Early prediction model for severe and critical old new crown patients and application of early prediction model

Abstract

The invention relates to the technical field of medical data analysis, and discloses a severe and critical early prediction model and application of an old new crown patient, and the method comprises the following steps: obtaining detection values of C-reactive protein, D-dimer and lymphocyte percentage of a to-be-detected old new crown patient; substituting the detection value into a preset linear equation to calculate a total risk score; and comparing the total risk score with a preset diagnosis critical value to assess the risk level of the patient progressing into a heavy type or a critical type. The system comprises an information acquisition unit, a calculation unit configured with a prediction model, an evaluation unit and a result display unit to cooperatively execute the method. According to the method, rapid and objective prediction of risks is realized through the quantitative model, a reliable basis is provided for early clinical intervention, accurate hierarchical management of patients is facilitated, and decision support data is provided for optimizing medical resource allocation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical data analysis, in particular to an early prediction model for severe and critical cases of elderly COVID-19 patients and application thereof. BACKGROUND

[0002] After COVID-19 infection, the clinical manifestations of patients are wide-ranging. A part of patients, especially elderly patients aged 60 years or older, will progress from mild or moderate to severe or critical. The condition of severe and critical patients often deteriorates rapidly, accompanied by respiratory failure or multiple organ dysfunction, and the mortality rate is significantly higher than that of non-severe patients. The elderly group is at high risk of developing serious complications due to their physiological characteristics and immune function, as well as the universality of multiple underlying diseases. Therefore, objectively assessing the risk of disease progression of elderly COVID-19 patients in the early stage of the disease and identifying individuals with a tendency to develop severe symptoms are a technical problem to be solved in clinical practice.

[0003] To solve this problem, some prediction models based on clinical characteristics, laboratory indicators or imaging data have been disclosed in the prior art. However, these technical solutions have certain limitations. Many models are not specifically designed for the physiological and pathological characteristics of the elderly group, and their training data come from patients of all ages, which affects the accuracy of their prediction when applied to the elderly sub-group. At the same time, some models rely on a single laboratory indicator for judgment, which may not fully reflect the complex pathophysiological process of the disease, resulting in insufficient prediction efficiency. Some other models combine multiple indicators, but the sample data on which they are based is single, which limits the universality of the model. In addition, although imaging examination can provide direct evidence of lung lesions, in the early stage of the disease, the lung images of some elderly patients may not change significantly, affecting the effectiveness of early warning. Some technical solutions use complex algorithm models, which require specific computing equipment to run, which poses obstacles to their deployment and application in primary medical institutions.

[0004] Therefore, the present application proposes an early prediction model for severe and critical cases of elderly COVID-19 patients and application thereof to solve the deficiencies of the prior art. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides an early prediction model for severe and critical cases of elderly COVID-19 patients and application thereof, which solves the problem of the lack of a rapid, objective and accurate early prediction of the risk of severe illness in elderly COVID-19 patients.

[0006] To solve the above technical problems, the present application provides the following technical solutions: The first aspect of the present application provides a method for constructing an early prediction model for severe and critical cases of elderly COVID-19 patients, comprising the following steps: S1, obtaining a clinical data set of a group of elderly COVID-19 patients, the clinical data set comprising demographic information, combined underlying disease information, and multiple laboratory index detection values covering inflammation, coagulation and blood routine, and dividing the clinical data set into a severe or critical group and a non-severe or critical group according to the severity of the disease of the patients; S2, using a single factor analysis method, performing correlation analysis on the multiple laboratory indexes and the disease severity grouping, and preliminarily screening out candidate indexes with statistical significance; S3, including the candidate indexes and other clinical related variables into a multi-factor logistic regression model for analysis, screening and finally determining that the combination of C-reactive protein, D-dimer and lymphocyte percentage is an independent risk index for predicting the progression of elderly COVID-19 patients to severe or critical cases; S4, based on the three independent risk indexes of C-reactive protein, D-dimer and lymphocyte percentage, establishing an early prediction model for calculating the total risk score by a linear equation.

