Down's syndrome screening system based on multiple key indexes

Abstract

The invention relates to the technical field of data analysis, and particularly discloses a Down's syndrome screening system based on multiple key indicators, which comprises the following steps: analyzing historical sample data including pregnant woman age, body weight, race background, pregnancy history record, serum PAPP-A concentration, free beta-hCG concentration, neck transparent layer thickness and the like; mining a chromosome abnormality risk distribution rule; standardizing detection data by adopting a principal component analysis method, grouping pregnant woman features by utilizing a clustering algorithm, and determining a feature offset coefficient; establishing a personalized index reference interval by combining factors such as fetal structure abnormality, chromosome abnormality and family genetic disease history; generating personalized risk probability distribution by fusing individual features and ultrasonic measurement data; according to the method, the risk value is calculated and the dynamically adjusted risk prediction curve is generated according to the real-time detection data and the personalized reference interval, so that an accurate Down's syndrome risk assessment basis is provided for clinical decisions.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular, to a Down syndrome screening system based on multiple key indicators. Background Art

[0002] During the process of Down syndrome screening, a series of interrelated technical challenges are currently faced. First, the individual differences among different pregnant women lead to significant variations in screening indicators, and how to establish personalized reference intervals is a key issue. Second, the serum test data and ultrasonic measurement data come from different devices, resulting in data non-uniformity problems, which need to be standardized. Moreover, factors such as maternal age, weight, and pregnancy history have complex effects on screening indicators, and how to accurately quantify these effects and make corrections is a major difficulty. In addition, as the gestational age changes, the screening indicators also change dynamically, and how to establish a dynamic risk assessment model becomes another challenge. Finally, in practical applications, how to convert the results of these complex data analyses into concise and easy-to-understand risk grades and give high-risk warnings in a timely manner is the core problem of the entire screening system. These problems are closely linked and jointly constitute a technical problem chain in Down syndrome screening. Summary of the Invention

[0003] In order to solve the above technical problems, the present invention provides a Down syndrome screening system based on multiple key indicators, and the following technical solutions: A Down syndrome screening system based on multiple key indicators, A first preprocessing module, configured to obtain data on pregnant women's age, weight difference, ethnic background, pregnancy history record, serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness from a historical sample database, and through statistical analysis, mine the distribution law of chromosomal abnormality risks under the combination of maternal age influence and ethnic differences to obtain a preliminary risk distribution feature set; A second preprocessing module, configured to map the serum test data and ultrasonic measurement data to a unified feature space to obtain a standardized index vector set, and perform offset correction processing on the standardized index vector set, and combine fetal structural abnormalities, chromosomal abnormalities, family genetic history, and gene mutations to determine a corrected personalized index reference interval set; A third preprocessing module, configured to extract the distribution characteristics of serum PAPP-A concentration and free β-hCG concentration corresponding to maternal age influence, weight difference, and gestational age dependence from the corrected personalized index reference interval set, and fuse individual characteristics and ultrasonic measurement data through a model fusion method based on conditional probability to obtain a personalized risk probability distribution for chromosomal abnormalities; The first screening module is used to obtain the serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness of a pregnant woman from a real-time detection device according to the personalized risk probability distribution. If the detected value exceeds the personalized reference interval, the risk calculation value is calculated by combining the maternal age influence and family genetic records, and a Down syndrome risk assessment curve is obtained. The second screening module is used to fit the dynamic monitoring data within consecutive gestational weeks by using the linear interpolation method through the change trend of the risk calculation value in the Down syndrome risk assessment curve, determine the dynamic distribution law of the risk probability under gestational week dependence, and obtain a dynamically adjusted risk prediction curve. The third screening module is used to classify the risk probability in the dynamically adjusted risk prediction curve by using the grading threshold determined based on clinical research data. If the risk probability is higher than the preset threshold, a high-risk prompt signal is generated, and the output classification of the screening result is determined.

[0004] The first preprocessing module is specifically used for: Obtain historical sample data including the age, weight difference, ethnic background, pregnancy history record, serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness data of the pregnant woman, and construct an original data set. Use statistical analysis methods to calculate the association strength between the age of the pregnant woman and chromosomal abnormalities in the original data set, and obtain the distribution characteristics of the maternal age influence. Perform cluster analysis on the ethnic background and chromosomal abnormalities in the original data set to determine the distribution pattern of ethnic differences. If the distribution characteristics of the maternal age influence exceed the preset threshold, a risk adjustment factor is calculated by combining the serum PAPP-A concentration and free β-hCG concentration to obtain an adjusted maternal age risk distribution. Determine the comprehensive risk distribution characteristics by fusing the nuchal translucency thickness and the adjusted maternal age risk distribution through regression analysis. Obtain the cross-statistical result of the comprehensive risk distribution characteristics and the pregnancy history record, and judge the multi-dimensional distribution law of chromosomal abnormalities. Generate feature set data according to the multi-dimensional distribution law, and output a preliminary risk distribution feature set.

