MCI risk assessment system and method based on neural immune multilevel weighted scoring

Abstract

The invention discloses a construction method of an MCI risk assessment system based on neural immune multilevel weighted scoring, which comprises the following steps of: respectively scoring virus load, virus type quantity, neural immunological response and neural immune cardiovascular indexes, comprehensively calculating a total score by adopting a preset weight, and constructing a multilevel weighted scoring prediction model; training is carried out through a machine learning algorithm, prediction model parameters with accuracy and specificity reaching 95% or above are obtained, an MCI risk assessment system is constructed, and the system is used for MCI risk assessment. According to the method, the problems of single index, fixed threshold and lack of multi-dimensional comprehensive evaluation in the prior art are solved, the scoring rule is scientific and has a dynamic adjustment mechanism, the prediction accuracy and specificity of MCI are greatly improved, early MCI patients can be more accurately identified, and powerful technical support is provided for early intervention and precision medical treatment.

Description

Technical Field

[0001] The present invention relates to a MCI risk assessment system and method based on neuroimmune multi-level weighted scoring. Background Art

[0002] Mild cognitive impairment (MCI) is a precursor to Alzheimer's disease, and its early and accurate identification has significant clinical value in slowing disease progression. Existing technologies have clear clinical diagnostic criteria for Alzheimer's disease, generally based on a 42 / 40 ratio of beta-amyloid protein. However, the diagnosis of MCI is currently primarily based on subjective judgment by doctors using standard cognitive scales and experience. There is still no unified clinical quantitative diagnostic standard for MCI.

[0003] The current expert consensus is that viral infection may cause neuroinflammation, which in turn develops into MCI. Existing technologies generally use real-time fluorescence quantitative PCR to detect the nucleic acid CT value of human herpes virus (HHV) and conduct risk assessments for MCI based on viral indicators. Testing items include detecting the nucleic acid CT value of individual viruses such as HHV1 and HHV2. A simple judgment standard is set based on the preset CT value range (such as CT value <36 or >36), which is directly associated with positive or negative results, and the risk of MCI is judged accordingly.

[0004] This single evaluation method has the following drawbacks:

[0005] 1. A single indicator has limitations: It relies solely on the CT value of viral nucleic acid for evaluation, and fails to comprehensively consider multi-dimensional information such as viral load, virus type, neuroimmune response, and cardiovascular disease. This leads to inaccurate prediction results, and it is difficult to achieve a high level of sensitivity and specificity (for example, above 95%).

[0006] 2. Fixed thresholds and scoring systems are inflexible: Using fixed thresholds or single-tier scoring systems makes it impossible to reasonably allocate weights and dynamically adjust the clinical significance of different indicators, making it difficult to take into account the contribution of different detection indicators in MCI risk prediction.

[0007] 3. Lack of collaborative evaluation: It is difficult to reflect the pathological status under the joint action of multiple factors by detecting a single indicator, resulting in a high risk of misdiagnosis or missed diagnosis, which affects clinical early warning and subsequent intervention.

[0008] Therefore, existing technologies have obvious limitations in the early diagnosis and risk assessment of MCI, and there is an urgent need for an assessment method that can integrate multi-dimensional information, flexibly adjust weights, and dynamically optimize. Summary of the Invention

[0009] The purpose of the present invention is to propose an MCI assessment system and method based on multi-level weighted scoring. The system includes four major modules, which score viral load, number of viral types, neuroimmunological response and neuroimmune cardiovascular indicators respectively, and use preset weights to comprehensively calculate the total score, thereby achieving high accuracy and high specificity of MCI risk prediction.

[0010] The technical solution adopted in the present invention is as follows:

[0011] A method for constructing an MCI risk assessment system based on a neuroimmune multi-level weighted scoring system, the method comprising the following steps:

[0012] S1. Collect sample data, including human herpes virus nucleic acid CT value, virus species and number, neuroimmune cardiovascular indicators and clinical MCI diagnosis data, and divide the data into training set and test set after preprocessing;

[0013] The human herpes virus nucleic acid CT value refers to the detected CT value of HHV1, HHV2, HHV3, HHV6, HHV8, HHV4, HHV5, and HHV7;

[0014] The number of virus species refers to the types and numbers of human herpes viruses detected;

[0015] The neuroimmune cardiovascular index was aldosterone concentration;

[0016] S2. Construct a multi-level weighted scoring prediction model. Divide the sample data in the training set into four scoring modules, namely: human herpes virus nucleic acid CT value, number of virus species, virus combination and CT value, and aldosterone concentration. Each module is scored according to a preset interval. The scores of the four modules are weighted according to the preset weights to calculate the total score:

[0017] Total score = (score of the first module × R1) + (score of the second module × R2) + (score of the third module × R3) + (score of the fourth module × R4)

[0018] R1, R2, R3, and R4 are the preset weight parameters of the first module, the second module, the third module, and the fourth module respectively, and the sum of R1, R2, R3, and R4 is 1;

[0019] Compare the total score with the preset threshold. If the total score is lower than the preset threshold, a high risk is output; if the total score is greater than or equal to the preset threshold, a low risk is output.

[0020] The output risk results were compared with actual clinical MCI diagnostic data, and the accuracy and specificity were calculated, with the lower value of the two used as the evaluation standard. The training set was trained using a machine learning algorithm to obtain prediction model parameters with accuracy and specificity exceeding 95%, and the scoring rules, segmentation thresholds, weight parameters, and preset thresholds of the optimized scoring module were obtained.

[0021] Accuracy is the ratio of the number of samples correctly predicted by the model to the total number of samples, and specificity refers to the ratio of the number of samples correctly predicted to have low MCI risk to the total number of samples with low MCI risk, which represents the ability of the model to correctly identify low-risk MC.

[0022] S3. Use the test set data to verify and evaluate the prediction model, further optimize the model, obtain the optimal combination of scoring rules, segmentation thresholds, weight parameters, and preset thresholds, and construct an MCI risk assessment system for MCI risk assessment;

[0023] Furthermore, in step S3, the accuracy, precision, and specificity are calculated using the data from the test set, the ROC curve is drawn, the AUC is calculated, and the model is verified and evaluated, requiring the AUC, accuracy, and specificity of the prediction model to meet the preset requirements; if the accuracy and specificity of the prediction model do not meet the preset requirements, the model is re-iterated until the accuracy and specificity of the prediction model meet the preset requirements.

[0024] Generally speaking, the preset requirements for AUC, accuracy, and specificity of the prediction model are:

[0025] AUC ≥ 0.80, accuracy ≥ 0.95, and specificity ≥ 0.95.

[0026] In step S3, the MCI risk assessment system can achieve an accuracy and specificity of more than 95% in different populations and scenarios.

[0027] The method may further include step S4: establishing a regular parameter correction and model iteration process, updating the evaluation system according to the newly input clinical sample data, and adjusting the parameters in real time to ensure that the system can be continuously optimized and improved according to the clinical data during actual application, and can achieve an accuracy and specificity of more than 95% in different populations and scenarios.

[0028] In step S2, the training set is trained using a machine learning algorithm. Commonly used machine learning algorithms include logistic regression, random forest, gradient boosting, etc. The clinical sample data of the training set are trained using the machine learning algorithm, and regression analysis and cross-validation are performed on the scoring rules of the scoring module, the segmentation threshold, and the weight parameters of the weighted calculation module. The parameters are dynamically adjusted and corrected to ensure that the prediction accuracy and specificity of the final evaluation system reach more than 95%.

[0029] In step S2, assigning points according to a preset interval means assigning points within a range of 0 to 100 to each scoring module.

[0030] The scoring principle is that the higher the score, the lower the risk.

[0031] Furthermore, the larger the CT value of viral nucleic acid, the higher the score, but a CT value of 0 is scored as 100;

[0032] When the number of virus types is 0, the score is 100. The more types there are, the lower the score.

[0033] Virus combinations and CT values are scored according to different virus combination types and CT value intervals; the virus combination and CT value represent the level of human neuroimmune response. The lower the immune capacity represented by the virus combination and CT value, the lower the score.