[0007] In a preferred embodiment, in the step S4, the early prediction model is defined by the following linear equation: ; wherein, represents the total risk score, represents the detection value of lymphocyte percentage, represents the detection value of C-reactive protein, represents the detection value of D-dimer; 、 、 、 is the model coefficient determined by the multi-factor logistic regression model analysis.

[0008] In a specific embodiment, the model coefficient is respectively: ; ; ; .

[0009] Further, the construction method further comprises: determining the diagnostic critical value of the total risk score based on the receiver operating characteristic curve analysis of the early prediction model; wherein the total risk score greater than 0.41 is determined as high risk, and the total risk score less than or equal to 0.41 is determined as low risk.

[0010] Preferably, in the step S1, the elderly COVID-19 patient is a person aged ≥ 60 years.

[0011] The second aspect of the present application provides an early prediction model for severe and critical COVID-19 in elderly patients, which is obtained by any of the construction methods of the first aspect.

[0012] The early prediction model takes C-reactive protein, D-dimer, and lymphocyte percentage as input variables, and calculates the total risk score by a linear equation.

[0013] The third aspect of the present application provides an early prediction method for severe and critical COVID-19 in elderly patients, which applies the early prediction model of the second aspect, and comprises the following steps: (1) obtaining the detection values of C-reactive protein, D-dimer, and lymphocyte percentage of the elderly COVID-19 patient to be tested; (2) substituting the detection values obtained into the linear equation of the early prediction model to calculate the total risk score; (3) comparing the total risk score with a preset diagnostic critical value to evaluate the risk level of the elderly COVID-19 patient to be tested developing into severe or critical type.

[0014] The fourth aspect of the present application provides an early prediction system for severe and critical COVID-19 in elderly patients, which is configured with the early prediction method of the third aspect, and comprises: a detection unit for detecting the levels of C-reactive protein, D-dimer, and lymphocyte percentage in the sample of the elderly COVID-19 patient to be tested; an information acquisition unit for acquiring the detection values of C-reactive protein, D-dimer, and lymphocyte percentage obtained by the detection unit; a calculation unit configured with the early prediction model, for calculating the total risk score by a linear equation according to the detection values acquired by the information acquisition unit; an evaluation unit for judging the risk level of the elderly COVID-19 patient to develop into severe or critical type according to the total risk score calculated by the calculation unit and a preset diagnostic critical value; a result display unit for displaying the risk level conclusion obtained by the evaluation unit; the result display unit displays the conclusion by screen display, voice broadcast, or printing.

[0015] The fifth aspect of the present application provides a computer readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the early prediction method of the third aspect.

[0016] The sixth aspect of the present application provides the use of a combination of C-reactive protein, D-dimer and lymphocyte percentage in the preparation of a prediction product for evaluating the risk of a severe or critical COVID-19 patient aged 60 or above developing into a severe or critical state.

[0017] The present application provides an early prediction model and application for severe and critical COVID-19 patients in the elderly. The following beneficial effects are provided: 1. The present application determines the specific combination of C-reactive protein, D-dimer and lymphocyte percentage from a plurality of laboratory indicators covering inflammation, coagulation and routine blood tests through a screening step, which are independent risk indicators for predicting the progression of elderly COVID-19 patients to severe or critical state. The technical solution provides a specific, objective and specific risk-related biomarker combination, providing a specific technical basis for establishing a quantitative early prediction model.

[0018] 2. The early prediction model established by the present application uses a linear equation containing specific coefficients to directly convert the detection values of the three indicators into a quantitative risk total score. The model structure is simple, the calculation process is clear, and an objective and continuous risk assessment result can be output, thereby avoiding the uncertainty that may be caused by relying entirely on the subjective experience of clinicians.