[0005] The second preprocessing module is specifically used for: Obtain serum test data and ultrasonic measurement data, and use the principal component analysis method to process the serum test data and ultrasonic measurement data, map the serum test data and ultrasonic measurement data to a unified feature space, and obtain a standardized index vector set. For the serum test data, calculate the variances of the PAPP-A concentration and free β-hCG concentration to determine the measurement deviation range caused by device differences. If the variance of the PAPP-A concentration or the free β-hCG concentration exceeds a preset threshold, normalize the serum test data to obtain an adjusted concentration dataset; According to the adjusted concentration dataset and the ultrasonic measurement data, use the principal component analysis method to extract the main feature components to obtain a dimensionality-reduced feature matrix; Calculate the distribution distance of each sample in the feature space through the feature matrix, judge the measurement consistency between samples, and obtain a consistency judgment result; According to the consistency judgment result, use the K-means clustering algorithm to group the samples to obtain a classified vector set; For the classified vector set, calculate the mean value of the standardized indicators for each group to obtain a standardized indicator vector set.

[0006] The second preprocessing module is specifically used for: Obtain the standardized indicator vector set, and perform offset correction processing on the standardized indicator vector set by using the least squares method to obtain a preliminary indicator set after correction; Obtain fetal structural abnormality data, extract abnormal characteristic values from the fetal structural abnormality data, and perform data fusion on the abnormal characteristic values and the preliminary indicator set to obtain a first fusion indicator set; Obtain chromosomal abnormality data, extract variant characteristic values from the chromosomal abnormality data, and perform weighted processing on the variant characteristic values and the first fusion indicator set to obtain a second fusion indicator set; Obtain family genetic history data, extract risk factors from the family genetic history data, and use the linear regression method to adjust the second fusion indicator set to obtain a third fusion indicator set; Obtain gene mutation data, extract mutation site information from the gene mutation data, judge whether the mutation site exceeds a preset threshold, and if so, perform weighted calibration on the third fusion indicator set to obtain a fourth fusion indicator set; Determine a personalized reference interval set according to the fourth fusion indicator set, and judge the deviation degree of abnormal indicators by comparing the fourth fusion indicator set with the personalized reference interval set to obtain a finally corrected personalized indicator set.

[0007] The third preprocessing module is specifically used for: Obtain a corrected reference interval set, and the reference interval set includes serum PAPP-A concentration and free β-hCG concentration data corresponding to maternal age influence, weight difference, and gestational week dependence; According to the reference interval set, extract the distribution characteristics of serum PAPP-A and free β-hCG through statistical analysis, and determine the correlation parameters with maternal age influence, weight difference, and gestational week dependence; Obtain individual characteristics and ultrasonic measurement data, integrate the distribution characteristics of the serum PAPP-A and free β-hCG through preprocessing, and obtain a standardized feature set; Use a conditional probability model to fuse the standardized feature set, and calculate the joint probability distribution of individual characteristics, ultrasonic measurement and chromosomal abnormalities; Update the conditional probability in the joint probability distribution through the Bayesian method to obtain a preliminary risk probability distribution for chromosomal abnormalities; Judge whether the preliminary risk probability exceeds a preset threshold. If it exceeds, adjust the fusion parameters of the conditional probability model in combination with the distribution characteristics to obtain a final personalized risk probability distribution; According to the final personalized risk probability distribution, obtain the personalized risk probability distribution characteristics of chromosomal abnormalities.

[0008] The first screening module is specifically used for: Obtain the serum PAPP-A concentration, free β-hCG concentration and nuchal translucency thickness data to obtain the original test values; If the original test values exceed a preset reference interval, obtain abnormal marker data; Obtain the maternal age influencing factor, and use a linear regression algorithm to calculate a preliminary risk adjustment value according to the abnormal marker data and the maternal age influencing factor; Obtain the family genetic record, and perform secondary correction on the preliminary risk adjustment value according to the family genetic record to obtain a comprehensive risk value; Use a Gaussian distribution model to process the comprehensive risk value to generate personalized risk probability distribution data; Draw a curve according to the personalized risk probability distribution data to obtain a risk assessment curve for Down syndrome; If the comprehensive risk value exceeds a preset threshold, trace the source of the abnormality of the original test value according to the abnormal marker data to determine the key influencing factor.

[0009] The second screening module is specifically used for: Obtain the initial calculated value of risk assessment within consecutive gestational weeks; Use linear interpolation to fit the data of the initial calculated value to obtain a smoothed change trend; According to the change trend, determine the distribution law characteristics of the risk probability dependent on gestational weeks; If the distribution law characteristics deviate from a preset threshold, dynamically adjust the risk probability to obtain an adjusted risk probability value; Draw a prediction curve according to the adjusted risk probability value; Obtain the time series data of the prediction curve and determine the change trend dependent on gestational weeks; Generate an updated strategy for dynamic monitoring in response to the described change trend to obtain an optimized risk assessment result.

