Gender stratification-based early pulmonary nodule disease differential diagnosis system

Abstract

The application provides a gender stratified lung nodule disease identification and diagnosis model and system. Specifically, based on a machine learning method, six male characteristic indexes and six female characteristic indexes are used to construct a gender stratified identification and diagnosis system. The results show that the model based on support vector machine has an AUC greater than 0.8 in the training set and the validation set, has good identification ability, and can accurately identify and judge whether the patient with normal serum angiotensin converting enzyme level is a lung nodule disease patient. The system effectively solves the short board of the existing diagnosis method of lung nodule disease, and realizes the identification and diagnosis by using a small amount of characteristic indexes, greatly shortens the diagnosis time, and has a wide clinical application prospect.

Description

Technical Field

[0001] This invention relates to the fields of machine learning and disease differential diagnosis, and more specifically, to a method and system for constructing an early pulmonary sarcoidosis differential diagnosis model based on machine learning for cases with normal serum angiotensin-converting enzyme levels, categorized by gender. Background Technology

[0002] Sarcoidosis is a systemic granulomatous disease characterized by non-caseating necrotizing epithelioid cell granulomas. It can affect multiple organs throughout the body, with the lungs being the most commonly affected site. Sarcoidosis is classified into five stages according to the Scadding staging system. Early pulmonary sarcoidosis (stages 0, I, and II) lacks specific imaging features on chest CT scans, making diagnosis difficult. Approximately 25% of patients experience chronic disease progression. Without timely intervention, these patients often progress to advanced stages, leading to increased mortality due to respiratory failure, infection, involvement of vital organs, or complications. Therefore, early diagnosis and treatment of pulmonary sarcoidosis are crucial.

[0003] While granulomas are a pathological feature of sarcoidosis, they can also be seen in the pathological changes of infectious diseases such as Mycobacterium tuberculosis, non-tuberculous mycobacteria, and fungi. The diagnosis of sarcoidosis is currently still exclusionary, requiring the exclusion of infectious diseases such as pulmonary tuberculosis, as well as non-infectious diseases such as pneumoconiosis and lung cancer. Serum angiotensin-converting enzyme (sACE) is elevated in active pulmonary sarcoidosis and is considered a potential biomarker for the disease; however, its low sensitivity and specificity limit its clinical application value, especially when sACE shows false negatives, further complicating diagnosis. Despite numerous studies on biomarkers for pulmonary sarcoidosis, the diagnostic efficacy of a single indicator is often insufficient to directly guide clinical decisions. The World Association for Sarcoidosis and Other Granulomatous Diseases (WASOG) first established diagnostic requirements for sarcoidosis in 1999 and updated its organ assessment tool for sarcoidosis in 2014. Researchers developed two sarcoidosis diagnostic scores (SDS Clinical Score and SDS Biopsy Score) based on this assessment tool. The SDS Clinical Score ≥3 had a sensitivity of 90.6% and a specificity of 88.5%. However, a multi-state, multicenter validation study demonstrated its insufficient ability to differentiate non-infectious granulomatous diseases (AUC, 0.684; 95% CI, 0.602–0.766). Furthermore, due to the lack of standardization in patient assessment, heterogeneity exists among physicians. While biopsy is often considered essential for diagnosis, its risks and benefits must be carefully weighed. The American Thoracic Society Clinical Practice Guidelines indicate that establishing reliable sarcoidosis prediction models could potentially allow some patients to avoid invasive biopsies. In addition, research translating biomarker research findings into assessment tools that practically guide clinical decision-making remains scarce.

[0004] Therefore, there is an urgent need in the field for a model or system with good specificity and high sensitivity, which can be used to identify and diagnose early pulmonary sarcoidosis patients whose angiotensin-converting enzyme levels are within the normal range, and a method for constructing such a model. Summary of the Invention

[0005] The purpose of this invention is to provide a diagnostic method and system for early pulmonary sarcoidosis based on gender-stratified angiotensin-converting enzyme levels at normal levels.

[0006] In a first aspect, the present invention provides a method for constructing a sex-specific diagnostic model for pulmonary sarcoidosis, the method comprising the steps of: (S1) Provide a clinical dataset of disease patients; the disease patients are those with atypical imaging manifestations, including patients with pulmonary sarcoidosis and patients without pulmonary sarcoidosis; divide the clinical dataset into a male clinical dataset and a female clinical dataset; further divide the male clinical dataset into a male training set and a male validation set, and further divide the female clinical dataset into a female training set and a female validation set; (S2) The clinical dataset is preprocessed to obtain a preprocessed male clinical dataset and a preprocessed female clinical dataset; (S3) Extract feature indicators from the preprocessed male clinical dataset and the preprocessed female clinical dataset respectively, and filter the feature indicators to obtain preferred feature indicators for males and preferred feature indicators for females. (S4) The preferred male feature indicators in the male training set and the preferred female feature indicators in the female training set are used to train the machine learning model, respectively; when the training result reaches the predetermined termination condition, the training of the model is terminated, thereby obtaining the male pulmonary sarcoidosis diagnosis model and the female pulmonary sarcoidosis diagnosis model.

[0007] In another preferred embodiment, the clinical dataset includes: imaging feature data and peripheral blood test data.

[0008] In another preferred embodiment, the imaging features include: the number of infiltrated lung fields and the extent of hilar / mediastinal lymph node enlargement.

[0009] In another preferred embodiment, "atypical imaging features" refers to imaging features that do not meet the typical diagnostic criteria for the disease, and / or imaging features that make it difficult to make a clear diagnosis based on them.

[0010] In another preferred embodiment, the pulmonary sarcoidosis patient includes a pulmonary sarcoidosis patient with normal serum angiotensin-converting enzyme (sACE) levels.

[0011] In another preferred embodiment, "normal serum angiotensin-converting enzyme level" means that the serum angiotensin-converting enzyme level is below the upper limit of the reference range (<70 IU / L).

[0012] In another preferred embodiment, the "upper limit of the reference range" refers to a serum angiotensin-converting enzyme level >70 IU / L.

[0013] In another preferred embodiment, the non-pulmonary sarcoidosis patients include: patients with mediastinal lymph node tuberculosis, patients with pneumoconiosis, patients with lung cancer, and patients with lymphoma.

[0014] In another preferred embodiment, the clinical dataset also includes the age and clinical symptoms of the patients with the disease.

[0015] In another preferred embodiment, the preprocessing includes: handling missing values, handling outliers, transforming data, and standardizing the clinical dataset.

[0016] In another preferred embodiment, the missing value processing employs a multiple interpolation method.