[0019] 3. The technical solution of the present application only needs to obtain C-reactive protein, D-dimer and lymphocyte percentage, which are three laboratory indicators that are popular and quickly detected in medical institutions at all levels. Based on these easily obtained input data, the model constructed can quickly calculate the risk level, which helps to stratify the management of the prognosis risk of patients at an early stage of hospitalization, and provides a decision basis for the rational allocation of medical resources and the timely adjustment of treatment plans. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 Flowchart for screening the research object of the present application; Figure 2 Nomogram for the prediction model of the present application; Figure 3 Receiver operating characteristic curve graph for the prediction model of the present application; Figure 4 System architecture diagram of the present application.

[0021] Among them, 10, detection unit; 20, information acquisition unit; 30, calculation unit; 40, evaluation unit; 50, result display unit. DETAILED DESCRIPTION

[0022] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0023] For the convenience of description, the specific terms and English abbreviations used in the embodiments of the present application are defined as follows: CRP: C-reactive protein; DD: D-dimer; LYM: percentage of lymphocytes; COPD: chronic obstructive pulmonary disease; Pro-BNP: brain natriuretic peptide precursor; IL-6: interleukin-6; RR: respiratory rate; PaO2: arterial oxygen partial pressure; FiO2: oxygen concentration; AUC: area under the curve.

[0024] The present application provides an embodiment of a method for early prediction of severe and critical cases of elderly COVID-19 patients, which is applied to an early prediction model and includes the following steps: Obtaining the detection values of C-reactive protein, D-dimer and percentage of lymphocytes of the elderly COVID-19 patient to be tested; Substituting the obtained detection values into the linear equation of the early prediction model to calculate the total risk score; Comparing the total risk score with a preset diagnostic critical value to evaluate the risk level of the elderly COVID-19 patient to be tested developing into a severe or critical case.

[0025] The prediction system described in the present application can be a standalone computing device or can be integrated into an existing medical information system (HIS) or laboratory information system (LIS). The information acquisition unit 20, the calculation unit 30, the evaluation unit 40 and the result display unit 50 can implement their respective functions by executing computer program instructions stored on a non-transitory computer readable storage medium through one or more processors. The system also includes a bus for data transmission and an interface for communication with external devices (such as detection devices, displays, printers).

[0026] Reference Figures 1-4 The present application also provides an early prediction system for severe and critical cases of elderly COVID-19 patients. In a specific embodiment, the system includes: A detection unit 10 for detecting the levels of C-reactive protein, D-dimer and percentage of lymphocytes in the sample of the elderly COVID-19 patient to be tested; An information acquisition unit 20 for obtaining the detection values of C-reactive protein, D-dimer and percentage of lymphocytes obtained by the detection unit 10; A computing unit 30 configured with the early prediction model built by the method of the present application, for calculating the total risk score by linear equation according to the detection values obtained by the information acquisition unit 20; An evaluation unit 40 for judging the risk level of the severe and critical progression of the elderly COVID-19 patient according to the total risk score calculated by the computing unit 30 and the preset diagnostic critical value; A result display unit 50 for displaying the risk level conclusion obtained by the evaluation unit 40. The result display unit 50 outputs the conclusion by screen display, voice broadcast or printing.

[0027] In a specific embodiment, the acquisition and grouping process of the clinical data set for building the prediction model is as follows.

[0028] Reference Figure 1 The flowchart shows the screening process of the research subjects. First, an initial cohort containing 722 elderly COVID-19 patients hospitalized in Nanfang Hospital is obtained. The elderly patients are all greater than or equal to 60 years old. The initial cohort is grouped according to the following diagnosis and classification criteria.

[0029] All the patients included in the study meet the following criteria for the diagnosis of COVID-19 infection: 1. Having clinical manifestations related to COVID-19 infection; 2. And having one or more of the following etiological or serological test results: (1) Positive result of COVID-19 nucleic acid detection; (2) Positive result of COVID-19 antigen detection; (3) Positive result of COVID-19 isolation and culture; (4) The level of COVID-19 specific IgG antibody in the recovery period is 4 times or more higher than that in the acute phase.