[0010] The third screening module is specifically used for: Obtain clinical data, use statistical analysis methods to determine the initial parameters of the prediction model, and obtain the basic curve for risk prediction; For the basic curve, use time series analysis methods for dynamic adjustment to obtain an adjusted prediction curve and determine the distribution characteristics of the risk probability; Extract the risk probability from the adjusted prediction curve. If the risk probability is higher than the preset threshold, generate a high-risk identifier and judge the potential abnormal state; According to the high-risk identifier, obtain the preliminary classification of the screening result through signal mapping technology; After obtaining the preliminary classification, use the decision tree algorithm to optimize the screening result to obtain the final output classification; Compare the output classification with the clinical data to obtain the deviation value of the classification result; According to the deviation value, use the iterative update method to optimize the grading threshold, and obtain an optimized risk prediction curve through loop processing.

[0011] It also includes: a psychological assessment module, which is used to, when the third screening module generates a high-risk prompt signal, real-time obtain the patient's physiological signal data and psychological assessment data, generate a multi-dimensional psychological state feature vector in combination with the risk level, and judge whether to trigger a psychological intervention signal; The psychological assessment module is specifically used for: Collect the patient's real-time physiological signals through wearable devices and extract the characteristic parameters related to the autonomic nervous system; Synchronously obtain the psychological state feedback information submitted by the patient and convert it into standardized psychological assessment parameters; Comprehensively analyze the physiological characteristic parameters, psychological assessment parameters and the current risk level to generate a psychological stress index; When the psychological stress index exceeds the preset range, trigger a psychological intervention signal and generate a state vector containing risk-related characteristics.

[0012] It also includes: an intervention decision module, which is used to match the corresponding intervention plan from a multi-level intervention strategy library according to the state vector; The intervention decision module is specifically used for: Establish a multi-level intervention strategy including instant consultation, self-training, and professional referral; Generate a matching degree parameter according to the multi-dimensional feature of the state vector and dynamically select the type of intervention strategy; When a strong correlation is detected between the psychological stress index and the high-risk level, the real-time interactive intervention of professionals is preferentially initiated; If the psychological stress index is within the adjustable range, customized training content and community support resources are pushed to the patient terminal; When it is monitored that the intervention response fails to meet the expectation, the family member collaborative intervention mechanism is automatically triggered.

[0013] It also includes: an effect feedback module, which is used to update the psychological state assessment model through dynamic monitoring of physiological signals and interactive feedback data; The effect feedback module is specifically used for: Continuously collect data on the changing trend of physiological signals during the intervention implementation period; Obtain the subjective experience feedback of the patient through the interactive interface and extract the emotional state characteristics; Conduct a correlation trend analysis on physiological and psychological data. When it is detected that the improvement rate is lower than expected, the intervention intensity level is automatically increased; Input the updated feature data into the adaptive model to optimize the psychological stress assessment parameters in the next stage.

[0014] The effect feedback module also includes a genetic counseling connection unit. When the high-risk prompt signal and the psychological intervention effect data meet the preset correlation conditions, a visual report integrating the screening results and psychological assessment is automatically generated, and the genetic counseling service channel is activated; The genetic counseling connection unit is specifically used for: Conduct a spatio-temporal correlation analysis on the dynamic risk prediction curve and the changing trend of psychological stress; When it is detected that there is a continuous positive correlation between the risk level and psychological stress, the priority consultation channel is triggered; Transmit the complete assessment record to the designated genetic counseling terminal through a secure encryption protocol.

[0015] On the other hand, a Down syndrome screening method based on multiple key indicators is provided, with the following technical solutions: Obtain data on the age of pregnant women, weight difference, ethnic background, pregnancy history record, serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness from the historical sample database. Through statistical analysis, mine the distribution law of chromosomal abnormality risks under the combination of maternal age influence and ethnic differences to obtain a preliminary risk distribution feature set; Map the serum test data and ultrasonic measurement data to a unified feature space to obtain a standardized index vector set, and perform an offset correction process on the standardized index vector set. Combine fetal structural abnormalities, chromosomal abnormalities, family genetic history, and gene mutations to determine a corrected personalized index reference interval set; The distribution characteristics of serum PAPP-A concentration and free β-hCG concentration corresponding to maternal age, weight difference and gestational age dependence were extracted from the corrected personalized index reference interval set. The individual characteristics and ultrasound measurement data were fused through a model fusion method based on conditional probability to obtain a personalized risk probability distribution for chromosomal abnormalities. According to the personalized risk probability distribution, the serum PAPP-A concentration, free β-hCG concentration and nuchal translucency thickness of pregnant women are obtained from the real-time detection equipment. If the test value exceeds the personalized reference interval, the risk calculation value is calculated in combination with the influence of maternal age and family genetic records to obtain the Down syndrome risk assessment curve; Through the changing trend of the risk calculation value in the Down syndrome risk assessment curve, the linear interpolation method is used to fit the dynamic monitoring data in consecutive gestational weeks, to determine the dynamic distribution law of risk probability under gestational age dependence, and to obtain the dynamically adjusted risk prediction curve; The risk probability in the dynamically adjusted risk prediction curve is graded using a grading threshold determined based on clinical research data. If the risk probability is higher than the preset threshold, a high-risk warning signal is generated to determine the output classification of the screening result.

[0016] Compared with the prior art, the Down syndrome screening system (method) based on multiple key indicators provided by the present invention has the following technical effects: The present invention mines the distribution pattern of chromosomal abnormality risk by analyzing historical sample data, including maternal age, weight, ethnic background, pregnancy history, serum PAPP-A concentration, free β-hCG concentration and nuchal translucency thickness.