[0017] In another preferred embodiment, step (S3) specifically includes the following steps: (S3.1) Univariate analysis was performed on the preprocessed male clinical dataset and the preprocessed female clinical dataset respectively, and indicators with P < 0.1 were retained; (S3.2) Perform LASSO regression analysis on the indicators obtained in step (S3.1) to obtain the regression coefficients of the male candidate indicator set and the female candidate indicator set and each indicator in the candidate indicator set respectively; (S3.3) Sort the male candidate indicator set and the female candidate indicator set according to the absolute value of the regression coefficient, and select the indicators with the highest absolute value of the regression coefficient and clinical significance, thereby obtaining the preferred characteristic indicators for males and the preferred diagnostic indicators for females respectively.

[0018] In another preferred embodiment, the optimal penalty coefficient is determined using 10-fold cross-validation in the LASSO regression analysis.

[0019] In another preferred embodiment, in step (S4), the training set is tuned using K-fold cross-validation.

[0020] In another preferred embodiment, in step (S4), the training set is tuned using 10-fold cross-validation.

[0021] In another preferred embodiment, the preferred male characteristic indicators are selected from the group consisting of: (A1) albumin (ALB); (A2) absolute lymphocyte count (Lym#); (A3) serum angiotensin-converting enzyme (sACE); (A4) red blood cell distribution width standard deviation (RDW-SD); (A5) erythrocyte sedimentation rate (ESR); (A6) CD4+ to CD8+ ratio (CD4+ / CD8+), or a combination thereof.

[0022] In another preferred embodiment, the preferred female characteristic indicators are selected from the group consisting of: (B1) serum angiotensin-converting enzyme (sACE); (B2) CD4+ to CD8+ ratio (CD4+ / CD8+); (B3) lymphocyte percentage (Lym%); (B4) total cholesterol (TC); (B5) serum prothrombin time (TTs); (B6) serum fibrinogen (FIBs), or a combination thereof.

[0023] In another preferred embodiment, the machine learning model includes: Support Vector Machine (SVM), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Naive Bayes (NBM), Logistic Regression (Logistics), and K-Nearest Neighbors (KNN).

[0024] In another preferred embodiment, the machine learning model is selected from the group consisting of: Support Vector Machine (SVM), Random Forest (RF), and Extreme Gradient Boosting (XGBoost).

[0025] In another preferred embodiment, the machine learning model is a support vector machine (SVM).

[0026] In another preferred embodiment, the method further includes the step of: (S5) validating the male pulmonary sarcoidosis diagnostic model and the female pulmonary sarcoidosis diagnostic model in the male validation set and the female validation set, respectively.

[0027] A second aspect of the present invention provides a sex-specific diagnostic system for pulmonary sarcoidosis, the diagnostic system comprising: An input unit configured to input data, the data including clinical data of the subject being tested; A diagnostic unit, configured as a male or female diagnostic model, diagnoses the subject based on the subject's gender and clinical data using the male or female diagnostic model, thereby obtaining a diagnostic result for the subject; wherein the diagnostic model is constructed using the method described in the first aspect of the present invention; An output unit is configured to output the diagnostic results of the diagnostic unit.

[0028] In another preferred embodiment, the subject of the test is a patient with a lung disease.

[0029] In another preferred embodiment, the subject of the test is a patient with pulmonary sarcoidosis who has atypical imaging findings and / or normal serum angiotensin-converting enzyme levels.

[0030] In another preferred embodiment, the clinical data includes peripheral blood test data.

[0031] A third aspect of the present invention provides an electronic device, including a processor and a memory, the memory having a plurality of executable instructions, the processor being configured to read the instructions and perform the following steps: (1) Provide clinical data of a subject to be tested; the clinical data includes characteristic index values; (2) Input the clinical data into the diagnostic model, and the diagnostic model determines whether the subject to be tested is a patient with pulmonary nodules based on the clinical data, thereby obtaining a risk probability value; If the subject of the test is female, the characteristic indicators include: serum angiotensin-converting enzyme level; CD4+ to CD8+ ratio, lymphocyte percentage, total cholesterol, serum prothrombin time, and serum fibrinogen. If the subject to be tested is male, the characteristic indicators include: albumin level, absolute lymphocyte count, serum angiotensin-converting enzyme level, standard deviation of red blood cell distribution width, erythrocyte sedimentation rate, and CD4+ to CD8+ ratio.

[0032] A fourth aspect of the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when read and executed by a processor, perform the following steps: (1) Provide clinical data of a subject to be tested; the clinical data includes characteristic index values; (2) Input the clinical data into the diagnostic model, and the diagnostic model determines whether the subject to be tested is a patient with pulmonary nodules based on the clinical data, thereby obtaining a risk probability value; If the subject of the test is female, the characteristic indicators include: serum angiotensin-converting enzyme level; CD4+ to CD8+ ratio, lymphocyte percentage, total cholesterol, serum prothrombin time, and serum fibrinogen. If the subject to be tested is male, the characteristic indicators include: albumin level, absolute lymphocyte count, serum angiotensin-converting enzyme level, standard deviation of red blood cell distribution width, erythrocyte sedimentation rate, and CD4+ to CD8+ ratio.

[0033] A fifth aspect of the present invention provides a computer program product comprising computer-executable instructions, which, when executed by a processor, perform the following steps: (1) Provide clinical data of a subject to be tested; the clinical data includes characteristic index values; (2) Input the clinical data into the diagnostic model, and the diagnostic model determines whether the subject to be tested is a patient with pulmonary nodules based on the clinical data, thereby obtaining a risk probability value; If the subject of the test is female, the characteristic indicators include: serum angiotensin-converting enzyme level; CD4+ to CD8+ ratio, lymphocyte percentage, total cholesterol, serum prothrombin time, and serum fibrinogen. If the subject to be tested is male, the characteristic indicators include: albumin level, absolute lymphocyte count, serum angiotensin-converting enzyme level, standard deviation of red blood cell distribution width, erythrocyte sedimentation rate, and CD4+ to CD8+ ratio.

[0034] It should be understood that, within the scope of this invention, the above-described technical features of this invention and the technical features specifically described below (such as in the embodiments) can be combined with each other to form new or preferred technical solutions. Due to space limitations, they will not be described in detail here. Attached Figure Description

[0035] Figure 1 The feature selection process and results for male (A, C, and E) and female (B, D, and F) models are shown.

[0036] Figure 2 The visualization analysis of the male model is shown (SHAP value visualization). A: Dependency graph; B: Histogram; C and E: Individual force graphs; D: Cellular graph.

