data.

[0008] Furthermore, the construction of the training set and validation set based on the multimodal diagnosis and treatment data includes: randomly dividing the multimodal diagnosis and treatment data into a training set and a validation set in a 7:3 ratio.

[0009] Furthermore, the preset time periods include: one month, two months, and three months.

[0010] Furthermore, the nomogram model maps multiple target variable values ​​after tocilizumab treatment to multiple variable scores, sums the multiple variable scores to obtain a total score, and calculates the probability of no response risk over multiple preset time periods based on the total score.

[0011] Furthermore, the multi-dimensional validation of the nomogram model based on the validation set includes: validating the nomogram model based on the validation set in terms of discrimination, calibration, and clinical net benefit.

[0012] Secondly, the present invention also provides a nomogram model construction system for predicting the efficacy of tocilizumab in treating thyroid ophthalmopathy, comprising: The data acquisition module is used to acquire multimodal diagnostic and treatment data for thyroid eye diseases; The variable screening module is used to construct a training set and a validation set based on the multimodal diagnosis and treatment data. In the training set, univariate Cox regression analysis, LASSO 10-fold cross-validation and multivariate Cox regression analysis are used to screen out combinations of predictive variables. The model building module is used to build a nomogram model based on the combination of the predictor variables. The nomogram model is used to predict the probability of no response risk at multiple preset time periods after tocilizumab treatment. The model validation module is used to perform multi-dimensional validation of the nomogram model based on the validation set to obtain the validated nomogram model.

[0013] Thirdly, the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements the nomogram model construction method for predicting the efficacy of tocilizumab in treating thyroid ophthalmopathy as described in any one of the first aspects.

[0014] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program, characterized in that, when the computer program is executed by a processor, it implements the nomogram model construction method for predicting the efficacy of tocilizumab in treating thyroid ophthalmopathy as described in the first aspect.

[0015] Compared with the prior art, the present invention has the following advantages: This invention acquires multimodal diagnostic and treatment data, and selects predictive variable combinations through univariate Cox regression analysis, LASSO 10-fold cross-validation, and multivariate Cox regression analysis. The selected variables are used to construct a nomogram model, and multidimensional tests are performed using the validation set. This can moderately improve the predictive accuracy of the probability of no response after tocilizumab treatment for thyroid eye disease, forming a predictive model adapted to tocilizumab medication scenarios, and providing a reference for clinical diagnosis and treatment judgment.

[0016] Based on the above reasons, this invention can be widely promoted in fields such as smart healthcare. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the method for constructing a nomogram model to predict the efficacy of tocilizumab in treating thyroid eye disease in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of the column line diagram in an embodiment of the present invention. Detailed Implementation

[0019] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0020] It should be noted that the terms "comprising" and "having" and any variations thereof in this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or device.

[0021] The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0022] Please see Figure 1 , Figure 1 This is a flowchart illustrating the method for constructing a nomogram model to predict the efficacy of tocilizumab in treating thyroid eye disease, as described in an embodiment of the present invention.

[0023] This invention provides a method for constructing a nomogram model to predict the efficacy of tocilizumab in treating thyroid ophthalmopathy, comprising the following steps: Step 101: Obtain multimodal diagnostic and treatment data for thyroid eye disease; By obtaining multimodal diagnostic and treatment data for thyroid eye disease, we can collect multidimensional diagnostic and treatment-related information of the subjects, objectively reflect the actual situation of inflammation in the deep orbital tissues, and reduce the one-sidedness of data acquisition caused by relying solely on surface inflammation indicators.

[0024] In some embodiments, multimodal diagnostic and treatment data includes at least one of demographic data, thyroid disease history data, ocular symptoms and signs data, serological index data, SPECT / CT functional imaging parameters, and efficacy evaluation data.

[0025] The demographic data include at least one of the following: age, sex, smoking index, drinking history, history of hypertension, history of diabetes, and history of treatment for thyroid eye disease.