[0034] Higher aldosterone concentrations are associated with higher scores, and lower aldosterone concentrations are associated with lower scores.

[0035] In a preferred embodiment, the scoring rules, segmentation thresholds, weight parameters and preset thresholds of the optimized model are as follows:

[0036] Module 1: CT value of human herpes virus nucleic acid

[0037] Any virus detected and CT value > 36 or CT value 0: score 100

[0038] CT value ≥31 and <36: score 80

[0039] CT value ≥ 25 and < 31: score 50

[0040] CT value <25: score 30

[0041] If multiple viruses are detected in a sample, the maximum CT value of the multiple viruses can be used for scoring.

[0042] Module 2: Number of virus types:

[0043] Quantity is 0: score 100 points

[0044] Quantity is 1: score 80 points

[0045] The number is 2: score 60 points.

[0046] The number is 3: score 40 points.

[0047] Quantity > 3: score 20 points.

[0048] Module 3: Neuroimmunological response assessment, scoring based on virus combination and CT value

[0049] When all virus detection indicators show a CT value of 0, the score is 100 points;

[0050] If one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and none of HHV4, HHV5, or HHV7 are detected, and the CT value of all detected viruses is greater than 36, the score is 90 points;

[0051] If one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and none of HHV4, HHV5, or HHV7 is detected, and the CT value of any of the detected viruses satisfies 0 < CT value < 36, the score is 80 points;

[0052] If one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and one of the following viruses is detected: HHV4, HHV5, HHV7, and the CT value of any of the detected viruses is between 0 < CT value < 36, the score is 70 points;

[0053] If HHV4 and HHV5, or HHV5 and HHV7, or HHV4, HHV5, and HHV7 are detected simultaneously, and the CT value of any of HHV4, HHV5, and HHV7 is between 0 and 36, the score is 70 points.

[0054] When HHV4 and HHV7 are detected at the same time, the CT values of HHV4 and HHV7 both meet 0<CT value<36, and the ratio of the CT value of HHV4 to the CT value of HHV7 is ≥1, the score is 60 points;

[0055] When the ratio of the CT value of HHV4 to the CT value of HHV7 is <1, the score is 40 points;

[0056] If both HHV4 and HHV7 are detected, and the CT value of either virus is less than 25, and the ratio of HHV4 to HHV7 is less than 1, the score is 30 points;

[0057] In the fourth module, the scoring rules for aldosterone concentration intervals are as follows:

[0058] Aldosterone concentration <50 pg / mL, score 40

[0059] Aldosterone concentration ≥50 pg / mL and <100 pg / mL, score 50

[0060] Aldosterone concentration ≥100 pg / mL and <150 pg / mL, score 60

[0061] Aldosterone concentration ≥150 pg / mL and <200 pg / mL, score 80

[0062] Aldosterone concentration ≥200 pg / mL and <250 pg / mL, score 100

[0063] When the aldosterone concentration is >310 pg / ml, it can also be adjusted to 50 points based on statistical data (this range is set according to the clinical warning significance).

[0064] Total score = (score of the first module × 0.10) + (score of the second module × 0.10) + (score of the third module × 0.50) + (score of the fourth module × 0.30)

[0065] The preset threshold is 60, that is:

[0066] If the total score is less than 60, the output is high risk;

[0067] The total score is greater than or equal to 60, and the output is low risk.

[0068] The present invention also provides an MCI risk assessment system based on neuroimmune multi-level weighted scoring, the system comprising:

[0069] (1) a data input module for reading sample data, including human herpes virus nucleic acid CT value, virus species and number, neuroimmune cardiovascular indexes, and clinical MCI diagnosis data; and performing preprocessing;

[0070] The preprocessing includes checking the integrity of the data, and the missing values can be filled or eliminated by the mean value; human herpes viruses include HHV1, HHV2, HHV3, HHV6, HHV8, HHV4, HHV5, HHV7

[0071] The human herpes virus nucleic acid CT value refers to the detected CT value of HHV1, HHV2, HHV3, HHV6, HHV8, HHV4, HHV5, and HHV7;

[0072] The number of virus species refers to the types and numbers of human herpes viruses detected;

[0073] The neuroimmune cardiovascular index was aldosterone concentration;

[0074] Clinical MCI diagnosis data refers to whether the patient was diagnosed with MCI, including MCI-positive (confirmed as MCI) and MCI-negative (not diagnosed as MCI);

[0075] (2) Scoring module, including:

[0076] The first module: viral load scoring module, scoring according to the viral nucleic acid CT value;

[0077] The second module: the virus type quantity scoring module, which scores according to the number of detected virus types;

[0078] The third module: the neuroimmune response scoring module, which scores according to the virus combination and the CT value range;

[0079] The fourth module: the cardiovascular index scoring module, which scores according to the aldosterone concentration range;

[0080] Furthermore, in the first module, the CT value of the virus nucleic acid is scored according to the preset CT value segmentation rule, and the preset CT value segmentation rule can generally be set as:

[0081] Any virus is detected and CT value > P1 or CT value is 0: score 100

[0082] CT value ≥ P2 and < P1: score 80

[0083] CT value ≥ P3 and < P2: score 50

[0084] CT value < P3: score 30

[0085] P1, P2, and P3 are the CT value scoring segmentation thresholds for virus load, and satisfy P1 > P2 > P3; the range is between 15 and 50;

[0086] The segmentation thresholds P1, P2, and P3 are obtained by training the clinical sample database through a machine learning algorithm and can be dynamically optimized to ensure that the model is dynamically adjusted according to the clinical database, guaranteeing high accuracy and specificity.

[0087] If multiple viruses are detected in the sample, the maximum CT value of the multiple viruses (or weighted average) can be used for scoring.

[0088] Furthermore, in the second module, the score is decreased according to the number of detected virus types, and the scoring rule can generally be set as:

[0089] The quantity is 0: score 100

[0090] The quantity is N1: score 80

[0091] The quantity is N2: score 60.

[0092] The quantity is N3: score 40.

[0093] The quantity > N3: score 20.

[0094] N1, N2, and N3 are the virus type quantity scoring segmentation thresholds, and satisfy N1 < N2 < N3; the range is between 1 and 10;

[0095] The segment thresholds N1, N2, and N3 are obtained by training a clinical sample database through a machine learning algorithm and can be dynamically optimized to ensure that the model is dynamically adjusted according to the clinical database, guaranteeing high accuracy and specificity.

[0096] Furthermore, in the third module, the rule for scoring the neuroimmune response is as follows:

[0097] When all detected virus indicators show a CT value of 0, the score is 100 points;

[0098] One or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, none of HHV4, HHV5, HHV7 is detected, and when the CT value of all detected viruses > 36, the score is 90 points;

[0099] One or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, none of HHV4, HHV5, HHV7 is detected, and when the CT value of any one of the detected viruses satisfies 0 < CT value < 36, the score is 80 points;

[0100] One or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and one of the following viruses is detected: HHV4, HHV5, HHV7, and when the CT value of any one of the detected viruses is between 0 < CT value < 36, the score is 70 points;

[0101] HHV4 and HHV5, or HHV5 and HHV7, or HHV4, HHV5, and HHV7 are detected simultaneously, and when the CT value of any one of HHV4, HHV5, and HHV7 is 0 < CT value < 36, the score is 70 points;

[0102] HHV4 and HHV7 are detected simultaneously, the CT values of HHV4 and HHV7 both satisfy 0 < CT value < 36, and when the ratio of the CT value of HHV4 to the CT value of HHV7 ≥ 1, the score is 60 points;

[0103] When the ratio of the CT value of HHV4 to the CT value of HHV7 < 1, the score is 40 points;

[0104] HHV4 and HHV7 are detected simultaneously, the CT values of HHV4 and HHV7 both satisfy 0 < CT value < 36, and when the CT value of any one of them < 25, and the ratio of HHV4 to HHV7 < π, the score is 30 points.