[0037] Figure 3 The visualization analysis of the female model is shown (SHAP value visualization). A: Dependency graph; B: Bar chart; C and E: Individual force graphs; D: Cellular graph.

[0038] Figure 4 The comparison of ROC curves, calibration curves, and decision curves for Logistic Regression (Logistics), K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Naive Bayes (NB) models on the training set is shown for both male and female models.

[0039] Figure 5The comparison of ROC curves, calibration curves, and decision curves for the male and female models of Logistic Regression (Logistics), K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Naive Bayes (NB) on the validation set is shown respectively. Detailed Implementation

[0040] Through extensive and in-depth research, the inventors have developed for the first time a gender-stratified model and system for the identification and diagnosis of pulmonary sarcoidosis. Specifically, this invention identified six indicators in the male dataset (absolute lymphocyte count (Lym#), standard deviation of red blood cell distribution width (RDW-SD), albumin (ALB), serum angiotensin-converting enzyme (sACE), CD4+ / CD8+ ratio (CD4+ / CD8+), and erythrocyte sedimentation rate (ESR)), and six indicators in the female dataset (lymphocyte percentage (Lym%), serum angiotensin-converting enzyme (sACE), total cholesterol (TC), CD4+ / CD8+ ratio (CD4+ / CD8+), serum fibrinogen (FIBs), and serum prothrombin time (TTs)). Based on these indicators, six machine learning algorithms were used to construct diagnostic models for pulmonary sarcoidosis. The support vector machine-based diagnostic model showed an AUC greater than 0.8 on both the training and validation sets, demonstrating good discriminative ability and the ability to sensitively and accurately identify whether patients with normal serum angiotensin-converting enzyme levels are pulmonary sarcoidosis patients. The model and system of this invention can complete differential diagnosis using only a few feature indicators, greatly shortening the time for relevant patient examination decisions. Based on this, this invention was completed.

[0041] It should be understood that the specific methods and experimental conditions of the invention described below in varying degrees of detail are intended to provide a substantive understanding of the invention. Definitions of certain terms used in this specification are provided below. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0042] the term As used herein, the terms “containing” or “including (comprise)” can be open-ended, semi-closed, or closed-ended. In other words, the terms also include “consistently made of” or “made of”.

[0043] As used herein, the term “and / or” refers to and covers any and all possible combinations of one or more of the related listed items.

[0044] As used in this article, the term "significant" means that, in a hypothesis test, the observed effect (such as the difference between the experimental and control groups) is unlikely to be caused solely by random error. A hypothesis test includes: the null hypothesis (H0), which assumes that the observed effect does not exist (such as no difference between the experimental and control groups); the p-value, which is the probability of observing the current or more extreme effect when H0 is true; and the significance threshold (α). The significance threshold is typically used to determine whether a hypothesis test is significant. Generally, the significance threshold is 0.05. If the p-value ≤ α, then H0 is rejected, meaning the observed effect exists, and the result is called "significant."

[0045] As used herein, the terms "gender stratification" and "gender specificity" are used interchangeably, referring to something that is specifically designed for / exists in a particular gender, or that has a prominent effect only in a particular gender. The diagnostic models and systems of this invention are gender specific; that is, this invention has developed separate diagnostic models for men and women, and these two models only have a prominent diagnostic effect within their respective gender populations.

[0046] Sarcoidosis Sarcoidosis is a systemic granulomatous disease characterized by non-caseating necrotizing epithelioid cell granulomas that can affect multiple organs throughout the body, with the lungs being the most commonly affected site. Sarcoidosis is classified into five stages according to the Scadding staging system. Early pulmonary sarcoidosis (stages 0, I, and II) lacks specific imaging features on chest CT, making diagnosis difficult; approximately 25% of patients experience chronic disease progression. Without timely intervention, these patients often experience increased mortality due to respiratory failure, infection, involvement of vital organs, or complications in later stages. Currently, the diagnosis of sarcoidosis faces several challenges: firstly, its pathological granulomatous features can also be seen in tuberculosis, non-tuberculous mycobacterial lung diseases, fungal infections, etc., requiring manual identification and exclusion; secondly, the sensitivity and specificity of serum angiotensin-converting enzyme (sACE), a potential biomarker for sarcoidosis, are insufficient for effective diagnosis; and thirdly, existing diagnostic assessment tools for sarcoidosis have insufficient diagnostic and differential diagnostic capabilities in multicenter cohorts of non-infectious granulomatous diseases. Therefore, there is an urgent need in this field for a scoring tool or system that can specifically and sensitively diagnose and differentiate sarcoidosis, especially one that can specifically and sensitively distinguish pulmonary sarcoidosis patients with normal sACE levels from other patients with lung diseases, thereby aiding in differential diagnosis and clinical decision-making.

[0047] As used in this article, the term "early pulmonary sarcoidosis" refers to pulmonary sarcoidosis patients with indistinct pathological features, normal serum angiotensin-converting enzyme levels, and radiological staging of stage 0-II.

[0048] The diagnostic model and its construction method of the present invention This invention provides a diagnostic model and a method for constructing the same. Specifically, the method includes the following steps: (S1) Provide a clinical dataset of disease patients; the disease patients are those with atypical imaging manifestations, including patients with pulmonary sarcoidosis and patients without pulmonary sarcoidosis; divide the clinical dataset into a male clinical dataset and a female clinical dataset; further divide the male clinical dataset into a male training set and a male validation set, and further divide the female clinical dataset into a female training set and a female validation set; (S2) The clinical dataset is preprocessed to obtain a preprocessed male clinical dataset and a preprocessed female clinical dataset; (S3) Extract feature indicators from the preprocessed male clinical dataset and the preprocessed female clinical dataset respectively, and filter the feature indicators to obtain preferred feature indicators for males and preferred feature indicators for females. (S4) The preferred male feature indicators in the male training set and the preferred female feature indicators in the female training set are used to train the machine learning model, respectively; when the training result reaches the predetermined termination condition, the training of the model is terminated, thereby obtaining the male pulmonary sarcoidosis diagnosis model and the female pulmonary sarcoidosis diagnosis model.

[0049] Preferably, "atypical imaging features" refers to imaging features that do not meet the typical diagnostic criteria for the disease, and / or imaging features that make it difficult to make a clear diagnosis based on them.