[0017] The principal component analysis method was used to standardize the test data, and the clustering algorithm was used to group the characteristics of pregnant women and determine the characteristic deviation coefficient. Combined with factors such as fetal structural abnormalities, chromosomal abnormalities, and family genetic history, personalized index reference intervals were established.

[0018] By integrating individual characteristics with ultrasound measurement data, a personalized risk probability distribution is generated. Finally, the present invention calculates the risk value and generates a dynamically adjusted risk prediction curve based on real-time detection data and personalized reference intervals, providing an accurate basis for Down syndrome risk assessment for clinical decision-making.

[0019] Through the fusion of physiological and psychological bimodal data, psychological status assessment is initiated simultaneously when high-risk screening results appear, so as to avoid patients' anxiety affecting subsequent diagnosis and treatment decisions and improve the clinical compliance of the screening system. A closed-loop association between intervention effects and physiological indicators is established to achieve real-time optimization of personalized psychological support plans and ensure dynamic matching of intervention measures with the actual status of patients.

[0020] Organically combine automated intervention resources with professional services, while ensuring the response speed, retain the flexibility of manual intervention, and form a multi-dimensional health management link of screening - psychology - genetics; through the spatio-temporal correlation analysis of the psychological stress index and the screening risk level, identify special cases that need key attention in advance, and provide decision-making support for subsequent genetic counseling. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art.

[0022] Figure 1 It is a schematic structural diagram of a Down syndrome screening system based on multiple key indicators. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] The following will clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments.

[0024] Embodiment 1 Figure 1 It shows a schematic structural diagram of an embodiment of a Down syndrome screening system 100 provided by the present invention. As Figure 1 shown, the system 100 includes: a first preprocessing module 110, a second preprocessing module 120, a third preprocessing module 130, a first screening module 140, a second screening module 150, and a third screening module 160; The first preprocessing module 110 is used to obtain data on pregnant women's age, weight difference, ethnic background, pregnancy history record, serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness from the historical sample database, and through statistical analysis, mine the distribution law of chromosomal abnormality risks under the combination of maternal age influence and ethnic differences, so as to obtain a preliminary risk distribution feature set; The second preprocessing module 120 is used to map the serum test data and ultrasonic measurement data to a unified feature space to obtain a set of standardized index vectors, and perform offset correction processing on the set of standardized index vectors. Combining fetal structural abnormalities, chromosomal abnormalities, family genetic history, and gene mutations, determine a set of corrected personalized index reference intervals; The third preprocessing module 130 is used to extract the distribution characteristics of serum PAPP-A concentration and free β-hCG concentration corresponding to maternal age influence, weight difference, and gestational age dependence from the set of corrected personalized index reference intervals, and fuse individual characteristics and ultrasonic measurement data through a model fusion method based on conditional probability to obtain a personalized risk probability distribution for chromosomal abnormalities; The first screening module 140 is configured to obtain the serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness of a pregnant woman from a real-time detection device according to the personalized risk probability distribution. If the detected value exceeds the personalized reference interval, a risk calculation value is calculated by combining the maternal age influence and family genetic records to obtain a Down syndrome risk assessment curve; The second screening module 150 is configured to fit the dynamic monitoring data within consecutive gestational weeks by using linear interpolation according to the change trend of the risk calculation value in the Down syndrome risk assessment curve, and determine the dynamic distribution law of the risk probability under gestational week dependence to obtain a dynamically adjusted risk prediction curve; The third screening module 160 is configured to classify the risk probability in the dynamically adjusted risk prediction curve by using a grading threshold determined based on clinical research data. If the risk probability is higher than a preset threshold, a high-risk prompt signal is generated to determine the output classification of the screening result.

[0025] In an optional manner, the first preprocessing module 110 is specifically configured to: Obtain historical sample data including the age, weight difference, ethnic background, pregnancy history record, serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness data of a pregnant woman, and construct an original data set; calculate the association strength between the age of the pregnant woman and chromosomal abnormalities in the original data set by using a statistical analysis method to obtain the distribution characteristics of the maternal age influence; perform cluster analysis on the ethnic background and chromosomal abnormalities in the original data set to determine the distribution pattern of ethnic differences; if the distribution characteristics of the maternal age influence exceed a preset threshold, calculate a risk adjustment factor by combining the serum PAPP-A concentration and free β-hCG concentration to obtain an adjusted maternal age risk distribution; determine the comprehensive risk distribution characteristics by regression analysis integrating the nuchal translucency thickness and the adjusted maternal age risk distribution; obtain the cross-statistical result of the comprehensive risk distribution characteristics and the pregnancy history record, and judge the multi-dimensional distribution law of chromosomal abnormalities; generate feature set data according to the multi-dimensional distribution law and output a preliminary risk distribution feature set.