[0026] Thyroid disease history data includes at least one of the following: thyroid disease course data, thyroid function normality indicators, and ocular symptom course data.

[0027] Data on ocular symptoms and signs include diplopia, best corrected visual acuity, intraocular pressure, eyeball protrusion, clinical activity score, palpebral fissure height, MRD-1, MRD-2, and at least one of the following indicators of limited eye movement.

[0028] Serological marker data include at least one of the following: T-cell tuberculosis infection markers, thyroid-stimulating hormone receptor antibody, thyroid-stimulating hormone, thyroglobulin, thyroid peroxidase antibody, free triiodothyronine, free thyroxine, total triiodothyronine, and total thyroxine.

[0029] SPECT / CT functional imaging parameters include the maximum standardized uptake values ​​of the periorbital soft tissue, extraocular muscles, lacrimal gland, and retrobulbar connective tissue regions.

[0030] Efficacy assessment data include at least one of clinical activity scores and treatment response outcome data.

[0031] Step 102: Construct training and validation sets based on multimodal medical data. In the training set, use univariate Cox regression analysis, LASSO 10-fold cross-validation, and multivariate Cox regression analysis to select combinations of predictive variables. Univariate Cox regression analysis uses p < 0.10 as the selection threshold, LASSO 10-fold cross-validation uses Lambda-1se as the optimal penalty parameter, and multivariate Cox regression analysis uses a backward stepwise regression method with the goal of minimizing the Akaike Information Criterion (AIC). By sequentially using univariate Cox regression analysis with a p < 0.10 selection threshold, LASSO 10-fold cross-validation with Lambda-1se as the optimal penalty parameter, and multivariate Cox regression analysis with backward stepwise regression aiming to minimize AIC, variables within the multimodal medical data are filtered to streamline the selection of predictive variables with high correlation to the outcome event, reducing the interference of irrelevant variables on prognostic judgment.

[0032] In some embodiments, constructing training and validation sets based on multimodal medical data includes: randomly dividing the multimodal medical data into training and validation sets in a 7:3 ratio. This allows for testing the stability of the selected predictive variable combinations based on two independent sets of multimodal medical data.

[0033] Step 103: Construct a nomogram model based on the combination of predictor variables. The nomogram model is used to predict the probability of no response risk within multiple preset time periods after tocilizumab treatment. By constructing a nomogram model based on the combination of predictor variables obtained through screening, the probability of no response risk corresponding to patients within multiple preset time periods after tocilizumab treatment can be quantified, so as to predict the effect of different patients receiving tocilizumab treatment.

[0034] In some embodiments, multiple preset time periods include one month, two months, and three months, thereby periodically assessing changes in response risk within a short timeframe after patients receive tocilizumab treatment.

[0035] In some embodiments, the nomogram model maps multiple target variable values ​​after tocilizumab treatment to multiple variable scores, sums the multiple variable scores to obtain a total score, and calculates the probability of no response risk over multiple preset time periods based on the total score. This step-by-step conversion method, which maps the patient's target variable values ​​after tocilizumab treatment to variable scores, summarizes them to obtain a total score, and then calculates the probability of no response risk using the nomogram model, simplifies the risk probability calculation process and facilitates intuitive acquisition of the patient's probability of no response over each preset time period.

[0036] Step 104: Perform multi-dimensional validation of the nomogram model based on the validation set to obtain the validated nomogram model. By using the validation set to perform multi-dimensional validation of the nomogram model, the generalization ability of the nomogram model in predicting the probability of no response risk in the tocilizumab-treated patient population is examined.

[0037] In some embodiments, the nomogram model is validated in multiple dimensions based on the validation set, including: validating the nomogram model in terms of discrimination, calibration and clinical net benefit dimensions based on the validation set.

[0038] The discrimination dimension can be validated using ROC curves (receiver operating characteristic curves) and AUC (area under the ROC curve), the calibration dimension can be validated using calibration curves, and the clinical net benefit dimension can be validated using decision-curve analysis (DCA).