[0105] Furthermore, in the fourth module, the rule for scoring the aldosterone concentration range is as follows:

[0106] Aldosterone concentration < Q1 pg / mL, score 40 points

[0107] The aldosterone concentration is ≥ Q1 pg / mL and < Q2 pg / mL, with a score of 50 points

[0108] The aldosterone concentration is ≥ Q2 pg / mL and < Q3 pg / mL, with a score of 60 points

[0109] The aldosterone concentration is ≥ Q3 pg / mL and < Q4 pg / mL, with a score of 80 points

[0110] The aldosterone concentration is ≥ Q4 pg / mL and < Q5 pg / mL, with a score of 100 points

[0111] When the aldosterone concentration > Q6 pg / ml, it can also be adjusted to 50 points according to statistical data (set according to the clinical warning significance within this range).

[0112] Q1, Q2, Q3, Q4, Q5, and Q6 are the threshold values for segmenting the aldosterone concentration scores, and Q1 < Q2 < Q3 < Q4 < Q5 < Q6; the range is between 30 and 310;

[0113] The segment threshold values Q1, Q2, Q3, Q4, Q5, and Q6 are obtained by training the clinical sample database through a machine learning algorithm and can be dynamically optimized.

[0114] (3) Weighted calculation module, which performs weighted summation according to the scores of each scoring module and the preset weights to calculate the total score;

[0115] Total score = (Score of the first module × R1) + (Score of the second module × R2) + (Score of the third module × R3) + (Score of the fourth module × R4)

[0116] R1, R2, R3, and R4 are the weight parameters of the first module, second module, third module, and fourth module respectively, and the sum of R1, R2, R3, and R4 is 1;

[0117] The weight parameters R1, R2, R3, and R4 are obtained by training the clinical sample database through a machine learning algorithm and can be dynamically optimized to ensure that the model is dynamically adjusted according to the clinical database, guaranteeing high accuracy and specificity.

[0118] (4) Risk assessment module: Compare the total score with the preset threshold and output the MCI risk level;

[0119] If the total score is lower than the preset threshold, output high risk;

[0120] If the total score is greater than or equal to the preset threshold, output low risk.

[0121] The preset threshold is generally between 40 and 80, but may also be further changed according to model optimization.

[0122] In a preferred embodiment, the preset threshold is 60.

[0123] The assessment system may further include a result output module for displaying and saving the scores of each module, the total score and the risk assessment results.

[0124] The scoring rules, segmentation thresholds P1, P2, P3, weight parameters R1, R2, R3, R4 and preset thresholds are obtained by training a clinical sample database through a machine learning algorithm. Specifically, a large-scale clinical database can be used to train, adjust and calibrate the scoring rules, segmentation thresholds, weight parameters and preset thresholds through regression analysis and cross-validation, thereby ensuring that the model can achieve an accuracy and specificity of more than 95% in different populations.

[0125] In a preferred embodiment, the scoring rules, segmentation thresholds, weight parameters, and preset thresholds of the model trained by the machine learning algorithm are as follows:

[0126] Module 1: CT value of human herpes virus nucleic acid

[0127] Any virus detected and CT value > 36 or CT value 0: score 100

[0128] CT value ≥31 and <36: score 80

[0129] CT value ≥ 25 and < 31: score 50

[0130] CT value <25: score 30

[0131] If multiple viruses are detected in a sample, the maximum CT value of the multiple viruses is used for scoring;

[0132] Module 2: Number of virus types:

[0133] Quantity is 0: score 100 points

[0134] Quantity is 1: score 80 points

[0135] Quantity is 2: score 60 points

[0136] Quantity is 3: score 40 points

[0137] Number > 3: score 20 points

[0138] Module 3: Neuroimmunological response assessment, scoring based on virus combination and CT value

[0139] When all virus detection indicators show a CT value of 0, the score is 100 points;

[0140] If one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and none of HHV4, HHV5, or HHV7 are detected, and the CT value of all detected viruses is greater than 36, the score is 90 points;

[0141] If one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and none of HHV4, HHV5, or HHV7 is detected, and the CT value of any of the detected viruses satisfies 0 < CT value < 36, the score is 80 points;

[0142] If one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and one of the following viruses is detected: HHV4, HHV5, HHV7, and the CT value of any of the detected viruses is between 0 < CT value < 36, the score is 70 points;

[0143] If HHV4 and HHV5, or HHV5 and HHV7, or HHV4, HHV5, and HHV7 are detected simultaneously, and the CT value of any of HHV4, HHV5, and HHV7 is between 0 and 36, the score is 70 points.

[0144] When HHV4 and HHV7 are detected at the same time, the CT values of HHV4 and HHV7 both meet 0<CT value<36, and the ratio of the CT value of HHV4 to the CT value of HHV7 is ≥1, the score is 60 points;

[0145] When the ratio of the CT value of HHV4 to the CT value of HHV7 is <1, the score is 40 points;

[0146] If both HHV4 and HHV7 are detected, and the CT value of either virus is less than 25, and the ratio of HHV4 to HHV7 is less than 1, the score is 30 points;

[0147] In the fourth module, the scoring rules for aldosterone concentration intervals are as follows:

[0148] Aldosterone concentration <50 pg / mL, score 40

[0149] Aldosterone concentration ≥50 pg / mL and <100 pg / mL, score 50

[0150] Aldosterone concentration ≥100 pg / mL and <150 pg / mL, score 60

[0151] Aldosterone concentration ≥150 pg / mL and <200 pg / mL, score 80

[0152] Aldosterone concentration ≥200 pg / mL and <250 pg / mL, score 100

[0153] When the aldosterone concentration was >310 pg / ml, the score was 50 points;

[0154] Total score = (score of the first module × 0.10) + (score of the second module × 0.10) + (score of the third module × 0.50) + (score of the fourth module × 0.30)

[0155] The preset threshold is 60, that is:

[0156] If the total score is less than 60, the output is high risk;

[0157] The total score is greater than or equal to 60, and the output is low risk.

[0158] Furthermore, the evaluation system can also include a model training and optimization module, which includes a machine learning algorithm library and a clinical database. The clinical database is trained through a machine learning algorithm to obtain model parameters with an accuracy and specificity of more than 95%, and the scoring rules, segmentation thresholds, and weight parameters of the weighted calculation module of the optimized scoring module are obtained. The system is optimized according to the updated clinical database, and the scoring rules, weight parameters and preset thresholds are updated regularly.

[0159] Furthermore, using the detection data of a large number of samples collected from the clinical database (viral CT values, number of virus types, immune response data, aldosterone concentration) and the final clinical MCI diagnosis data, a machine learning algorithm was used to train a large amount of clinical data, and regression analysis and cross-validation were performed on the scoring rules, segmentation thresholds, and weight parameters of the weighted calculation module of the scoring module. The parameters were dynamically adjusted and corrected to ensure that the final model prediction accuracy and specificity reached more than 95%.

[0160] Commonly used algorithms include logistic regression, random forest, gradient boosting, etc. These algorithms can be trained on large amounts of clinical data and optimize parameters to achieve the desired prediction accuracy and specificity.

[0161] The model training and optimization module is equipped with a parameter update mechanism: a regular parameter correction and model iteration process is established, the model is updated according to the newly input clinical sample data, and the parameters are adjusted in real time to ensure that the system can be continuously optimized and improved according to clinical data during actual application, and can achieve an accuracy and specificity of more than 95% in different populations and scenarios.

[0162] The evaluation system may further include a biological detection module: including a real-time fluorescence quantitative PCR instrument and an immunoassay detection device;

[0163] The real-time fluorescence quantitative PCR instrument is used to detect the CT value of human herpesvirus (HHV) nucleic acid in a sample. Human herpesviruses include HHV1, HHV2, HHV3, HHV6, HHV8, HHV4, HHV5, and HHV7. The real-time fluorescence quantitative PCR instrument has the characteristics of high sensitivity, low noise, and rapid detection.