[0050] Preferably, the pulmonary sarcoidosis patients include those with normal serum angiotensin-converting enzyme (sACE) levels. Here, "normal serum angiotensin-converting enzyme level" means that the serum angiotensin-converting enzyme level is below the upper limit of the reference range, i.e., <70 IU / L. In this invention, regardless of whether it is the male or female training and validation sets, the subjects' serum angiotensin-converting enzyme levels are all below 50 IU / L, classifying them as pulmonary sarcoidosis patients who are difficult to diagnose based on serum angiotensin-converting enzyme levels.

[0051] Preferably, in step (S3), the optimal set of characteristic indicators for males and females can be obtained through univariate analysis and LASSO regression analysis. Specifically, in the LASSO regression, K-fold cross-validation is used to optimize the training set.

[0052] Preferably, this paper uses six machine learning models for model training. The machine learning models include: logistic regression, support vector machine (SVM), Naive Bayes (NB), K nearest neighbors (KNN), extreme gradient boosting (XGBoost), and random forest (RF).

[0053] The term "logistic regression" is a classic statistical model that uses the sigmoid function to map the linear regression results to the (0,1) interval, establishing a non-linear relationship between features and target probabilities. Its core lies in constructing a linear decision boundary. ; through the sigmoid function σ(z) = 1 / (1+e -z To achieve probability transformation, the maximum likelihood estimation is used to optimize the weight parameter w.

[0054] The term "Support Vector Machine" is a binary classification model. Its basic idea is to find a hyperplane that separates data points of different classes while maximizing the distance (margin) between the data points of each class to the hyperplane. SVMs can handle not only linearly separable problems but also non-linearly separable problems through kernel functions.

[0055] The term "Naive Bayes," also known as the Naive Bayes classifier, is based on Bayes' theorem and the assumption of conditional independence of features. Bayes' theorem is an important theorem in probability theory that describes the probability of an event occurring given certain conditions. The calculation formula is: .

[0056] The term "K-Nearest Neighbors" is a basic instance-based supervised learning algorithm that can be used for both classification and regression tasks. Its core idea is "like attracts like"—similar feature vectors tend to have similar output values, and the prediction result is determined by "voting" or "averaging" the neighboring samples.

[0057] The term "extreme gradient boosting" can be used interchangeably with "XGBoost," and it's an efficient implementation of the Gradient Boosting Decision Tree (GBDT) algorithm. It incorporates several optimizations and improvements over traditional GBDT, using an additive model to sequentially build multiple weak learners (usually decision trees) and combine them into a strong predictor. Each new tree aims to correct the residual error of the previous tree. The calculation formula is: , ,in, Let L be the predicted output of the Kth tree, and let Ω be the loss function. L measures the prediction error (such as mean squared error or cross-entropy). Ω controls the complexity of a single tree.

[0058] The term "random forest" is an ensemble learning method based on Bagging (Bootstrap Aggregating) and Random Subspace, which improves the robustness of the model by constructing multiple decision trees and combining their predictions.

[0059] Preferably, the machine learning model is a support vector machine.

[0060] Finally, this invention constructs male and female diagnostic models using male-preferred and female-preferred indicators, respectively. The male-preferred indicators include: (A1) albumin (ALB); (A2) absolute lymphocyte count (Lym#); (A3) serum angiotensin-converting enzyme (sACE); (A4) standard deviation of erythrocyte distribution width (RDW-SD); (A5) erythrocyte sedimentation rate (ESR); (A6) CD4+ / CD8+ ratio. The female-preferred indicators include: (B1) serum angiotensin-converting enzyme (sACE); (B2) CD4+ / CD8+ ratio (CD4+ / CD8+); (B3) percentage of lymphocytes (Lym%); (B4) total cholesterol (TC); (B5) serum prothrombin time (TTs); (B6) serum fibrinogen (FIBs).

[0061] To verify the necessity of constructing a sex-specific pulmonary sarcoidosis model, this study mixed male and female datasets and used the same feature selection and model building methods to construct a sex-neutral diagnostic model, identifying a common combination of feature indicators. The diagnostic model based on these common feature indicators could only diagnose male subjects more accurately, but could not improve the diagnostic sensitivity and accuracy for female subjects. Although there were common indicators, such as sACE, between the preferred male and female feature indicators, their contributions to the model predictions were different for males and females, meaning they affected the model's prediction results to varying degrees. Furthermore, stratifying the dataset by sex and modeling separately for each sex yielded better differential diagnostic results.

[0062] Preferably, the diagnostic model of the present invention is applicable to patients with lung diseases whose imaging manifestations are atypical, and is particularly suitable for differentiating between patients with pulmonary sarcoidosis and those without. In practical use, the diagnostic model of the present invention involves collecting peripheral blood (such as blood or plasma) from the subject, detecting the level / concentration / proportion of various characteristic indicators in the peripheral blood, and inputting these indicators into the model to diagnose the subject.

[0063] The sex-specific pulmonary sarcoidosis diagnostic system of the present invention This invention provides a gender-specific diagnostic system for pulmonary sarcoidosis. The system can be embodied as an electronic device, including but not limited to: smartphones, tablets, personal computers, servers, terminals, or other intelligent terminals with data processing capabilities. The system can also be embodied as a computer-readable storage medium, such as a disk, optical disk, solid-state drive, read-only memory, or flash memory, on which a computer program is stored. When the program is executed by one or more processors, it can implement the data processing flow. Furthermore, the system can also be embodied as a computer program product containing computer instructions that, when executed by a computer, cause the computer to perform all or part of the steps of the data processing flow.

[0064] The data processing flow can be loaded, deployed, or run on any of the aforementioned electronic devices. Through the collaboration of hardware resources and the software logic defined in the flow, the system can complete specific data acquisition, transmission, calculation, analysis, storage, or presentation tasks.

[0065] The main advantages of this invention include: (1) This invention is the first to develop a sex-based model for the identification and diagnosis of pulmonary sarcoidosis. The model has an AUC greater than 0.8 in both the training and validation sets for different sexes, and has good diagnostic and identification capabilities for early pulmonary sarcoidosis when serum angiotensin-converting enzyme is still at a normal level.

[0066] (2) Compared with the diagnostic model that has not been constructed with gender stratification, the diagnostic model based on gender stratification of the present invention has better sensitivity and accuracy.

[0067] (3) Through feature extraction and screening, this invention identified six feature indicators that significantly contribute to model performance in both men and women. The male indicators include absolute lymphocyte count (Lym#), red blood cell distribution width standard deviation (RDW-SD), albumin (ALB), serum angiotensin-converting enzyme (sACE), CD4+ / CD8+ ratio (CD4+ / CD8+), and erythrocyte sedimentation rate (ESR). The female indicators include lymphocyte percentage (Lym%), serum angiotensin-converting enzyme (sACE), total cholesterol (TC), CD4+ / CD8+ ratio (CD4+ / CD8+), serum fibrinogen (FIBs), and serum prothrombin time (TTs). The model constructed using these feature indicators exhibits superior performance, overcoming the limitations of single-indicator diagnostic efficacy.