[0026] Specifically, data such as pregnant women's age, weight difference, racial background, pregnancy history records, serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness are extracted from the historical sample database. First, the data is cleaned to remove missing values and outliers. For example: ① In the pregnant women's age data, there are records where the age is less than 15 years old or greater than 50 years old, and these records are regarded as outliers and excluded. Then, the data is processed using a standardization method to convert the serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness into Z-scores to ensure the comparability of data with different dimensions. ② The average value of the serum PAPP-A concentration is 8 MoM, and the standard deviation is 2. After Z-score conversion, the values are distributed between -3 and 3. Subsequently, a multiple regression analysis model is used, with the pregnant women's age and racial background as independent variables and the chromosomal abnormality risk as the dependent variable for statistical analysis. ③ The model results show that the chromosomal abnormality risk of pregnant women over 35 years old increases significantly, and the risk ratio (OR value) is 5, while the risk ratio (OR value) of Asian pregnant women is 8. Further, through the clustering analysis method, the pregnant women are grouped according to age and racial background to obtain the risk distribution characteristics under different combinations. ④ Asian pregnant women over 35 years old have the highest chromosomal abnormality risk, with a risk value of 2, while African pregnant women under 30 years old have the lowest risk value, which is 7. Finally, the decision tree algorithm is used to further mine the risk distribution characteristic set to identify high-risk groups. ⑤ The decision tree model shows that pregnant women with a serum PAPP-A concentration lower than 5 MoM and a nuchal translucency thickness greater than 5 mm have a significantly increased chromosomal abnormality risk, with a risk value of 5. Through the above statistical analysis, the distribution law of chromosomal abnormality risk under the combination of maternal age influence and racial differences is initially obtained, providing data support for subsequent clinical decisions.

[0027] In an alternative manner, the second preprocessing module 120 is specifically configured to: Obtain serum test data and ultrasonic measurement data, and process the serum test data and ultrasonic measurement data using the principal component analysis method, map the serum test data and ultrasonic measurement data to a unified feature space, and obtain a standardized index vector set; for the serum test data, calculate the variances of the PAPP-A concentration and the free β-hCG concentration, and determine the measurement deviation range caused by equipment differences; if the variances of the PAPP-A concentration or the free β-hCG concentration exceed a preset threshold, perform normalization processing on the serum test data to obtain an adjusted concentration data set; according to the adjusted concentration data set and the ultrasonic measurement data, use the principal component analysis method to extract the main feature components to obtain a dimensionality-reduced feature matrix; calculate the distribution distances of each sample in the feature space through the feature matrix, judge the measurement consistency between samples, and obtain a consistency judgment result; according to the consistency judgment result, use the K-means clustering algorithm to group the samples to obtain a classified vector set; for the classified vector set, calculate the standardized index mean of each group to obtain a standardized index vector set; obtain the standardized index vector set, and perform offset correction processing on the standardized index vector set using the least squares method to obtain a preliminarily corrected index set; obtain fetal structural anomaly data, extract abnormal feature values from the fetal structural anomaly data, and perform data fusion on the abnormal feature values and the preliminarily corrected index set to obtain a first fusion index set; obtain chromosomal anomaly data, extract variant feature values from the chromosomal anomaly data, and perform weighted processing on the variant feature values and the first fusion index set to obtain a second fusion index set; obtain family genetic history data, extract risk factors from the family genetic history data, and use the linear regression method to adjust the second fusion index set to obtain a third fusion index set; obtain gene mutation data, extract mutation site information from the gene mutation data, judge whether the mutation sites exceed a preset threshold, and if so, perform weighted calibration on the third fusion index set to obtain a fourth fusion index set; determine a personalized reference interval set according to the fourth fusion index set, and judge the deviation degree of abnormal indicators by comparing the fourth fusion index set with the personalized reference interval set to obtain a finally corrected personalized index set.

[0028] Specifically, during the measurement of serum PAPP-A concentration and free β-hCG concentration, data inconsistency may occur due to differences in detection methods or equipment. First, the dimensionality reduction process is performed on the serum test data and ultrasonic measurement data through the principal component analysis (PCA). Suppose we have a set of serum test data containing 100 samples, each sample includes two features: PAPP-A concentration and free β-hCG concentration, and 50 ultrasonic measurement data, each sample includes two features: fetal head circumference and abdominal circumference. First, these data are combined into a 150×4 matrix, and then standardized so that the mean of each feature is 0 and the standard deviation is 1. Then, the covariance matrix is calculated and its eigenvalue decomposition is performed to obtain eigenvalues and eigenvectors. Suppose the cumulative contribution rate of the first two principal components reaches 90%, and these two principal components are selected as the new feature space. The original data is projected onto these two principal components to obtain a set of 150×2 standardized index vectors.

[0029] For example, the projection value of the first sample on the first principal component is 85, and the projection value on the second principal component is -45. In this way, the serum test data and ultrasonic measurement data are mapped to a unified feature space, solving the problem of measurement inconsistency and providing a unified standardized index for subsequent risk assessment.