[0039] As a specific example: In this embodiment, three types of data to be analyzed are acquired: clinical indicator data, serological indicator data, and SPECT / CT examination data.

[0040] Clinical indicator data includes demographic data, thyroid disease-related medical history data, and ocular symptoms and signs data. Among them, demographic data may include age, sex, smoking index, alcohol consumption history, history of hypertension, history of diabetes, and history of treatment for thyroid-related eye diseases. Thyroid disease-related medical history data includes the duration of thyroid disease, whether thyroid function has returned to normal, and the duration of ocular symptoms. Ocular symptoms and signs data include diplopia, best corrected visual acuity, intraocular pressure, eyeball protrusion, clinical activity score and each item corresponding to the clinical activity score, palpebral fissure height, distance from the upper eyelid margin to the pupillary midline -1, distance from the upper eyelid margin to the pupillary midline -2, and limitation of eye movement.

[0041] Serological data include T-cell tuberculosis infection status, thyroid-stimulating hormone receptor antibody, thyroid-stimulating hormone, thyroglobulin, thyroid peroxidase antibody, free triiodothyronine, free thyroxine, total triiodothyronine, and total thyroxine.

[0042] The SPECT / CT data acquisition process was as follows: 99mTc-diethylenetriaminepentaacetic acid was selected as the tracer. After intravenous injection of 555MBq (15mCi) of the tracer, single-photon emission computed tomography (SPECT) was performed 30 minutes later. The maximum standard uptake values ​​were quantitatively calculated for four regions: periorbital soft tissue, extraocular muscles, lacrimal gland, and retrobulbar connective tissue using Q.Metrix software. 1.85 was set as the cutoff value, and the maximum standard uptake values ​​corresponding to each region were binary classified and analyzed.

[0043] Data related to tocilizumab use were collected, and efficacy evaluation data were collected at the time points of the 1st, 2nd and 3rd months after medication. The response status was defined as follows: the data of at least two of the four evaluation indicators of the affected eye after treatment showed improvement, and the data of any one of the four indicators did not deteriorate. The specific contents of the four evaluation indicators are: (1) the value of palpebral fissure height decreased by more than or equal to 2 mm; (2) the value of eyeball protrusion decreased by more than or equal to 3 mm; (3) the value of eyeball movement amplitude improved by more than or equal to 8 degrees; (4) the overall score of the five clinical activity scoring items, excluding the pain after spontaneous eyeball movement and the pain during eyeball movement, decreased by more than or equal to 1 point. The criteria for deterioration were set: when thyroid-associated ophthalmopathy compressive optic neuropathy appeared in the data, or when the data of at least two of the aforementioned four evaluation indicators deteriorated, it was determined to be a deterioration status. All sample data that did not meet the criteria for response were classified into the non-response category.

[0044] In the training set, combinations of predictive variables are selected, and the specific selection method is as follows: First, univariate Cox regression analysis was used, with a p-value less than 0.10 as the screening threshold, to identify candidate indicators associated with non-response tocilizumab treatment. These candidate indicators included: thyroid disease duration, thyroid function recovery status, intraocular pressure, eyeball protrusion, clinical activity score, conjunctival hyperemia, eyelid edema, thyroid-stimulating hormone receptor antibody, thyroglobulin, maximum standardized uptake value in the extraocular muscle region, and maximum standardized uptake value in the lacrimal gland region. Then, LASSO regression combined with 10-fold cross-validation was used, selecting Lambda-1-se as the optimal penalty parameter. After screening, four variables were retained: intraocular pressure, thyroglobulin, maximum standardized uptake value in the extraocular muscle region, and maximum standardized uptake value in the lacrimal gland region. Finally, multivariate Cox regression analysis was used: the data of the four selected variables were included in the analysis, and a combination of predictive variables was constructed using a backward stepwise regression method. The minimum value of the Akaike Information Criterion was used as the selection target. The analysis results are as follows: maximum standard uptake value of extraocular muscle region (hazard ratio = 4.489, 95% confidence interval: 1.769-11.348, p = 0.002); maximum standard uptake value of lacrimal gland region (hazard ratio = 6.695, 95% confidence interval: 2.646-16.939, p < 0.001); intraocular pressure (hazard ratio = 0.368, 95% confidence interval: 0.158-0.859, p = 0.021); thyroglobulin (hazard ratio = 0.996, 95% confidence interval: 0.993-1.000, p = 0.033).