[0068] (4) The system of the present invention can complete identification and diagnosis using a small number of feature indicators, which greatly shortens the time for relevant patients to make examination decisions.

[0069] The present invention will be further illustrated below with reference to specific embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. It is also understood that the purpose of describing the present invention in conjunction with the embodiments is to cover other options or modifications that may be derived based on the claims of the present invention. To provide a deep understanding of the invention, many specific details will be included in the following description. The invention may also be practiced without using these details. Furthermore, to avoid confusion or obscuring the focus of the invention, some specific details will be omitted in the description.

[0070] Data sources and methods 1. Data Sources and Arrangement: This is a retrospective observational study. Data were obtained from the electronic medical record system of Shanghai Pulmonary Hospital affiliated to Tongji University. The study included patients with early pulmonary sarcoidosis and non-pulmonary sarcoidosis (tuberculosis, lung cancer, pneumoconiosis, lymphoma) who were first hospitalized between January 2012 and May 2021 and whose imaging manifestations were atypical and difficult to diagnose. The following indicators were independently assessed by two physicians at the deputy director level or above from the tuberculosis department and the respiratory department: difficulty of disease diagnosis (based on clinical manifestations, imaging features and pathological results), hilar / mediastinal lymph node enlargement (assessed by HRCT or PET-CT), and number of infiltrated lung fields.

[0071] Enter the patient’s demographic characteristics (age and gender), clinical diagnosis, pathological diagnosis, and first peripheral blood test data upon admission into the hospital’s electronic medical record system. This includes complete blood count (white blood cells, neutrophils, lymphocytes, etc.), biochemical indicators (TP, LDL-C, etc.), inflammatory indicators (erythrocyte sedimentation rate, C-reactive protein, etc.), and coagulation (PT, APTT, D-dimer, etc.).

[0072] The criteria for determining the difficulty of disease diagnosis and the inclusion criteria for cases in this study are as follows: According to pathology, the staging of mediastinal lymph node tuberculosis can be divided into four stages. In pathological stages 1 and 2, the CT manifestations of mediastinal lymph node tuberculosis are relatively atypical, which may make it difficult to distinguish from the mediastinal lymph node enlargement in sarcoidosis. At the same time, extensively calcified lymph nodes are considered as easy to distinguish, while partially calcified lymph nodes are considered as difficult to distinguish. Sarcoidosis can be divided into 5 stages according to the Scadding staging system. In stages 0, I, and II, the CT and imaging manifestations are relatively atypical. At the same time, lymph node calcification presents as amorphous or punctate calcification (partial calcification), and hilar / mediastinal lymph node enlargement occurs. Pneumoconiosis is difficult to diagnose and differentiate from other diseases based solely on imaging findings. Lung cancer cases are characterized by multiple invasive lesions in the lungs and involvement of multiple lymph nodes, and are morphologically similar to sarcoidosis cases. Lymphoma (especially Hodgkin's lymphoma (HL) and non-Hodgkin's lymphoma (NHL)) presents with enlargement and fusion of hilar / mediastinal lymph nodes; intrapulmonary lymphatic infiltration. On enhanced CT, lymphoma appears as mild to moderate homogeneous enhancement, similar to proliferative tuberculous lymphadenopathy and sarcoidosis.

[0073] Inclusion criteria: newly diagnosed cases with no history of related diseases (such as tuberculosis or tumors) or treatment (hormones, antibiotics not used within two weeks) before admission. The initial imaging findings were atypical (without a clear diagnosis), and the peripheral blood test results were complete at the first admission. All cancer markers tested were negative, and the patient met the following criteria: (1) Sarcoidosis group: meets the international consensus diagnostic criteria for stage 0-II pulmonary sarcoidosis; (2) Tuberculosis group: negative Mycobacterium tuberculosis smear at admission, negative molecular diagnosis (Mycobacterium tuberculosis DNA or RNA test), but positive Mycobacterium tuberculosis culture and species identification during the follow-up period, or positive smear after at least 1 month of follow-up; (3) Lymphoma group: selected cases with uniform enhancement and less fusion on imaging, and confirmed by histopathology during the follow-up period; (4) Lung cancer group: clear pathological diagnosis during the follow-up period; (5) Pneumoconiosis group: clear history of occupational exposure, but atypical imaging findings (other lung diseases need to be excluded), and confirmed as pneumoconiosis during the follow-up period.

[0074] Exclusion criteria (those meeting any of the following conditions will be excluded): (1) Treatment interference: those who have received anti-tuberculosis, anti-tumor or hormone therapy or have been diagnosed with a confirmed disease before admission, or those who have used antibiotics other than cephalosporins within 2 weeks; (2) Typical imaging findings (such as lobulation sign, spiculation sign, etc.); (3) Any of the diseases in the differential diagnosis or other malignant tumors or active infections (such as HIV, HBV); (4) Data missing, key clinical indicators (such as complete blood count, angiotensin-converting enzyme, imaging report) are incomplete.

[0075] 2. Model selection and evaluation: This study selected six machine learning algorithms—Naive Bayes, Logistic Regression, Support Vector Machine, XGBoost, K-Nearest Neighbors, and Random Forest—to develop a differential diagnostic model for pulmonary sarcoidosis among lung diseases. The diagnostic model was built using data from various indicators obtained through infiltrative lung fields, hilar / mediastinal lymph node enlargement, clinical symptoms, age, sex, and peripheral blood tests. The dataset was randomly divided into a training set (model building) and a validation set (independent testing) at a ratio of 7:3. The model's discriminative ability was evaluated using the following metrics: Area Under the Receiver Operating Characteristic (AUC), Sensitivity, Specificity, Positive Predictive Value (PPV), Negative Predictive Value (NPV), Recall, F1 score, and Accuracy. Clinical applicability: Decision curve analysis (DCA) was used to assess the model's net benefit. Interpretability: Shapley Additive Explanations (SHAP) analysis was used to visualize the model and enhance its interpretability.