[0030] After being determined as a COVID-19 patient, the patient is grouped according to the severity of the disease at the time of admission. The grouping criteria are as follows: The criteria for determining a severe patient are as follows: any one of the following conditions is met, and cannot be explained by other causes other than COVID-19 infection: 1. Shortness of breath, respiratory rate (RR) ≥ 30 times / min; 2. In a resting state, the oxygen saturation is ≤ 93% when inhaling air; 3. The ratio of arterial oxygen partial pressure (Pa02) to oxygen concentration (Fi02) is ≤ 300 mmHg (1 mmHg = 0.133 kPa). For areas above 1000 meters above sea level, the following correction formula is used to correct Pa02 / Fi02: ; wherein, representative arterial partial pressure of oxygen detection value, representative oxygen concentration, actual atmospheric pressure measurement value of the patient's location, in millimeters of mercury; 4. Progressive clinical deterioration and radiographic evidence of > 50% progression of the lesion within 24 to 48 hours.

[0031] The criteria for determining a critical patient are as follows: 1. Respiratory failure and the need for mechanical ventilation; 2. Shock; 3. Combined with other organ failure, requiring intensive care unit (ICU) monitoring and treatment.

[0032] According to the above criteria, patients who meet any of the criteria for severe or critical illness are classified as severe or critical, and the remaining patients are classified as non-severe or non-critical.

[0033] For all patients included, the clinical data set was collected by checking the electronic medical record system. The data set includes demographic information, combined with basic disease information, and multiple laboratory index detection values completed within 48 hours after admission. Among them, the demographic information includes gender, age, vital signs (respiration, pulse, temperature, systolic pressure, diastolic pressure) at admission, smoking history and drinking history. The combined basic disease information includes the prevalence of hypertension, diabetes, coronary heart disease, chronic obstructive pulmonary disease (COPD), chronic kidney disease, hepatitis B, fatty liver, rheumatic immune disease, hematopathy and cancer.

[0034] Laboratory indicators cover multiple aspects such as inflammation, coagulation and blood routine, including: white blood cell count, total monocyte count, hematocrit, red blood cell count, lymphocyte percentage (LYM), total lymphocyte count, total neutrophil count, hemoglobin determination value, platelet count, C-reactive protein (CRP), procalcitonin, brain natriuretic peptide precursor (Pro-BNP), interleukin-6 (IL-6), high-sensitivity troponin, D-dimer (DD), albumin, alanine aminotransferase, aspartate aminotransferase, bilirubin, total protein, electrolytes (potassium, chloride, magnesium, sodium, calcium), renal function indicators (urea, creatinine, glomerular filtration rate), etc. These data constitute the original data set for subsequent analysis.

[0035] After obtaining and grouping the clinical data set, in order to screen out indicators with independent correlation with disease severity from multiple laboratory indicators and clinical information, the embodiment performs further data analysis steps.

[0036] Firstly, all collected variables, including demographic information, combined basic disease information and various laboratory indicators, were analyzed for correlation with disease severity groups (severe or critical group vs. non-severe or critical group) one by one using single-factor analysis method to preliminarily screen out candidate indicators with statistical significance. Table 1 shows the comparison of baseline clinical characteristics between 403 non-severe or critical patients and 319 severe or critical patients.

[0037] Table 1: Baseline table On the basis of single-factor analysis, to exclude the confounding effects between variables, the candidate indicators with statistical significance preliminarily screened out and other clinically relevant variables were included in a multivariate ordinal logistic regression model for analysis to screen and finally determine independent risk indicators. Table 2 lists the results of single-factor and multivariate ordinal logistic regression analysis of factors affecting the severity of COVID-19 in elderly patients, where OR represents odds ratio and 95% CI represents 95% confidence interval.

[0038] Table 2: Ordinal logistic regression analysis of factors affecting the severity of COVID-19 in elderly patients According to the multivariate analysis results shown in Table 2, variables with P value less than 0.05 were determined as independent influencing factors. The analysis results showed that the combination of C-reactive protein (CRP), D-dimer (DD) and lymphocyte percentage (LYM) was an independent risk indicator for predicting the progression of elderly COVID-19 patients to severe or critical type. These indicators will be used as core input variables for building subsequent prediction models.