[0030] Specifically, first, the offset correction process is performed on the set of standardized index vectors, and the least squares method is used to fit the data. Suppose there is a set of fetal structural anomaly index vectors, including head circumference, abdominal circumference, and femur length, which are standardized to [8, 2, 9] respectively. Through the least squares fitting, the correction coefficients are obtained as [95, 05, 98], and the corrected index vectors are [76, 26, 88]. Then, combined with chromosomal anomaly data, such as the chromosomal anomaly probability of trisomy 21 syndrome is 01, the corrected probability is calculated as 015 through Bayes' theorem. Further, considering the family genetic history, such as a family history of congenital heart disease with a genetic risk coefficient of 3, through the weighted average method, the corrected index vectors are adjusted to [76*3, 26*3, 88*3], obtaining [988, 638, 144]. Finally, combined with gene mutation data, such as a risk increase coefficient of 2 caused by a certain gene mutation, through the multiplication model, the corrected index vectors are further adjusted to [988*2, 638*2, 144*2], obtaining [186, 966, 373]. Finally, the set of corrected personalized index reference intervals is determined as [186, 966, 373] for subsequent fetal health assessment and diagnosis.

[0031] It should be noted that the second preprocessing module 120 is also used to: according to the correlation between the maternal age influence, weight difference, pregnancy history record and screening indicators in the standardized index vector set, group the individual characteristics of pregnant women through a clustering algorithm, judge the distribution offset of the nuchal translucency thickness and serum test data within each group, and obtain the grouped characteristic offset coefficient set. Obtain an index vector set, where the index vector set includes the correlation coefficients of maternal age influence, weight difference, pregnancy history record and screening indicators; perform standardization processing on the index vector set to obtain an initial correlation data set; according to the initial correlation data set, use the K-means clustering algorithm to group the pregnant women data according to individual characteristics to obtain a clustering grouping result set; for the clustering grouping result set, calculate the data distribution offset of the nuchal translucency within each group to obtain a nuchal translucency offset set; for the clustering grouping result set, calculate the data distribution offset of the serum test within each group to obtain a serum test offset set; according to the nuchal translucency offset set and the serum test offset set, judge the characteristic change trend of the distribution offset to obtain a characteristic offset coefficient set; obtain the characteristic offset coefficient set, if the characteristic offset coefficient exceeds the preset threshold, adjust the clustering grouping parameters, recalculate the distribution offset, and obtain an optimized characteristic offset coefficient set; according to the optimized characteristic offset coefficient set, determine the association mode between the individual characteristics of each group of pregnant women and the screening indicators to obtain a final grouped characteristic set.

[0032] Specifically, in the standardized index vector set, first perform a correlation analysis on the maternal age influence, weight difference and pregnancy history record, and use the Pearson correlation coefficient to calculate the correlation between each index and the screening index. For example, the correlation coefficient between maternal age and serum test data is 45, the correlation coefficient between weight difference and nuchal translucency thickness is 32, and the correlation coefficient between pregnancy history record and screening index is 28. Then, use the K-means clustering algorithm to group the individual characteristics of pregnant women, set the number of clusters to 3, calculate the similarity between samples through the Euclidean distance, and obtain three cluster centers after 100 iterations, which are [5, 6, 7], [3, 4, 5] and [8, 9, 0] respectively. Then, calculate the distribution offset of the nuchal translucency thickness and serum test data within each clustering group, and use the Kolmogorov-Smirnov test to obtain the offset of the first group as 12, the second group as 08, and the third group as 15. Finally, calculate the characteristic offset coefficient set according to the offset. The characteristic offset coefficient of the first group is 2, the second group is 9, and the third group is 5. These coefficients are used for the subsequent formulation of personalized screening strategies.

[0033] In an alternative manner, the third preprocessing module 130 is specifically used for: Obtain the corrected reference interval set, where the reference interval set includes data on serum PAPP-A concentration and free β-hCG concentration corresponding to maternal age influence, weight difference, and gestational age dependence; according to the reference interval set, extract the distribution characteristics of serum PAPP-A and free β-hCG through statistical analysis, and determine the correlation parameters with maternal age influence, weight difference, and gestational age dependence; obtain individual characteristics and ultrasonic measurement data, and through preprocessing, integrate the distribution characteristics of the serum PAPP-A and free β-hCG to obtain a standardized feature set; use a conditional probability model to fuse the standardized feature set, and calculate the joint probability distribution of individual characteristics and ultrasonic measurement and chromosomal abnormalities; update the conditional probability in the joint probability distribution through the Bayesian method to obtain a preliminary risk probability distribution for chromosomal abnormalities; determine whether the preliminary risk probability exceeds a preset threshold, if it exceeds, then adjust the fusion parameters of the conditional probability model in combination with the distribution characteristics to obtain a final personalized risk probability distribution; according to the final personalized risk probability distribution, obtain the personalized risk probability distribution characteristics of chromosomal abnormalities.