[0076] 3. Statistical analysis and feature selection: Missing data were imputed using multiple imputation methods. For measurement data (such as BMI and laboratory indicators) that conformed to a normal distribution, the values ​​were expressed as mean ± standard deviation (Mean ± SD) and analyzed using an independent samples t-test; for those that did not conform to a normal distribution, the values ​​were expressed as median (M) and interquartile range (IQR) and analyzed using the Mann-Whitney U test. Categorical data (such as gender and past medical history) were described as frequencies (percentages) and analyzed using a chi-square test (χ²) or Fisher's exact test (if the expected frequency < 5). This study employed a two-stage feature selection strategy: the first stage used univariate analysis (P < 0.10) to initially screen potential predictive indicators; the second stage used LASSO regression for feature selection and ranking, where LASSO regression used 10-fold cross-validation to determine the optimal penalty coefficient (λ), retaining variables with non-zero coefficients. The final indicators were selected based on the ranking of coefficients (β values) and clinical significance. Model evaluation and visualization used ROC curves to evaluate the diagnostic efficacy of individual indicators and combined models. SPSS 26.0 was used for basic statistical analysis, and R 4.4.3 (RStudio) was used for machine learning modeling, SHAP analysis, and visualization. SPSS 26.0 and RStudio were used for data analysis, and for creating images and charts.

[0077] Example 1: Characteristics of Research Participants A total of 1142 cases of various lung diseases were collected for the study. Based on the inclusion and exclusion criteria and randomization, 319 cases were ultimately included (156 males and 163 females). Among them, 119 cases were pulmonary sarcoidosis, and 200 cases were in the control group (78 cases of tuberculosis, 78 cases of lung cancer, 13 cases of lymphoma, and 32 cases of pneumoconiosis). The training set for males consisted of 112 cases, and the validation set consisted of 44 cases; the training set for females consisted of 115 cases, and the validation set consisted of 48 cases. The mean age of males with pulmonary sarcoidosis was 42.81 years (SE: 11.31), and the median age of females was 54.00 years (IQR: 48.00–62.00). The mean age of males without sarcoidosis was 50.48 years (17.84 years), and the median age of females was 54.00 years (40.75–60.00).

[0078] Table 1. Univariate analysis p<0.10 and baseline clinical characteristics (training set) - male Table 2. Univariate analysis p<0.10 and baseline clinical characteristics (training set) - female Example 2: Feature Extraction LASSO analysis is used to reduce the dimensionality of indicators that are statistically significant in univariate analysis, thereby enabling the selection of indicators and adjustment of complexity.

[0079] Figure 1 The process and results of feature selection for male and female models are shown. The feature variables are sorted according to the absolute value of the coefficients (β values) in the LASSO regression results. Combined with clinical significance, six feature variables were finally selected for model construction.

[0080] Example 3: SHAP Analysis and Interpretation SHAP was used to analyze each feature. The SHAP visualization results are as follows: Figure 2 and Figure 3 As shown.

[0081] Figure 2 A and Figure 3 A presents a scatter plot showing partial dependencies among the six most influential variables to illustrate the relationships between them. Figure 2 B and Figure 3B shows the mean absolute SHAP value of each feature across all predictions, a metric for feature importance. The six extracted features were ranked according to their discriminative factors, with albumin (ALB) being the most important factor for males and serum angiotensin-converting enzyme (sACE) being the most important factor for females. Furthermore, Figure 2 C and Figure 3 C and Figure 2 E and Figure 3 E shows the waterfall plot and individual force plot, providing representative prediction cases of early pulmonary sarcoidosis to illustrate the interpretability of the model. Yellow arrows indicate an increase in predicted values, and purple arrows indicate a decrease in predicted values. Figure 2 D and Figure 3 Figure D shows the peak group plot, which also shows the impact of multiple features and reveals the relationship between features and predicted values. The darker the yellow, the higher the risk value, while the darker the purple, the lower the risk value.

[0082] Example 4: Model Construction This embodiment involves training a model on a training set and optimizing the model through cross-validation.

[0083] The collected data undergoes data cleaning, including handling missing values, outlier handling, data transformation, and standardization; standardization formula. Where, min(X): minimum value of data, max(X): maximum value of data.

[0084] All participants were divided into training and validation sets in a 7:3 ratio. Six variables, determined by LASSO regression combined with clinical significance, were used. Six machine learning algorithms—logistics, support vector machine (SVM), k-nearest neighbor (KNN), Naive Bayes (NB), random forest (RF), and gradient boosting (XGBoost)—were employed to construct and validate predictive models for the training set based on routine peripheral blood biometrics analysis of pulmonary nodules and non-pulmonary nodule (pneumoconiosis, lung cancer, tuberculosis, lymphoma) data. During training, 10-fold cross-validation was used to fine-tune the models.

[0085] The AUC curves of the features used in the training of each model showed that Random Forest, XGBoost, and Support Vector Machine performed well. Table 1 summarizes the overall performance of each model. Figure 4 A and Figure 4 B shows the ROC curves of each model on the male and female training sets. The calibration curves show the relationship between the predicted probabilities of each model and the actual observations. Figure 4 C and Figure 4D). The final DCA results show that when the threshold probabilities are between 0.10 and 0.08, all six models predict a net return for STME greater than 0 in the training set. Among the training set models, Random Forest and XGBoost provide the highest net returns for most threshold ranges. Figure 4 E and Figure 4 F).

[0086] Table 3. Overall performance of each model in the training set Example 5: Validation of Model Performance This embodiment involves validating the model obtained in Example 4 on a validation set. The results show that the Support Vector Machine model performs best on the validation set, with an AUC of 0.888 for males and 0.825 for females. The specificity for discrimination and diagnosis on both the training and validation sets is higher than 70%. Table 2 lists a summary of the overall performance of each model on the validation set. Figure 5 The ROC curves for each model on the validation set are shown. The calibration curves show the relationship between the predicted probabilities of each model and the actual observations. Figure 4 C and Figure 4 D). DCA results show that when the threshold probability is between 0.10 and 0.08, the net return of STME predicted by all six models is also greater than 0 in the validation set.

[0087] The support vector machine (SVM) model provided the highest net benefit on the validation set, indicating that it has good clinical utility. Figure 5 ).

[0088] Table 4. Overall performance of each model in the validation set Example 6: Validation of External Data According to the inclusion and exclusion criteria, the accuracy of the detection model was collected from 97 sarcoidosis cases (65 females and 32 males) between May 2021 and April 2025. The accuracy of the male model was 72% and the accuracy of the female model was 68%.