[0039] Based on the three independent risk indicators of C-reactive protein (CRP), D-dimer (DD) and lymphocyte percentage (LYM) screened and determined in the foregoing steps, the embodiment constructs a quantitative model for early prediction.

[0040] Reference Figure 2FIG. 1 is a schematic diagram of a nomogram prediction model based on three indicators constructed using R software in one embodiment of the present application. The nomogram converts a complex regression equation into a visual graph, including multiple scale axes. Specifically, the top of the graph is the Points score axis; below it are the scale axes of the three input variables LYM, CRP, and D_Dimer; below it is the Total Points axis; at the bottom are the Linear-Predictor axis corresponding to the total score and the final risk probability axis.

[0041] The operation steps for risk assessment using the nomogram are as follows: first, obtain the actual detection values of LYM, CRP, and D_Dimer of the patient to be tested, and find the position of the value on the respective scale axis in Figure 2 the middle; then, draw a vertical line from the position to the top Points score axis, and read the score corresponding to the variable; add the respective scores of the three variables to obtain a total score; finally, find the position of the total score on the Total-Points axis, draw a vertical line from the position to the bottom risk probability axis, and read the value as the predicted risk probability of the patient developing into severe or critical type.

[0042] To facilitate direct deployment and execution in a computing device, the core linear equation is extracted from the nomogram model. Based on the model intercept and regression coefficients of each variable obtained by multi-factor logistic regression analysis, a linear prediction equation for calculating the risk total score is constructed, as follows: ; wherein, represents the risk total score, represents the detection value of the percentage of lymphocytes, represents the detection value of C-reactive protein, represents the detection value of D-dimer; 、 、 、 is the model coefficient determined by multi-factor logistic regression model analysis.

[0043] In one specific embodiment, the model coefficients are as follows: ; ; ; .

[0044] To verify the prediction performance of the constructed model, the receiver operating characteristic curve (ROC) method is used for evaluation. Referring to Figure 3FIG. 4 is a receiver operating characteristic curve plot drawn after the model was validated using the clinical dataset of Southern Hospital in one embodiment of the present application. The area under the curve (AUC) value is 0.753.

[0045] Based on the analysis of the receiver operating characteristic curve, a diagnostic critical value for distinguishing risk levels was determined. In this embodiment, the diagnostic critical value is 0.41. Accordingly, the following risk stratification criteria were established: when the total risk score calculated by the linear equation above is greater than 0.41, the patient is assessed as high risk (suggesting a trend of progression to severe or critical type); when the calculated total risk score is less than or equal to 0.41, the patient is assessed as low risk (suggesting mild or moderate type).

[0046] To more specifically illustrate the application process of the prediction method of the present application, an application example is given below.

[0047] In a specific application scenario, a 68-year-old male patient was diagnosed with COVID-19 infection, and the prediction method of the present application was performed to assess the risk of early progression of his hospitalization.

[0048] First, laboratory test information within 48 hours after admission of the patient was obtained. Through the test, the detection values of three core indicators were obtained: lymphocyte percentage (LYM) was 8.0%, C-reactive protein (CRP) was 70 mg / L, and D-dimer (DD) was 2.0 mg / L.

[0049] Second, the obtained three detection values above were substituted into the preset total risk score calculation equation. The equation is: ; Substituting the detection values of the patient, the calculation process is as follows: ; Thus, the total risk score of the patient was calculated to be 0.267.

[0050] Third, the calculated total risk score was compared with the preset diagnostic critical value. In this embodiment, the preset diagnostic critical value is 0.41.