[0164] The immunoassay detection device includes an enzyme-linked immunosorbent assay (ELISA) or a chemiluminescence immunoassay detector, which is used to detect neuroimmune cardiovascular indicators in serum, and the neuroimmune cardiovascular indicators include the concentration of aldosterone.

[0165] The sample is a blood or tissue sample, and a suitable sample can be selected according to the detection target.

[0166] The biological detection module may also include a virus detection kit, which contains specific primers, probes, reaction buffer, etc. for human herpes virus, and is used to prepare a PCR reaction system for detecting HHV1, HHV2, HHV3, HHV6, HHV8, HHV4, HHV5, and HHV7 viruses, and perform detection on a real-time fluorescence quantitative PCR instrument.

[0167] The biological detection module may further include a sample pretreatment system, which includes nucleic acid extraction reagents, a centrifuge, sample processing tubes, etc., for extracting nucleic acids from samples to ensure that the nucleic acid extraction has high efficiency and purity.

[0168] The present invention also provides a method for assessing MCI risk based on neuroimmune multi-level weighted scoring. The method utilizes the above-mentioned MCI risk assessment system based on neuroimmune multi-level weighted scoring to assess MCI risk. The method comprises the following steps:

[0169] (1) Input the CT value and aldosterone concentration value of human herpesvirus (HHV) nucleic acid in the sample to be tested. Human herpesviruses include HHV1, HHV2, HHV3, HHV6, HHV8, HHV4, HHV5, and HHV7;

[0170] The CT value of the human herpes virus (HHV) nucleic acid can be obtained by real-time fluorescence quantitative PCR detection, and the aldosterone concentration value can be obtained by immunoassay;

[0171] (2) Execute four-dimensional scoring:

[0172] a) Viral load scoring: Select the highest CT value and score according to the preset segmentation rules;

[0173] b) Virus type score: score is given in descending order of the number of virus types detected;

[0174] c) Neuroimmune response score: score is calculated based on the virus combination and CT value range;

[0175] d) Cardiovascular index score: score is divided into sections according to aldosterone concentration interval;

[0176] (3) The four scores are weighted and summed according to the weight parameters to obtain the total score;

[0177] (4) Compare the total score with the preset threshold and output the risk level.

[0178] The present invention constructs a multi-level scoring system based on multiple HHV nucleic acid CT values, which has the following advantages over traditional detection methods:

[0179] 1. High-precision risk assessment: The system integrates the synergistic effects of multiple indicators to improve the accuracy of MCI risk prediction. Through modular scoring and weighted calculation, this system comprehensively reflects the impact of viral load, virus type, neuroimmune response, and cardiovascular status on MCI risk. After optimization, it can achieve a prediction accuracy and specificity of over 95%.

[0180] 2. Establish a dynamic scoring system and use multi-level scoring rules to fully reflect the clinical significance of each indicator. At the same time, dynamically adjust the weight of each indicator to meet the personalized needs of different patients and different testing platforms, making the system more flexible and adaptable under different clinical conditions.

[0181] 3. Real-time early warning and clinical decision support: The system can automatically collect data, calculate scores and give risk assessment results, providing real-time early warning information to clinicians, assisting in early intervention and individualized treatment decisions, improving the deficiencies of existing technologies in specificity and sensitivity, and providing more reliable technical support for the early diagnosis of MCI.

[0182] 4. Wide applicability and convenience: The system and method of the present invention are not only applicable to hospital clinical testing, community health monitoring and telemedicine systems, but can also be used as sensitive indicators in drug development and efficacy evaluation, and have good promotion prospects.

[0183] The present invention effectively solves the problems of single indicators, fixed thresholds and lack of multidimensional comprehensive evaluation in the existing technology. It constructs an MCI assessment system with a clear structure, clear steps, strict process conditions, scientific scoring rules and a dynamic adjustment mechanism. It greatly improves the prediction accuracy and specificity of MCI, can more accurately identify early MCI patients, and provides strong technical support for early intervention and precision medicine. BRIEF DESCRIPTION OF THE DRAWINGS

[0184] Figure 1 The accuracy results of the prediction model on the training set and test set during iterative training.

[0185] Figure 2 This is a graph of the specificity results during iterative training of the prediction model.

[0186] Figure 3 This is the ROC curve of the prediction model. DETAILED DESCRIPTION

[0187] The technical solution of the present invention is further described below with reference to specific embodiments, but the protection scope of the present invention is not limited thereto.

[0188] Example 1

[0189] A method for constructing an MCI risk assessment system based on a neuroimmune multi-level weighted scoring system includes the following steps:

[0190] S1. Collect sample data, including human herpesvirus nucleic acid CT values (HHV1-HHV8), virus species and number, neuroimmune cardiovascular indicators, and clinical MCI diagnosis data; a total of 500 sample data were collected and divided into training and test sets after data preprocessing;

[0191] Clinical MCI diagnostic data is diagnosed according to the existing technology for the seven clinical stages of AD combined with the blood beta-amyloid protein 42 / 40 ratio to determine whether it is a high risk or a low risk of MCI.

[0192] Data preprocessing includes checking data integrity and filling or removing missing values with the mean.

[0193] The dataset is divided into stratified random sampling groups: 70% is used as the training set and 30% is used as the test set. The ratio of positive and negative samples is kept consistent during the division (Stratified Split) to avoid bias caused by class imbalance and ensure classification balance.

[0194] S2. Construct a multi-level weighted scoring prediction model, dividing the sample data in the training set into four scoring modules, namely: human herpes virus nucleic acid CT value, number of virus species, virus combination and CT value, and aldosterone concentration. Scores are assigned according to preset intervals. Each scoring module is scored within a range of 0 to 100. The scoring principle is that the higher the score, the lower the risk.

[0195] The larger the viral nucleic acid CT value, the higher the score, but a CT value of 0 is a score of 100;

[0196] When the number of virus types is 0, the score is 100. The more types there are, the lower the score.

[0197] Virus combinations and CT values are scored according to different virus combination types and CT value intervals; the virus combination and CT value represent the level of human neuroimmune response. The lower the immune capacity represented by the virus combination and CT value, the lower the score.

[0198] Higher aldosterone concentrations are associated with higher scores, and lower aldosterone concentrations are associated with lower scores.

[0199] The scores of the four modules are weighted according to the preset weights to calculate the total score:

[0200] Total score = (score of the first module × R1) + (score of the second module × R2) + (score of the third module × R3) + (score of the fourth module × R4)

[0201] R1, R2, R3, and R4 are the preset weight parameters of the first module, the second module, the third module, and the fourth module respectively, and the sum of R1, R2, R3, and R4 is 1;

[0202] The weight parameter represents the relative importance of different modules to MCI risk. The higher the weight, the greater the impact on MCI risk.

[0203] Compare the total score with the preset threshold. If the total score is lower than the preset threshold, a high risk is output; if the total score is greater than or equal to the preset threshold, a low risk is output.

[0204] The output risk results were compared with actual clinical MCI diagnostic data, and the accuracy and specificity were calculated, with the lower value of the two being used as the evaluation standard. A machine learning algorithm was used to train the training set, and K-fold cross-validation was performed within the training set. The model parameters were optimized through multiple iterations to obtain the prediction model parameters of the scoring module with an accuracy and specificity of over 95%. The optimized scoring rules, segmentation thresholds, weight parameters, and preset thresholds were then obtained.

[0205] Accuracy is the ratio of the number of samples correctly predicted by the model to the total number of samples, and specificity refers to the ratio of the number of samples correctly predicted to have low MCI risk to the total number of samples with actual low MCI risk, which represents the ability of the model to correctly identify low-risk MC.

[0206] S3. Use the test set data to verify and evaluate the prediction model, further optimize the model, and obtain the optimal combination of the final optimized scoring rules, segmentation thresholds, weight parameters, and preset thresholds to construct an MCI risk assessment system for MCI risk assessment;

[0207] The model performance was evaluated using the data from the test set, and the accuracy, precision, and specificity were calculated. The ROC curve was drawn, and the AUC was calculated. The model was validated and evaluated, and the AUC, accuracy, and specificity of the model were required to meet the preset requirements. If the AUC, accuracy, and specificity of the model did not meet the preset requirements, the model was retrained until the accuracy and specificity of the prediction model met the preset requirements, and an MCI risk assessment system was constructed.