[0034] Specifically, from the corrected personalized index reference interval set, first extract the distribution characteristics of serum PAPP-A concentration and free β-hCG concentration corresponding to maternal age influence, weight difference, and gestational age dependence. For example, for a pregnant woman with a maternal age of 30 years, the reference interval for serum PAPP-A concentration is 5-5 MoM, and the free β-hCG concentration is 8-2 MoM; for a pregnant woman with a weight of 70 kg, the PAPP-A concentration is adjusted to 6-4 MoM, and the free β-hCG concentration is 7-1 MoM; for a pregnant woman with a gestational age of 12 weeks, the PAPP-A concentration is further adjusted to 7-3 MoM, and the free β-hCG concentration is 9-3 MoM. These adjustments are based on statistical analysis of a large amount of clinical data to ensure the accuracy of the reference interval. Then, through a model fusion method based on conditional probability, fuse the individual characteristics with the ultrasonic measurement data. For example, for a pregnant woman aged 35 years, weighing 65 kg, and with a gestational age of 11 weeks, the ultrasonic measurement data includes a fetal nuchal translucency thickness of 5 mm. Combining her serum PAPP-A concentration of 8 MoM and free β-hCG concentration of 5 MoM, use Bayes' theorem to calculate the risk probability of chromosomal abnormalities. The specific algorithm is: P(chromosomal abnormality|data)=P(data|chromosomal abnormality)*P(chromosomal abnormality) / P(data), where P(data|chromosomal abnormality) is estimated through historical case data, P(chromosomal abnormality) is the population base probability, and P(data) is the sum of the data distributions in normal and abnormal cases. Finally, obtain the personalized risk probability distribution for this pregnant woman. For example, the risk probability of trisomy 21 syndrome is 1:250, and the risk probability of trisomy 18 syndrome is 1:1000, thus providing an accurate basis for clinical decision-making.

[0035] It should be noted that the third preprocessing module 130 is further configured to: according to the data of the difference in weight affected by maternal age and the corresponding serum PAPP-A concentration and free β-hCG concentration depending on gestational week, extract the distribution characteristics of serum PAPP-A and free β-hCG through statistical analysis, adopt a conditional probability model to fuse individual characteristics and ultrasonic measurement data, determine whether the preliminary risk probability exceeds a threshold, and adjust the fusion parameters of the conditional probability model in combination with the distribution characteristics to obtain the final personalized risk probability distribution for chromosomal abnormalities. Obtain the serum PAPP-A and free β-hCG concentration data corresponding to maternal age, weight and gestational week, extract their distribution characteristics through statistical methods to obtain a standardized distribution parameter set; adopt a conditional probability model to fuse individual characteristics and ultrasonic measurement data, and calculate a joint probability distribution in combination with the standardized distribution parameter set to obtain a preliminary probability estimate; adopt a Bayesian method to update the conditional probability in the joint probability distribution, and determine a preliminary risk probability value in combination with the distribution characteristics; judge whether the preliminary risk probability value exceeds a preset threshold, and if it exceeds, adjust the fusion parameters to obtain optimized conditional probability model parameters; according to the optimized conditional probability model parameters, re-fuse individual characteristics and ultrasonic measurement data, calculate an adjusted joint probability distribution to obtain a final risk probability distribution; extract the probability distribution characteristics of chromosomal abnormalities for the final risk probability distribution to determine a personalized risk feature set; analyze the personalized risk feature set to obtain a distribution rule related to chromosomal abnormalities to obtain the final service output data.

[0036] Specifically, when analyzing the effects of maternal age, weight, and gestational week on the concentrations of serum PAPP-A and free β-hCG, the distribution characteristics are first extracted through big data analysis. For example, for pregnant women with a maternal age of 28 years, the mean value of the serum PAPP-A concentration distribution is 8 MoM, and the standard deviation is 2; the mean value of the free β-hCG concentration is 1 MoM, and the standard deviation is 3. For pregnant women with a weight of 60 kg, the mean value of the PAPP-A concentration distribution is adjusted to 9 MoM, and the standard deviation is 15; the mean value of the free β-hCG concentration is adjusted to 0 MoM, and the standard deviation is 25. For pregnant women with a gestational week of 10 weeks, the mean value of the PAPP-A concentration distribution is further adjusted to 0 MoM, and the standard deviation is 1; the mean value of the free β-hCG concentration is adjusted to 2 MoM, and the standard deviation is 2. These distribution characteristics are fitted through a Gaussian distribution model to ensure the accuracy and reliability of the data. Then, a conditional probability model is used to fuse the individual characteristics with the ultrasound measurement data. For example, for a pregnant woman aged 32 years, weighing 58 kg, and with a gestational week of 9 weeks, whose ultrasound measurement data shows the absence of the fetal nasal bone, combined with a serum PAPP-A concentration of 7 MoM and a free β-hCG concentration of 3 MoM, the conditional probability formula P(Chromosomal abnormality|Data) = P(Data|Chromosomal abnormality) * P(Chromosomal abnormality) / P(Data) is used for calculation, where P(Data|Chromosomal abnormality) is estimated to be 0.5 through historical case data, P(Chromosomal abnormality) is the population-based probability of 0.1, and P(Data) is the sum of the data distributions in normal and abnormal cases of 0.6. The preliminary calculation result shows that the chromosomal abnormality risk probability is 0.083, exceeding the preset threshold of 0.05. According to the distribution characteristics, the fusion parameters of the conditional probability model are adjusted, P(Data|Chromosomal abnormality) is corrected to 0.6, and the final risk probability is 0.1 after recalculation, thereby providing a more accurate basis for clinical decision-making.