[0089] Example 7: Feature Model Built Based on Mixed Gender Dataset and Its Performance Based on the aforementioned feature selection and model construction methods, a machine learning differential diagnostic model was developed using the full dataset. The final included features were: triglyceride to high-density lipoprotein cholesterol ratio (TG / HDL-C), transferrin (TRF), sACE, absolute lymphocyte count (Lym), prothrombin time (TTs), fibrinogen (FIBs), and CD4+ / CD8+. Table 5 shows that after gender stratification, the male model showed significant improvement, while the diagnostic efficacy of the female model remained similar.

[0090] Table 5. Performance of the entire dataset on the training and validation sets discuss This study constructed a differential diagnostic model for early-stage pulmonary sarcoidosis (PSD) with other lung diseases exhibiting atypical imaging features. This is the first machine learning-based differential diagnostic model for PSD based on gender differences. Focusing on real-world diagnostic challenges in clinical practice, this study validated the value of using combined biomarkers with machine learning algorithms in improving differential diagnostic efficacy and clinical application through cases with insignificant imaging features rigorously selected by two clinicians, using readily available routine laboratory indicators. In this study, the population was stratified by gender for modeling. The indicators used in the male model were absolute lymphocyte count, erythrocyte distribution width, albumin, sACE, CD4+ / CD8+, and erythrocyte sedimentation rate (ESR); the indicators used in the female model were Lym%, sACE, total cytogen (TC), CD4+ / CD8+, fibrillary intraepithelial blots (FIBs), and total thromboses (TTs). Finally, using the optimal machine learning model—support vector machine (AUC of 0.888 for male validation set and 0.825 for female validation set)—a website for the differential diagnosis of early-stage PSD was constructed for clinical use.

[0091] Sarcoidosis has an unknown etiology and lacks a gold standard diagnosis, making it difficult to differentiate clinically from other diseases with similar imaging findings. Furthermore, the risks and benefits of invasive biopsies are difficult to weigh. Machine learning, a crucial branch of artificial intelligence, has yielded numerous breakthroughs in laboratory medicine. Machine learning techniques possess powerful computational methods capable of processing complex and high-dimensional data, identifying and summarizing meaningful information from datasets to facilitate the development of clinical diagnostic models.

[0092] The potential influence of biological sex on disease mechanisms is the fundamental reason for the sex-stratified study design used in this study (Çolak Y et al., *Thorax*, 2025;80(8):512). Nevertheless, male and female models share some common features: lower lymphocyte counts, higher sACE levels (still within the normal range), and higher CD4+ / CD8+ ratios. The reduced lymphocyte count may be associated with more severe cellular immune dysfunction in sarcoidosis compared to other differentially identified diseases (Chailleux E et al., *Thorax*, 1985;40(10):768-73). The presence of elevated sACE levels and CD4+ / CD8+ ratios in sarcoidosis patients has been documented in previous studies. A meta-analysis concluded that the sensitivity and specificity of sACE in diagnosing sarcoidosis were 60% (95% CI: 52%-68%) and 93% (95% CI: 88%-96%), respectively. Some sarcoidosis patients have sACE levels at the high end of the normal range, which is related to ACE gene insertion (I) / deletion (D) polymorphism (ACE I / D) (Kruit A, Respir Med, 2007;101(3):510-5). A study by Csongrádi et al. found that combining sACE activity detection with ACE I / D genotyping can increase the specificity of sarcoidosis diagnosis to 100%, but the sensitivity is relatively low at 42.5% (Csongrádi A et al., Clin Chem Lab Med, 2018;56(7):1117-25).

[0093] In this invention, although the sACE levels in all cases were within the normal range, sACE ranked third in importance in the male model and first in the female model. This may be due to the higher prevalence of pneumoconiosis (which also leads to elevated sACE) in men (Blanco-Pérez J et al., Pulmonology, 2024;30(4):370-7), which indirectly emphasizes the necessity of gender stratification when constructing diagnostic models. Gender stratification and sACE level restrictions improved model performance, resulting in validation set sensitivity of 0.818 (male) and 0.950 (female), and specificity of 0.882 (male) and 0.714 (female).

[0094] Multiple studies have shown that an elevated CD4+ / CD8+ ratio in bronchoalveolar lavage fluid (BAL) has diagnostic value for sarcoidosis. However, the invasive risks of BAL sampling limit its clinical application. This study demonstrates that combining the peripheral blood CD4+ / CD8+ ratio with other readily available indicators can construct a high-performance diagnostic model.

[0095] Studies have shown that patients with sarcoidosis have dyslipidemia. An observational and Mendelian randomized metabolite study on the relationship between lipids and sarcoidosis confirmed that high-density lipoprotein cholesterol (HDL-C) and total cholesterol (TC) are closely associated with the disease (Zhan Y et al., *Respir Res*, 2024;25(1):50), which may stem from abnormal lipid metabolism in macrophages during the pathogenesis of sarcoidosis (Hutton A et al., *Am J Respir Crit Care Med*, 2024;209(9):1064-6). Consistent with this, this study screened two indicators reflecting lipid metabolism status using LASSO (Shin HG et al., *Korean J FamMed*, 2017;38(6):352-7): apolipoprotein E (APOE) and the triglyceride to HDL ratio (TG / HDL). In the female cohort, replacing APOE with TC, which significantly contributed to the female model (ranked third in importance), improved model performance. Lipid metabolism and its relationship with sarcoidosis warrant further investigation.

[0096] The coagulation markers selected in both models exhibited interesting heterogeneity in male and female patients: antithrombin III levels were elevated in men, while lower in women in the training set of this study. The study by TW Meade et al. (Meade TW et al., *Br J Haematol*, 1990;74(1):77-81) showed that premenopausal women had lower antithrombin III activity than age-matched men, but menopause was associated with a significant increase in antithrombin III levels in women. In this study cohort, the median age of female patients with pulmonary sarcoidosis was 54.0 years (IQR 48.0–62.0), falling within the perimenopausal age range. In contrast, male patients were younger, with a mean age of 42.81 years (standard deviation 11.31), approximately ten years younger than women—a pattern consistent with the age characteristics of sarcoidosis diagnosis reported in Swiss and Italian populations (Arkema EV et al., *Ther Adv Chronic Dis*, 2018;9(11):227-40). While age differences may partially explain the differences in antithrombin III between the two models, the possibility that coagulation markers have a sex-specific role in the pathophysiology of sarcoidosis cannot be ruled out.

[0097] This article is the first to focus on the differential diagnosis of lung disease patients whose angiotensin-converting enzyme levels are within the normal range and difficult to identify. The model and system described in this article are constructed and have strong clinical reference value for research in lung disease specialty hospitals.