[0051] Fourth, the risk level conclusion was output according to the comparison result. Since the total risk score of the patient 0.267 is less than the diagnostic critical value 0.41, according to the preset risk stratification criteria, the patient is assessed as low risk, suggesting that the possibility of his condition progressing to severe or critical type at the current stage is low. This conclusion can provide an objective basis for the subsequent treatment decision and resource allocation of the clinician. ​​​

[0052] The present application also provides a system for performing the above-mentioned prediction method. The specific workflow of the system is completed through the cooperative operation of its internal units, which will be described below in conjunction with an application example.

[0053] With reference to Figure 4 In a specific application scenario, the system is deployed in the clinical laboratory or clinical information system of a hospital.

[0054] First, the detection unit 10 (for example, a fully automatic biochemical analyzer or a blood cell analyzer) processes the blood sample of a to-be-tested elderly COVID-19 patient, detects and outputs the levels of C-reactive protein (CRP), D-dimer (DD) and lymphocyte percentage (LYM). Assuming that the detection results are: CRP = 150 mg / L, DD = 4.0 mg / L, and LYM = 5.0%.

[0055] Subsequently, the information acquisition unit 20, which can be a data interface connected to the detection device or a manual input interface, acquires the detection values of the above-mentioned three indicators. The unit transmits the acquired values (150, 4.0, 5.0) to the calculation unit 30 in the form of structured data.

[0056] The calculation unit 30 is a processor or calculation module configured with the prediction model of the present application. After receiving the data transmitted by the information acquisition unit 20, the calculation unit 30 automatically performs the preset linear equation to calculate: ; The calculation unit 30 obtains the total risk score of the patient as 1.19 and transmits this calculation result to the evaluation unit 40.

[0057] The evaluation unit 40 has a preset diagnosis critical value of 0.41. After receiving the total risk score of 1.19, the unit performs a comparison operation: comparing 1.19 with 0.41. Since 1.19 is greater than 0.41, the evaluation unit 40 determines that the patient belongs to the high-risk level according to the internal logic rules.

[0058] Finally, the evaluation unit 40 transmits the conclusion of high risk to the result display unit 50. The result display unit 50 can be a display screen, a printer or a voice broadcast module. The received conclusion is output in a preset form, for example, displaying the risk level in a prominent way on the display screen: high risk, or directly printing a paper report containing patient information, indicator values, total risk score and final risk level. The output result can be directly viewed by medical staff for clinical decision-making assistance.

[0059] The embodiment of the present application also provides a computer readable storage medium, which stores a computer program. The computer program is configured to implement the aforementioned early prediction method of severe and critical cases of elderly COVID-19 patients when executed by one or more processors.

[0060] The present application also provides an embodiment relating to a computer readable storage medium. The medium stores a computer program, which, when executed by a processor of a computing device, enables the computing device to perform the aforementioned early prediction method of severe and critical cases of elderly COVID-19 patients. The method comprises: obtaining the detection values of C-reactive protein, D-dimer and lymphocyte percentage of a subject to be tested; based on the detection values, calculating the total risk score by a linear equation and comparing the score with a preset diagnostic threshold value 0.41 to determine the risk level. The computer readable storage medium can be any electronic, magnetic, optical or other physical device or means of storing program codes, such as read-only memory (ROM), random access memory (RAM), flash memory, hard disk or optical disc.

[0061] The present application also provides an embodiment relating to a kit for early prediction of severe and critical cases of elderly COVID-19 patients. In a specific implementation, the kit comprises: a detection reagent for detecting the level of C-reactive protein (CRP); a detection reagent for detecting the level of D-dimer (DD); a reagent for counting or determining the percentage of lymphocytes in the sample, such as a hemolytic agent or a fluorescently labeled antibody; and an instruction manual, which records the operation steps of the detection using the kit and clearly gives the linear equation for calculating the total risk score and the risk stratification standard based on the diagnostic threshold value 0.41.