[0045] Based on the multi-factor Cox regression coefficients, corresponding scores are assigned to each indicator, and a visual nomogram is constructed. The visual nomogram is shown below. Figure 2 As shown.

[0046] The nomogram model underwent multi-dimensional validation, which consisted of three parts: discrimination validation, calibration validation, and clinical net benefit validation. (1) Discrimination validation was performed using receiver operating characteristic (ROC) curves. In the training set data, the area under the curve (AUC) for 1 month after medication was 0.787 (95% confidence interval: 0.681-0.893), the AUC for 2 months after medication was 0.747 (95% confidence interval: 0.641-0.853), and the AUC for 3 months after medication was 0.834 (95% confidence interval: 0.734-0.934). In the validation set data, the AUC for 1 month after medication was 0.761 (95% confidence interval: 0.567-0.955), the AUC for 2 months after medication was 0.799 (95% confidence interval: 0.663-0.934), and the AUC for 3 months after medication was 0.833 (95% confidence interval: 0.655-1).

[0047] (2) Calibration verification was achieved through calibration curves; the actual non-reaction event occurrence probabilities at 1 month, 2 months and 3 months after medication followed the trend of predicted probability and had a high degree of fit with the 45° ideal reference line, indicating that the constructed nomogram model has a good calibration effect and the predicted probability output by the model can objectively reflect the actual risk of non-reaction events.

[0048] (3) Clinical net benefit verification was achieved through decision curve analysis; under the conditions of 1 month and 2 months after medication, within the corresponding threshold probability interval, the clinical net benefit value of the nomogram model constructed in this application is higher than any single predictor variable included in the model; at the same time, within most threshold probability ranges, the net benefit values ​​of the nomogram model and each single predictor variable are higher than the total intervention baseline curve and the total non-intervention baseline curve, thus indicating that the nomogram model and the single predictor variables included in the model have clinical decision reference value. Analysis of the decision curve results for 3 months of medication revealed the following: Within the threshold probability range of 0-0.8, the net benefit values ​​corresponding to the nomogram model and each individual indicator included in the model were consistently higher than the baseline curve for all non-intervention models. Specifically, the net benefit of the nomogram model in the training set was consistently higher than the baseline curve for all intervention models, demonstrating a stable decision advantage. In the validation set, the net benefit of the nomogram model gradually exceeded the baseline curve for all intervention models as the threshold probability increased. However, within certain threshold probability ranges, the net benefit of the nomogram model was lower than that of some individual predictor variables included in the model. This result indicates that under specific decision threshold conditions corresponding to 3 months of medication, it is necessary to comprehensively evaluate the results of individual predictor variables to optimize clinical decision-making.

[0049] This invention also provides a nomogram model construction system for predicting the efficacy of tocilizumab in treating thyroid eye disease, comprising: a data acquisition module for acquiring multimodal diagnostic and treatment data of thyroid eye disease; a variable screening module for constructing a training set and a validation set based on the multimodal diagnostic and treatment data, wherein univariate Cox regression analysis, LASSO 10-fold cross-validation, and multivariate Cox regression analysis are used to screen for combinations of predictive variables in the training set; a model construction module for constructing a nomogram model based on the combinations of predictive variables, wherein the nomogram model is used to predict the probability of no response risk at multiple preset time intervals after tocilizumab treatment; and a model validation module for performing multidimensional validation of the nomogram model based on the validation set to obtain a validated nomogram model.

[0050] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the nomogram model construction method for predicting the efficacy of tocilizumab in treating thyroid ophthalmopathy as described above.