[0098] All documents mentioned in this invention are incorporated herein by reference as if each document were individually incorporated by reference. Furthermore, it should be understood that after reading the foregoing teachings of this invention, those skilled in the art can make various alterations or modifications to this invention, and these equivalent forms also fall within the scope defined by the appended claims.

Claims

1. A method for constructing a sex-specific diagnostic model for pulmonary sarcoidosis, characterized in that, The method includes the following steps: (S1) Provide a clinical dataset of disease patients; the disease patients are those with atypical imaging manifestations, including patients with pulmonary sarcoidosis and patients without pulmonary sarcoidosis; divide the clinical dataset into a male clinical dataset and a female clinical dataset; further divide the male clinical dataset into a male training set and a male validation set, and further divide the female clinical dataset into a female training set and a female validation set; (S2) The clinical dataset is preprocessed to obtain a preprocessed male clinical dataset and a preprocessed female clinical dataset; (S3) Extract feature indicators from the preprocessed male clinical dataset and the preprocessed female clinical dataset respectively, and filter the feature indicators to obtain preferred feature indicators for males and preferred feature indicators for females. (S4) The preferred male feature indicators in the male training set and the preferred female feature indicators in the female training set are used to train the machine learning model, respectively; when the training result reaches the predetermined termination condition, the training of the model is terminated, thereby obtaining the male pulmonary sarcoidosis diagnosis model and the female pulmonary sarcoidosis diagnosis model.

2. The method as described in claim 1, characterized in that, The patients with pulmonary sarcoidosis include those with pulmonary sarcoidosis and normal serum angiotensin-converting enzyme (sACE) levels.

3. The method as described in claim 2, characterized in that, The phrase "normal serum angiotensin-converting enzyme level" means that the serum angiotensin-converting enzyme level is below the upper limit of the reference range (<70 IU / L).

4. The method as described in claim 1, characterized in that, Step (S3) specifically includes the following steps: (S3.1) Univariate analysis was performed on the preprocessed male clinical dataset and the preprocessed female clinical dataset respectively, and indicators with P < 0.1 were retained; (S3.2) Perform LASSO regression analysis on the indicators obtained in step (S3.1) to obtain the regression coefficients of the male candidate indicator set and the female candidate indicator set and each indicator in the candidate indicator set respectively; (S3.3) Sort the male candidate indicator set and the female candidate indicator set according to the absolute value of the regression coefficient, and select the indicators with the highest absolute value of the regression coefficient and clinical significance, thereby obtaining the preferred characteristic indicators for males and the preferred diagnostic indicators for females respectively.

5. The method as described in claim 1, characterized in that, The preferred male characteristic indicators are selected from the following group: (A1) Albumin (ALB); (A2) Absolute lymphocyte count (Lym#); (A3) Serum angiotensin-converting enzyme (sACE); (A4) Standard deviation of red blood cell distribution width (RDW-SD); (A5) Erythrocyte sedimentation rate (ESR); (A6) CD4+ to CD8+ ratio (CD4+ / CD8+), or a combination thereof.

6. The method as described in claim 1, characterized in that, The preferred female characteristics are selected from the following group: (B1) serum angiotensin-converting enzyme (sACE); (B2) CD4+ to CD8+ ratio (CD4+ / CD8+); (B3) lymphocyte percentage (Lym%); (B4) total cholesterol (TC); (B5) serum prothrombin time (TTs); (B6) serum fibrinogen (FIBs), or a combination thereof.

7. A sex-specific diagnostic system for pulmonary sarcoidosis, characterized in that, The pulmonary sarcoidosis diagnostic system includes: An input unit configured to input data, the data including clinical data of the subject being tested; A diagnostic unit, configured as a male or female diagnostic model, diagnoses the subject based on the subject's gender and clinical data using the male or female diagnostic model, thereby obtaining a diagnostic result for the subject; wherein the diagnostic model is constructed using the method described in claim 1. An output unit configured to output the diagnostic results of the diagnostic unit; In another preferred embodiment, the subject of the test is a patient with a lung disease; In another preferred embodiment, the subject of the test is a patient with pulmonary sarcoidosis who has atypical imaging findings and / or normal serum angiotensin-converting enzyme levels.

8. An electronic device comprising a processor and a memory, characterized in that, The memory contains multiple executable instructions, and the processor is used to read the instructions and perform the following steps: (1) Provide clinical data of a subject to be tested; the clinical data includes characteristic index values; (2) Input the clinical data into the diagnostic model, and the diagnostic model determines whether the subject to be tested is a patient with pulmonary nodules based on the clinical data, thereby obtaining a risk probability value; If the subject of the test is female, the characteristic indicators include: serum angiotensin-converting enzyme level; CD4+ to CD8+ ratio, lymphocyte percentage, total cholesterol, serum prothrombin time, and serum fibrinogen. If the subject to be tested is male, the characteristic indicators include: albumin level, absolute lymphocyte count, serum angiotensin-converting enzyme level, standard deviation of red blood cell distribution width, erythrocyte sedimentation rate, and CD4+ to CD8+ ratio.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when read and executed by a processor, perform the following steps: (1) Provide clinical data of a subject to be tested; the clinical data includes characteristic index values; (2) Input the clinical data into the diagnostic model, and the diagnostic model determines whether the subject to be tested is a patient with pulmonary nodules based on the clinical data, thereby obtaining a risk probability value; If the subject of the test is female, the characteristic indicators include: serum angiotensin-converting enzyme level; CD4+ to CD8+ ratio, lymphocyte percentage, total cholesterol, serum prothrombin time, and serum fibrinogen. If the subject to be tested is male, the characteristic indicators include: albumin level, absolute lymphocyte count, serum angiotensin-converting enzyme level, standard deviation of red blood cell distribution width, erythrocyte sedimentation rate, and CD4+ to CD8+ ratio.

10. A computer program product comprising computer-executable instructions, characterized in that, When the computer-executable instructions are executed by the processor, the following steps are performed: (1) Provide clinical data of a subject to be tested; the clinical data includes characteristic index values; (2) Input the clinical data into the diagnostic model, and the diagnostic model determines whether the subject to be tested is a patient with pulmonary nodules based on the clinical data, thereby obtaining a risk probability value; If the subject of the test is female, the characteristic indicators include: serum angiotensin-converting enzyme level; CD4+ to CD8+ ratio, lymphocyte percentage, total cholesterol, serum prothrombin time, and serum fibrinogen. If the subject to be tested is male, the characteristic indicators include: albumin level, absolute lymphocyte count, serum angiotensin-converting enzyme level, standard deviation of red blood cell distribution width, erythrocyte sedimentation rate, and CD4+ to CD8+ ratio.

Images