[0208] The preset requirements generally set the accuracy and specificity to be above 95%, and the AUC to be above 0.8.

[0209] The scoring rules, segmentation thresholds, weight parameters, and preset thresholds of the optimized model are as follows:

[0210] Module 1: CT value of human herpes virus nucleic acid

[0211] Any virus detected and CT value > 36 or CT value 0: score 100

[0212] CT value ≥31 and <36: score 80

[0213] CT value ≥ 25 and < 31: score 50

[0214] CT value <25: score 30

[0215] If multiple viruses are detected in a sample, the maximum CT value of the multiple viruses (or weighted average) can be used for scoring.

[0216] Module 2: Number of virus types:

[0217] Quantity is 0: score 100 points

[0218] Quantity is 1: score 80 points

[0219] The number is 2: score 60 points.

[0220] The number is 3: score 40 points.

[0221] Quantity > 3: score 20 points.

[0222] Module 3: Virus Combination and CT Value

[0223] When all virus detection indicators show a CT value of 0, the score is 100 points;

[0224] If one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and none of HHV4, HHV5, or HHV7 are detected, and the CT value of all detected viruses is greater than 36, the score is 90 points;

[0225] If one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and none of HHV4, HHV5, or HHV7 is detected, and the CT value of any of the detected viruses satisfies 0 < CT value < 36, the score is 80 points;

[0226] If one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and one of the following viruses is detected: HHV4, HHV5, HHV7, and the CT value of any of the detected viruses is between 0 < CT value < 36, the score is 70 points;

[0227] If HHV4 and HHV5, or HHV5 and HHV7, or HHV4, HHV5, and HHV7 are detected simultaneously, and the CT value of any of HHV4, HHV5, and HHV7 is between 0 and 36, the score is 70 points.

[0228] When HHV4 and HHV7 are detected at the same time, the CT values of HHV4 and HHV7 both meet 0<CT value<36, and the ratio of the CT value of HHV4 to the CT value of HHV7 is ≥1, the score is 60 points;

[0229] When the ratio of the CT value of HHV4 to the CT value of HHV7 is <1, the score is 40 points;

[0230] When HHV4 and HHV7 are detected at the same time, the CT values of HHV4 and HHV7 both meet 0<CT value<36, and the CT value of any one of the viruses is <25, and the ratio of HHV4 to HHV7 is <1, the score is 30 points.

[0231] In the fourth module, the scoring rules for aldosterone concentration intervals are as follows:

[0232] Aldosterone concentration <50 pg / mL, score 40

[0233] Aldosterone concentration ≥50 pg / mL and <100 pg / mL, score 50

[0234] Aldosterone concentration ≥100 pg / mL and <150 pg / mL, score 60

[0235] Aldosterone concentration ≥150 pg / mL and <200 pg / mL, score 80

[0236] Aldosterone concentration ≥200 pg / mL and <250 pg / mL, score 100

[0237] When the aldosterone concentration is >310 pg / ml, it can also be adjusted to 50 points based on statistical data (this range is set according to the clinical warning significance).

[0238] When the aldosterone concentration is >310pg / ml, it represents primary hypertension, which is abnormal aldosterone secretion, equivalent to excessive secretion, and is also an abnormal condition, generally assigned 50 points.

[0239] Total score = (score of the first module × 0.10) + (score of the second module × 0.10) + (score of the third module × 0.50) + (score of the fourth module × 0.30)

[0240] The preset threshold is 60, that is:

[0241] If the total score is less than 60, the output is high risk;

[0242] The total score is greater than or equal to 60, and the output is low risk.

[0243] The accuracy, specificity and ROC curves of the above model training are as follows: Figure 1 、 2 , as shown in 3.

[0244] Figure 1 It shows that after the 4th round, the accuracy steadily exceeded 95% and reached 97.5% in the 10th round, indicating that the model has converged and fits well.

[0245] Figure 2 The results show that the validation specificity exceeded 95% after the fifth round and eventually approached 98%, indicating that the model had an extremely low false positive rate when identifying low-risk (negative) samples.

[0246] Figure 3 The results show that AUC≈0.94, further proving that the model can maintain high accuracy and specificity at different thresholds.

[0247] According to the above method, an MCI risk assessment system based on neuroimmune multi-level weighted scoring was constructed, which includes:

[0248] (1) Biological detection module: including real-time fluorescence quantitative PCR instrument and immunoassay detection device;

[0249] The real-time fluorescence quantitative PCR instrument is used to detect the CT value of human herpes virus (HHV) nucleic acid in a blood or tissue sample to be tested. Human herpes viruses include HHV1, HHV2, HHV3, HHV6, HHV8, HHV4, HHV5, and HHV7.

[0250] The immunoassay detection device includes an enzyme-linked immunosorbent assay (ELISA) or a chemiluminescence immunoassay detector, which is used to detect the neuroimmune cardiovascular index in serum, namely the concentration of aldosterone.

[0251] In this embodiment, the concentration of aldosterone was detected by chemiluminescence immunoassay.

[0252] Real-time fluorescence quantitative PCR instruments are typically used in conjunction with virus detection kits. These kits contain specific primers, probes, and reaction buffers for human herpes viruses. These kits are used to prepare PCR reaction systems for detecting HHV1, HHV2, HHV3, HHV6, HHV8, HHV4, HHV5, and HHV7. Detection is performed on the real-time fluorescence quantitative PCR instrument. Accuracy and reproducibility of all test parameters must be ensured.

[0253] The biological detection module may further include a sample pretreatment system, which includes nucleic acid extraction reagents, a centrifuge, sample processing tubes, etc., for extracting nucleic acids from samples to ensure that the nucleic acid extraction has high efficiency and purity.

[0254] (2) a data input module for reading sample data, including human herpes virus nucleic acid CT value, virus species and number, and neuroimmune cardiovascular indicators, and performing preprocessing;

[0255] The preprocessing includes checking data integrity, and missing values can be filled or eliminated by the mean;

[0256] (3) Scoring module, including:

[0257] The first module: viral load scoring module, scoring according to the viral nucleic acid CT value;

[0258] The second module: virus type quantity scoring module, which scores according to the number of detected virus types;

[0259] The third module: neuroimmune response scoring module, which scores according to the virus combination and CT value range;

[0260] The fourth module: cardiovascular index scoring module, scoring according to the aldosterone concentration range;

[0261] The following is a detailed explanation of why the following four detection modules were selected from a neuroimmunological perspective, and their basic principles are explained:

[0262] 1. Viral load scoring module

[0263] The principle and neuroimmune mechanism are: viral reactivation and neuroinflammation:

[0264] Human herpes simplex virus, which remains latent in the body, can easily reactivate when immune regulation is imbalanced or the body is under stress. The viral load (such as the CT value obtained by PCR testing) can directly reflect the level of viral activity.

[0265] After viral reactivation, pathogen components (such as viral DNA or related antigens) can pass through the blood-brain barrier and enter the central nervous system, activating microglia and other immune cells, leading to an inflammatory response and the release of pro-inflammatory cytokines. This persistent neuroinflammation can damage nerve cells and thus affect cognitive function.

[0266] By detecting the viral load, the activity level of the virus in the body can be assessed, which in turn indirectly reflects the possibility of it causing neuroimmune abnormalities (such as excessive activation of microglia and inflammation levels), providing a biomarker basis for MCI risk.

[0267] 2. Number of virus types

[0268] The principle and neuroimmune mechanism lies in the cumulative effect of multiple viruses:

[0269] Not only a single virus, but also the coexistence and reactivation of other latent viruses may have synergistic negative effects on the nervous system. The simultaneous activation of different types of viruses may produce cumulative pro-inflammatory effects.