[0051] This invention also provides a computer-readable storage medium storing a computer program, characterized in that, when executed by a processor, the computer program implements the nomogram model construction method for predicting the efficacy of tocilizumab in treating thyroid ophthalmopathy as described above.

[0052] The same or similar parts among the various embodiments in this specification can be referred to mutually, and will not be repeated here.

[0053] This invention acquires multimodal diagnostic and treatment data, and selects predictive variable combinations through univariate Cox regression analysis, LASSO 10-fold cross-validation, and multivariate Cox regression analysis. The selected variables are used to construct a nomogram model, and multidimensional tests are performed using the validation set. This can moderately improve the predictive accuracy of the probability of no response after tocilizumab treatment for thyroid eye disease, forming a predictive model adapted to tocilizumab medication scenarios, and providing a reference for clinical diagnosis and treatment judgment.

[0054] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing a nomogram model to predict the efficacy of tocilizumab in treating thyroid ophthalmopathy, characterized in that, include: Acquire multimodal diagnostic and treatment data for thyroid eye diseases; Based on the multimodal diagnosis and treatment data, a training set and a validation set were constructed. In the training set, univariate Cox regression analysis, LASSO 10-fold cross-validation, and multivariate Cox regression analysis were used to select combinations of predictive variables. A nomogram model is constructed based on the combination of the predictor variables. The nomogram model is used to predict the probability of no response risk at multiple preset time periods after tocilizumab treatment. The nomogram model is validated in multiple dimensions based on the validation set to obtain the validated nomogram model.

2. The method for constructing a nomogram model to predict the efficacy of tocilizumab in treating thyroid ophthalmopathy according to claim 1, characterized in that, The multimodal diagnostic and treatment data includes at least one of the following: demographic data, thyroid disease history data, ocular symptoms and signs data, serological index data, SPECT / CT functional imaging parameters, and efficacy evaluation data.

3. The method for constructing a nomogram model to predict the efficacy of tocilizumab in treating thyroid ophthalmopathy according to claim 1, characterized in that, The construction of training and validation sets based on the multimodal diagnosis and treatment data includes: randomly dividing the multimodal diagnosis and treatment data into training and validation sets in a 7:3 ratio.

4. The method for constructing a nomogram model to predict the efficacy of tocilizumab in treating thyroid ophthalmopathy according to claim 1, characterized in that, The preset time periods include: one month, two months, and three months.

5. The method for constructing a nomogram model to predict the efficacy of tocilizumab in treating thyroid ophthalmopathy according to claim 1, characterized in that, The nomogram model maps multiple target variable values ​​after tocilizumab treatment to multiple variable scores, sums the multiple variable scores to obtain a total score, and calculates the probability of no response risk over multiple preset time periods based on the total score.

6. The method for constructing a nomogram model to predict the efficacy of tocilizumab in treating thyroid ophthalmopathy according to claim 1, characterized in that, The multi-dimensional validation of the nomogram model based on the validation set includes: validating the nomogram model based on the validation set in terms of discrimination, calibration, and net clinical benefit.

7. A nomogram model construction system for predicting the efficacy of tocilizumab in treating thyroid eye disease, characterized in that, include: The data acquisition module is used to acquire multimodal diagnostic and treatment data for thyroid eye diseases; The variable screening module is used to construct a training set and a validation set based on the multimodal diagnosis and treatment data. In the training set, univariate Cox regression analysis, LASSO 10-fold cross-validation and multivariate Cox regression analysis are used to screen out combinations of predictive variables. The model building module is used to build a nomogram model based on the combination of the predictor variables. The nomogram model is used to predict the probability of no response risk at multiple preset time periods after tocilizumab treatment. The model validation module is used to perform multi-dimensional validation of the nomogram model based on the validation set to obtain the validated nomogram model.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for constructing a nomogram model for predicting the efficacy of tocilizumab in treating thyroid eye disease according to any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for constructing a nomogram model for predicting the efficacy of tocilizumab in treating thyroid eye disease according to any one of claims 1 to 6.

Images