[0062] Although the embodiments of the present application have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and alterations can be made without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for constructing an early prediction model for severe and critically ill elderly COVID-19 patients, characterized in that, Includes the following steps: S1. Obtain a set of clinical datasets of elderly COVID-19 patients. The clinical datasets include demographic information, information on underlying diseases, and multiple laboratory indicators covering inflammation, coagulation, and routine blood tests. Based on the severity of the patients' diseases, the clinical datasets are divided into severe or critical group and non-severe or critical group. S2. Univariate analysis was used to analyze the correlation between multiple laboratory indicators and disease severity groups, and candidate indicators with statistical significance were preliminarily screened. S3. The candidate indicators and other clinically relevant variables are included in a multivariate logistic regression model for analysis. The combination of C-reactive protein, D-dimer and lymphocyte percentage is finally determined to be an independent risk indicator for predicting the progression of elderly COVID-19 patients to severe or critical illness. S4. Based on the three independent risk indicators of C-reactive protein, D-dimer and lymphocyte percentage, establish an early prediction model that calculates the total risk score using a linear equation.

2. The method for constructing an early prediction model for severe and critically ill elderly COVID-19 patients according to claim 1, characterized in that, Step S4, which involves establishing an early prediction model that calculates the total risk score using a linear equation, includes the following steps: The early prediction model is defined by the following linear equation: ； in, Represents the total risk score. The test value representing the percentage of lymphocytes. The measured value represents C-reactive protein. The detection value represents D-dimer; , , , These are the model coefficients determined through multifactor logistic regression model analysis.

3. The method for constructing an early prediction model for severe and critically ill elderly COVID-19 patients according to claim 2, characterized in that, The model coefficients are as follows: ； ； ； 。 4. The method for constructing an early prediction model for severe and critically ill elderly COVID-19 patients according to claim 3, characterized in that, The construction method also includes: Based on the receiver operating characteristic curve analysis of the early prediction model, the diagnostic threshold of the total risk score is determined. Specifically, a total risk score greater than 0.41 is considered high risk, while a total risk score less than or equal to 0.41 is considered low risk.

5. The method for constructing an early prediction model for severe and critically ill elderly COVID-19 patients according to claim 1, characterized in that, Step S1, which involves obtaining a clinical dataset of elderly COVID-19 patients, includes the following steps: The elderly COVID-19 patients mentioned refer to people aged 60 years and older.

6. An early prediction model for severe and critical COVID-19 patients in the elderly, applied to the construction method described in any one of claims 1-5, characterized in that, The early prediction model uses C-reactive protein, D-dimer, and lymphocyte percentage as input variables and calculates the total risk score through a linear equation.

7. A method for early prediction of severe and critical cases of COVID-19 in elderly patients, applied to the early prediction model described in claim 6, characterized in that, Includes the following steps: (1) Obtain the detection values ​​of C-reactive protein, D-dimer and lymphocyte percentage in elderly COVID-19 patients to be tested; (2) Substitute the obtained detection values ​​into the linear equation of the early prediction model to calculate the total risk score; (3) Compare the total risk score with a preset diagnostic threshold to assess the risk level of the elderly COVID-19 patient to be tested progressing to severe or critical condition.

8. An early prediction system for severe and critical COVID-19 patients in the elderly, applied to the early prediction method described in claim 7, characterized in that, include: The detection unit is used to detect the levels of C-reactive protein, D-dimer, and lymphocyte percentage in samples from elderly COVID-19 patients. An information acquisition unit is used to acquire the detection values ​​of C-reactive protein, D-dimer, and lymphocyte percentage obtained by the detection unit; The calculation unit is equipped with an early prediction model, which is used to calculate the total risk score through a linear equation based on the detection values ​​obtained by the information acquisition unit. An assessment unit is used to determine the risk level of severe and critical COVID-19 progression in elderly COVID-19 patients based on the total risk score calculated by the calculation unit and a preset diagnostic threshold. The results display unit is used to display the risk level conclusions reached by the assessment unit. The result display unit displays the conclusions through screen display, sound broadcast, or printing.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the early prediction method as described in claim 7.

10. Application of the combination of C-reactive protein, D-dimer and lymphocyte percentage in the preparation of a predictive product for assessing the risk of progression to severe or critical COVID-19 in elderly patients aged ≥60 years.

Images