[0270] When multiple viruses are present in the body, the immune system may be in a state of chronic overactivation, increasing the immune burden and leading to chronic low-grade inflammation. The different antigenic stimulations carried by multiple viruses make it difficult for the immune regulatory mechanism to maintain a balanced state, thereby exacerbating neuroinflammation and cell damage.

[0271] Detecting the number of viral types helps reflect the "viral load" status of the immune system, indicating whether there are multiple pathogen interferences, thereby reflecting the overall burden of the neuroimmune system and inflammatory risk, which is particularly critical for assessing cognitive impairment.

[0272] 3. Neuroimmune response

[0273] The principle and neuroimmune mechanism lies in the immune regulation state:

[0274] The neuroimmune response is proposed for the first time in the present invention, and the human body's immune response ability to neuroviruses is evaluated through the neuroimmune response. Different human bodies have different immune response abilities to neuroviruses, which are affected by age, health status, genetics, immune memory, etc. Viruses can induce immune activation in the human body, and differences in immune response abilities can lead to different degrees of viral replication and spread in the body, resulting in differences in the types and loads of viruses in the body. This application not only evaluates the types and loads of viruses on the surface, but also further evaluates the combination of detected viruses as immune ability. Each virus has its own neural affinity and immune latency, and different virus combinations may reflect different response patterns of the human immune system. For the first time, the present invention also included the human body's immune regulation ability in the scope of investigation. A high immune response score represents a better human immune response, and a low score represents a poor immune self-regulation ability.

[0275] By detecting the combination of CT values and specific ratios of multiple viruses (such as the ratio of HHV-4 to HHV-7), an indicator reflecting the overall state of the neuroimmune response can be obtained.

[0276] Testing viral load and quantity alone is not sufficient to fully reflect the complex neuroimmune response, while comprehensive neuroimmune response indicators can more accurately evaluate the inflammatory state of the nervous system and provide direct information for predicting changes in cognitive function.

[0277] 4. Detection of brain microvascular circulation disorders (aldosterone)

[0278] The principle and neuroimmune mechanism lies in the role of blood-brain barrier and microcirculation:

[0279] The state of brain microvascular circulation has a significant impact on neuronal function. Microcirculatory disorders can cause ischemia, insufficient oxygen supply, and disruption of the blood-brain barrier.

[0280] Neuroinflammation can lead to increased permeability of the blood-brain barrier, making it easier for peripheral inflammatory factors and immune cells to enter the central nervous system, further triggering or aggravating neuroinflammation.

[0281] The RAAS system maintains microvascular function by regulating vasoconstriction and blood flow; changes in aldosterone levels not only reflect RAAS activity, but are also closely related to regulating the integrity of the blood-brain barrier and the state of cerebral microcirculation.

[0282] Aldosterone also affects cognitive function in the apical part of the frontal lobe through various mechanisms. Aldosterone may trigger or exacerbate neuroinflammation through the following mechanisms: excessive activation of microglia, increasing the release of proinflammatory cytokines; disruption of the blood-brain barrier, allowing peripheral inflammatory factors to easily enter the central nervous system and trigger a local immune response; interference with neurotransmitter regulation, affecting synaptic plasticity and disrupting normal information transmission in neural networks; and impacting endocrine-immune feedback, causing HPA axis dysregulation and further exacerbating the imbalance in the neuronal environment.

[0283] Including brain microvascular circulation (aldosterone concentration) in the test helps assess the spread of neuroinflammation and neurological dysfunction that may be caused by microcirculatory disorders and blood-brain barrier disruption. This module provides another important physiological information for assessing MCI risk, in addition to viral and immune responses.

[0284] This application is the first to incorporate aldosterone into an MCI risk assessment system, avoiding the one-sidedness and singularity of traditional methods that rely solely on viral indicators. Conventional techniques typically focus solely on viral-related indicators (such as viral CT values), while ignoring other important factors that may affect cognitive function. By incorporating aldosterone concentration as a physiological indicator, the present invention enables a more comprehensive assessment of MCI risk, comprehensively considering the combined effects of viral and physiological factors.

[0285] In summary, from the perspective of neuroimmunology, the four major modules selected in this invention reflect the following key information:

[0286] Viral load: represents the level of local inflammation triggered by viral reactivation;

[0287] Virus types: This suggests that multiple pathogens may jointly cause immune system burden and chronic inflammation risk;

[0288] Neuroimmune response: a comprehensive reflection of the activation of immune cells in the central nervous system;

[0289] Brain microvascular circulation: Assess blood-brain barrier integrity and microcirculatory function, and their regulatory effects on inflammatory signaling.

[0290] These four modules capture the three major links of the pathological process, namely viral infection, neuroimmune response and microvascular circulatory disorders, from different perspectives, providing a multi-dimensional biomarker basis for MCI risk assessment and helping to build a comprehensive evaluation system that reflects the patient's overall neuroimmune status, thereby more accurately predicting changes in cognitive function.

[0291] The evaluation system of the present invention can more accurately reflect the patient's overall health status and cognitive function risk, improve the predictive accuracy of the system model, and enhance its clinical applicability and interpretability.

[0292] (4) Weighted calculation module, which calculates the total score by weighted summation of the scores of each scoring module and the preset weights;

[0293] Total score = (score of the first module × 0.10) + (score of the second module × 0.10) + (score of the third module × 0.50) + (score of the fourth module × 0.30)

[0294] After model optimization, the third module has the highest weight coefficient, indicating that the human immune regulation ability has a greater impact on the risk of MCI.

[0295] (5) Risk assessment module: compares the total score with the preset threshold and outputs the MCI risk level;

[0296] If the total score is lower than the preset threshold, the risk is high;

[0297] If the total score is greater than or equal to the preset threshold, the risk is low.

[0298] In this embodiment, the preset threshold is set to 60.

[0299] The evaluation system also includes a result output module for displaying and saving the scores of each module, the total score and the risk evaluation results.

[0300] The evaluation system also includes a model optimization module, which includes a machine learning algorithm library and a clinical database, and is used to regularly update scoring rules, weight parameters and preset thresholds.

[0301] Furthermore, using the detection data of a large number of samples collected from the clinical database (viral CT values, number of virus types, immune response data, aldosterone concentration) and the final clinical MCI diagnosis data, the scoring rules, segmentation thresholds, and weight parameters of the weighted calculation module of the scoring module were subjected to regression analysis and cross-validation through a machine learning algorithm, and the parameters were dynamically adjusted and corrected to ensure that the final model prediction accuracy and specificity both reached more than 95%.

[0302] The model optimization module is equipped with a parameter update mechanism: a regular parameter correction and model iteration process is established to ensure that the system can be continuously optimized and improved based on clinical data during actual application.

[0303] Example 2

[0304] Using the MCI risk assessment system based on neuroimmune multi-level weighted scoring constructed in Example 1, a risk assessment of a clinical sample is performed, including the following steps:

[0305] (1) Sample collection: Collect venous blood or other appropriate samples from the subject.

[0306] Nucleic acid extraction: Use automated extraction equipment to extract nucleic acids from blood or tissues to ensure that the concentration and purity of the extract meet PCR testing requirements.

[0307] Serum separation: Centrifugation is used to separate serum for subsequent immunoassays of aldosterone and other indicators. Real-time fluorescence quantitative PCR is used to detect the CT value of human herpesvirus (HHV) nucleic acid in the sample to be tested. Human herpesviruses include HHV1, HHV2, HHV3, HHV6, HHV8, HHV4, HHV5, and HHV7.

[0308] The HHV test results of this sample are shown in the following table:

[0309]

[0310]

[0311] The patient's beta-amyloid test results are as follows: the beta-amyloid protein 42 / 40 ratio is 0.171, which is within the safe range. Generally speaking, a ratio of 0.15-0.169 indicates that the patient has a moderate risk of AD pathology.

[0312] Project Name Project Name result unit Reference interval Methodology β-amyloid 42 Aβ42 41.598 pg / mL IP-MS β-amyloid 40 Aβ40 243.387 pg / mL IP-MS β-amyloid 42 / 40 Aβ42 / 40 0.171 ≥0.17 IP-MS

[0313] According to the diagnostic criteria of existing expert consensus, this patient cannot be considered to be at risk of AD for the time being, and his or her MCI risk cannot be determined.

[0314] (2) The aldosterone concentration was measured using a chemiluminescence immunoassay; the patient's aldosterone concentration was 166.13 pg / ml.

[0315] (3) Execute four-dimensional scoring:

[0316] a) Module 1: Viral load scoring: Select the highest CT value and score according to the preset segmentation rules;

[0317] According to the scoring rules, HSV1, HHV4, and HHV7 were detected, and their maximum CT value of 33.02 was in the range of 31–36. According to the scoring rule of CT value ≥ 31 and < 36: score 80, the corresponding score was 80 points.

[0318] b) Module 2: Virus type scoring: Scoring is based on the number of detected virus types in descending order;

[0319] A total of three viruses were detected: HSV1, HHV-4, and HHV-7, with a corresponding score of 40 points.

[0320] c) Module 3: Neuroimmune response scoring: Scoring is based on virus combination and CT value range;

[0321] Group 1 (HSV1, HSV2, HSV3, HSV6, HSV8): The non-zero value is 33.02, all within (0, 36), and the basic score is 80 points.

[0322] Group 2 (HHV-4, HHV-5, HHV-7): There were non-zero values of 24.41 and 27.76, and all non-zero values were within (0, 36), with a preliminary adjusted score of 70 points.

[0323] Consider further the ratio between HHV-4 and HHV-7:

[0324] Ratio = 24.41 ÷ 27.76 ≈ 0.88 (less than 1),

[0325] At the same time, the HHV-4CT value was lower than 25 (24.41<25), so the final score was adjusted to 30 points.

[0326] d) Module 4: Cardiovascular index score: Score is divided into sections according to aldosterone concentration range;

[0327] The aldosterone concentration of 166.13 pg / ml falls within the range of 151 to 200 pg / ml, corresponding to a score of 80 points.

[0328] (4) Weight the four scores in order of 10%, 10%, 50%, and 30% to obtain the total score; Total score = 80 × 0.10 + 40 × 0.10 + 30 × 0.50 + 80 × 0.30 = 8 + 4 + 15 + 24 = 51 points

[0329] (5) Compare the total score with the preset threshold and output the risk level. The preset threshold is 60.

[0330] The total score of this sample is 51 points, which is less than 60. Therefore, the output risk level is high risk, and the possibility of MCI is high.

[0331] The MCI risk assessment system of the present invention comprehensively reflects the overall neuroimmune status of the patient, thereby more accurately predicting changes in cognitive function, which is conducive to auxiliary diagnosis and early intervention treatment.

Claims

1. A method for constructing an MCI risk assessment system based on neuroimmune multi-level weighted scoring, characterized by The method comprises the following steps: S1. Collect sample data, including human herpes virus nucleic acid CT value, virus species and number, neuroimmune cardiovascular indicators and clinical MCI diagnosis data, and divide the data into training set and test set after preprocessing; The human herpes virus nucleic acid CT value refers to the detected CT value of HHV1, HHV2, HHV3, HHV6, HHV8, HHV4, HHV5, and HHV7; The number of virus species refers to the types and numbers of human herpes viruses detected; The neuroimmune cardiovascular index was aldosterone concentration; S2. Construct a multi-level weighted scoring prediction model. Divide the sample data in the training set into four scoring modules, namely: human herpes virus nucleic acid CT value, number of virus species, virus combination and CT value, and aldosterone concentration. Each module is scored according to a preset interval. The scores of the four modules are weighted according to the preset weights to calculate the total score: Total score = (score of the first module × R1) + (score of the second module × R2) + (score of the third module × R3) + (score of the fourth module × R4) R1, R2, R3, and R4 are the preset weight parameters of the first module, the second module, the third module, and the fourth module respectively, and the sum of R1, R2, R3, and R4 is 1; Compare the total score with the preset threshold. If the total score is lower than the preset threshold, a high risk is output; if the total score is greater than or equal to the preset threshold, a low risk is output. The output risk results were compared with actual clinical MCI diagnostic data, and the accuracy and specificity were calculated, with the lower value of the two used as the evaluation standard. The training set was trained using a machine learning algorithm to obtain prediction model parameters with accuracy and specificity exceeding 95%, and the scoring rules, segmentation thresholds, weight parameters, and preset thresholds of the optimized scoring module were obtained. S3. Use the test set data to verify and evaluate the prediction model, further optimize the model, obtain the best combination of scoring rules, segmentation thresholds, weight parameters, and preset thresholds, and construct an MCI risk assessment system.

2. The method according to claim 1, wherein In step S3, the accuracy, precision, and specificity are calculated using the data from the test set, the ROC curve is drawn, the AUC is calculated, and the model is verified and evaluated, requiring that the accuracy and specificity of the prediction model meet the preset requirements; if the accuracy and specificity of the prediction model do not meet the preset requirements, the model is re-trained until the accuracy and specificity of the prediction model meet the preset requirements.

3. The method according to claim 2, wherein The preset requirements for the accuracy and specificity of the prediction model are: AUC ≥ 0.80, accuracy ≥ 0.95, and specificity ≥ 0.

95.

4. The method according to claim 1, wherein The method includes step S4: establishing a regular parameter calibration and model iteration process, updating the evaluation system based on newly input clinical sample data, and adjusting parameters in real time to ensure that the system can be continuously optimized and improved based on clinical data during actual application, and can achieve an accuracy and specificity of more than 95% in different populations and scenarios.

5. The method according to claim 1, wherein In step S2, assigning points according to a preset interval means assigning points within a range of 0 to 100 for each scoring module; The principle of assigning points is that the higher the score, the lower the risk; The higher the CT value of the viral nucleic acid, the higher the score, but when the CT value is 0, the score is 100; When the number of virus species is 0, the score is 100. The more the number of species, the lower the score; The virus combination and CT value are scored according to different virus combination types and CT value ranges; The higher the aldosterone concentration, the higher the score, and the lower the concentration, the lower the score.

6. A MCI risk assessment system based on neuroimmune multi-level weighted scoring, characterized by The system includes: (1) A data input module for reading sample data, where the sample data includes the CT value of human herpesvirus nucleic acid, the number of virus species, neuroimmune cardiovascular indicators, and clinical MCI diagnosis data; and performing preprocessing; The preprocessing includes checking data integrity, and for missing values, mean filling or deletion processing can be used; Human herpesviruses include HHV1, HHV2, HHV3, HHV6, HHV8, HHV4, HHV5, HHV7. The CT value of human herpesvirus nucleic acid refers to the CT values of detected HHV1, HHV2, HHV3, HHV6, HHV8, HHV4, HHV5, HHV7; The number of virus species refers to the types and numbers of detected human herpesviruses; The neuroimmune cardiovascular indicator is the aldosterone concentration; Clinical MCI diagnosis data refers to the result of whether the patient is diagnosed with MCI, including being diagnosed with MCI and not being diagnosed with MCI; (2) A scoring module, including: The first module: a viral load scoring module for scoring according to the CT value of the viral nucleic acid; The second module: a virus type number scoring module for scoring according to the detected number of virus species; The third module: a neuroimmune response scoring module for scoring according to the virus combination and CT value range; The fourth module: a cardiovascular indicator scoring module for scoring according to the aldosterone concentration range; (3) A weighted calculation module for performing weighted summation according to the scores of each scoring module and preset weights to calculate the total score; Total score = (score of the first module × R1) + (score of the second module × R2) + (score of the third module × R3) + (score of the fourth module × R4) R1, R2, R3, and R4 are the weight parameters of the first module, the second module, the third module, and the fourth module respectively, and the sum of R1, R2, R3, and R4 is 1; The weight parameters R1, R2, R3, and R4 are obtained by training the clinical sample database through a machine learning algorithm; (4) A risk assessment module: comparing the total score with a preset threshold and outputting the MCI risk level; If the total score is lower than the preset threshold, output high risk; If the total score is greater than or equal to the preset threshold, output low risk.

7. The MCI risk assessment system according to claim 6, wherein In the scoring module, in the first module, according to the preset CT value segmentation rule, the CT value of the viral nucleic acid is scored, and the preset CT value segmentation rule is set as: Any virus is detected and CT value > P1 or CT value is 0: score 100 CT value ≥ P2 and < P1: score 80 CT value ≥ P3 and < P2: score 50 CT value < P3: score 30 P1, P2, and P3 are the viral load scoring segmentation thresholds, and P1 > P2 > P3; the range is between 15 and 50; The segmentation thresholds P1, P2, and P3 are obtained by training the clinical sample database through a machine learning algorithm; In the second module, a decreasing score is given according to the number of detected virus species, and the scoring rules are set as follows: When the number is 0: the score is 100 points When the number is N1: the score is 80 points When the number is N2: the score is 60 points When the number is N3: the score is 40 points When the number > N3: the score is 20 points N1, N2, and N3 are the threshold values for dividing the number of virus species for scoring, and N1 < N2 < N3; the range is between 1 and 10; The threshold values N1, N2, and N3 for segmentation are obtained by training a clinical sample database through a machine learning algorithm; In the third module, the scoring rules for the neuroimmune response are as follows: When all detected virus indicators show a CT value of 0, the score is 100 points; When one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and none of HHV4, HHV5, HHV7 are detected, and the CT value of all detected viruses > 36, the score is 90 points; When one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and none of HHV4, HHV5, HHV7 are detected, and the CT value of any one of the detected viruses satisfies 0 < CT value < 36, the score is 80 points; When one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and one of the following viruses is detected: HHV4, HHV5, HHV7, and the CT value of any one of the detected viruses is between 0 < CT value < 36, the score is 70 points; When HHV4 and HHV5, or HHV5 and HHV7, or HHV4, HHV5, and HHV7 are detected simultaneously, and the CT value of any one of HHV4, HHV5, HHV7 is between 0 < CT value < 36, the score is 70 points; When HHV4 and HHV7 are detected simultaneously, the CT values of HHV4 and HHV7 both satisfy 0 < CT value < 36, and the ratio of the CT value of HHV4 to the CT value of HHV7 ≥ 1, the score is 60 points; When the ratio of the CT value of HHV4 to the CT value of HHV7 < 1, the score is 40 points; When HHV4 and HHV7 are detected simultaneously, and the CT value of any one virus < 25, and the ratio of HHV4 to HHV7 < 1, the score is 30 points; In the fourth module, the scoring rules for the aldosterone concentration range are as follows: When the aldosterone concentration < Q1 pg / mL, the score is 40 points When the aldosterone concentration ≥ Q1 pg / mL and < Q2 pg / mL, the score is 50 points When the aldosterone concentration ≥ Q2 pg / mL and < Q3 pg / mL, the score is 60 points When the aldosterone concentration ≥ Q3 pg / mL and < Q4 pg / mL, the score is 80 points When the aldosterone concentration ≥ Q4 pg / mL and < Q5 pg / mL, the score is 100 points When the aldosterone concentration > Q6 pg / ml, the score is 50 points; Q1, Q2, Q3, Q4, Q5, and Q6 are the threshold values for dividing the aldosterone concentration for scoring, and Q1 < Q2 < Q3 < Q4 < Q5 < Q6; the range is between 30 and 310; The segmentation thresholds Q1, Q2, Q3, Q4, Q5, and Q6 were obtained by training the clinical sample database using a machine learning algorithm; The preset threshold is between 40 and 80.

8. The MCI risk assessment system according to claim 7, wherein The scoring rules, segmentation thresholds, weight parameters, and preset thresholds of the model trained by the machine learning algorithm are as follows: Module 1: CT value of human herpes virus nucleic acid Any virus detected and CT value > 36 or CT value 0: score 100 CT value ≥31 and <36: score 80 CT value ≥ 25 and < 31: score 50 CT value <25: score 30 If multiple viruses are detected in a sample, the maximum CT value of the multiple viruses is used for scoring; Module 2: Number of virus types: Quantity is 0: score 100 points Quantity is 1: score 80 points Quantity is 2: score 60 points Quantity is 3: score 40 points Number > 3: score 20 points Module 3: Neuroimmunological response assessment, scoring based on virus combination and CT value When all virus detection indicators show a CT value of 0, the score is 100 points; If one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and none of HHV4, HHV5, or HHV7 are detected, and the CT value of all detected viruses is greater than 36, the score is 90 points; If one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and none of HHV4, HHV5, or HHV7 is detected, and the CT value of any of the detected viruses satisfies 0 < CT value < 36, the score is 80 points; If one or more of the following viruses are detected: HHV1, HHV2, HHV3, HHV6, HHV8, and one of the following viruses is detected: HHV4, HHV5, HHV7, and the CT value of any of the detected viruses is between 0 < CT value < 36, the score is 70 points; If HHV4 and HHV5, or HHV5 and HHV7, or HHV4, HHV5, and HHV7 are detected simultaneously, and the CT value of any of HHV4, HHV5, and HHV7 is between 0 and 36, the score is 70 points. When HHV4 and HHV7 are detected at the same time, the CT values of HHV4 and HHV7 both meet 0<CT value<36, and the ratio of the CT value of HHV4 to the CT value of HHV7 is ≥1, the score is 60 points; When the ratio of the CT value of HHV4 to the CT value of HHV7 is <1, the score is 40 points; If both HHV4 and HHV7 are detected, and the CT value of either virus is less than 25, and the ratio of HHV4 to HHV7 is less than 1, the score is 30 points; In the fourth module, the scoring rules for aldosterone concentration intervals are as follows: Aldosterone concentration <50 pg / mL, score 40 Aldosterone concentration ≥50 pg / mL and <100 pg / mL, score 50 Aldosterone concentration ≥100 pg / mL and <150 pg / mL, score 60 Aldosterone concentration ≥150 pg / mL and <200 pg / mL, score 80 Aldosterone concentration ≥200 pg / mL and <250 pg / mL, score 100 When the aldosterone concentration was >310 pg / ml, the score was 50 points; Total score = (score of the first module × 0.10) + (score of the second module × 0.10) + (score of the third module × 0.50) + (score of the fourth module × 0.30) The preset threshold is 60, that is: If the total score is less than 60, the output is high risk; The total score is greater than or equal to 60, and the output is low risk.

9. The MCI risk assessment system according to claim 6, wherein The system includes a model training and optimization module, which contains a machine learning algorithm library and a clinical database. The clinical database is trained through a machine learning algorithm to obtain model parameters with an accuracy and specificity of more than 95%, and the scoring rules, segmentation thresholds, and weight parameters of the weighted calculation module are optimized. The system is optimized according to the updated clinical database, and the scoring rules, weight parameters and preset thresholds are regularly updated.

10. A method for MCI risk assessment based on multi-level weighted neuroimmune assessment, wherein the method uses the MCI risk assessment system based on multi-level weighted neuroimmune assessment as described in any one of claims 6 to 9 to assess MCI risk, characterized in that The method comprises the following steps: (1) Input the CT value and aldosterone concentration value of human herpesvirus nucleic acid in the sample to be tested. Human herpesviruses include HHV1, HHV2, HHV3, HHV6, HHV8, HHV4, HHV5, and HHV7; (2) Execute four-dimensional scoring: a) Viral load scoring: Select the highest CT value and score according to the preset segmentation rules; b) Virus type score: score is given in descending order of the number of virus types detected; c) Neuroimmune response score: score is calculated based on the virus combination and CT value range; d) Cardiovascular index score: score is divided into sections according to aldosterone concentration interval; (3) The four scores are weighted and summed according to the weight parameters to obtain the total score; (4) Compare the total score with the preset threshold and output the risk level.

Images