[0037] In an alternative manner, the first screening module 140 is specifically configured to: Obtain the data of serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness to obtain the original test values; if the original test values exceed the preset reference interval, obtain abnormal marker data; obtain the maternal age influencing factor, and use a linear regression algorithm to calculate a preliminary risk adjustment value according to the abnormal marker data and the maternal age influencing factor; obtain the family genetic record, and perform secondary correction on the preliminary risk adjustment value according to the family genetic record to obtain a comprehensive risk value; use a Gaussian distribution model to process the comprehensive risk value to generate personalized risk probability distribution data; draw a curve according to the personalized risk probability distribution data to obtain a Down syndrome risk assessment curve; if the comprehensive risk value exceeds the preset threshold, trace the source of the abnormality of the original test values according to the abnormal marker data to determine the key influencing factor.

[0038] Specifically, in a real-time detection device, biomarker data such as the serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness of a pregnant woman are first obtained. For example, the serum PAPP-A concentration of a certain pregnant woman is 8 MoM (multiple of the median), the free β-hCG concentration is 2 MoM, and the nuchal translucency thickness is 5 mm. Then, the system compares these detected values with the personalized reference interval. If it exceeds the interval, risk calculation is further performed in combination with the maternal age and family genetic records. Suppose the pregnant woman is 35 years old and there is a history of Down syndrome in the family. The system will use the Bayesian algorithm to adjust the risk probability in combination with these factors. The specific calculation process is as follows: The baseline risk based on the maternal age is 1 / 250, the family genetic record increases the risk coefficient by 5 times, the abnormal serum PAPP-A and free β-hCG concentrations contribute risk coefficients of 2 and 3 respectively, and the abnormal nuchal translucency thickness contributes a risk coefficient of 4. Through the multiplication model, the final risk value is 1 / 250×5×2×3×4 = 1 / 96. The system plots this risk value on the Down syndrome risk assessment curve to form a personalized risk assessment report, providing a basis for clinical decision-making.

[0039] In an alternative approach, the second screening module is specifically configured to: Obtain the initial calculated value of the risk assessment within consecutive gestational weeks; perform data fitting on the initial calculated value using the linear interpolation method to obtain the smoothed change trend; determine the distribution law characteristics of the risk probability depending on the gestational week according to the change trend; if the distribution law characteristics deviate from the preset threshold, dynamically adjust the risk probability to obtain the adjusted risk probability value; draw a prediction curve according to the adjusted risk probability value; obtain the time series data of the prediction curve and determine the change trend depending on the gestational week; generate an updated strategy for dynamic monitoring for the change trend to obtain an optimized risk assessment result.

[0040] Specifically, in the risk assessment of Down syndrome, it is first necessary to obtain the dynamic monitoring data of pregnant women at different gestational weeks. For example, the risk calculation value at 12 weeks of gestation is 1:500, and at 16 weeks of gestation is 1:300. Through the linear interpolation method, these data points can be fitted. Assuming that at 14 weeks of gestation, the risk calculation value is 1:400, which is calculated by the linear interpolation formula (y = y1+(x - x1)*(y2 - y1) / (x2 - x1)), where x1 = 12, y1 = 500, x2 = 16, y2 = 300, and x = 14. Substituting these values into the formula gives y = 400. Next, analyze the dynamic distribution law of the risk probability depending on the gestational week. For example, between 12 weeks and 16 weeks of gestation, the risk probability shows a linear downward trend, and for each additional week, the risk probability decreases by approximately 50. Based on this law, the risk prediction curve can be dynamically adjusted. For example, at 18 weeks of gestation, the risk calculation value is expected to be 1:200, which is obtained by continuing to apply the linear interpolation method and combining the known data points and trend prediction. Finally, through information technology processing, a smooth risk prediction curve can be generated to provide a scientific basis for clinical decision-making.

[0041] In an alternative approach, the third screening module 160 is specifically configured to: Obtain clinical data, determine the initial parameters of the prediction model using statistical analysis methods to obtain the basic curve of risk prediction; for the basic curve, perform dynamic adjustment using time series analysis methods to obtain the adjusted prediction curve and determine the distribution characteristics of the risk probability; extract the risk probability from the adjusted prediction curve. If the risk probability is higher than the preset threshold, generate a high-risk label and judge the potential abnormal state; according to the high-risk label, obtain the preliminary classification of the screening result through signal mapping technology; after obtaining the preliminary classification, optimize the screening result using the decision tree algorithm to obtain the final output classification; compare the output classification with the clinical data to obtain the deviation value of the classification result; according to the deviation value, optimize the grading threshold using the iterative update method, and obtain the optimized risk prediction curve through cyclic processing.

[0042] Specifically, based on clinical research data, first, a logistic regression model is used to dynamically adjust the risk probability of patients. The model inputs include multiple characteristic variables such as age, gender, blood pressure, and blood sugar. The model parameters are fitted by the maximum likelihood estimation method. For example, the regression coefficient of age is 15, and the regression coefficient of systolic blood pressure is 08. The final output risk probability range is from 0 to 1. Then, according to the clinical guidelines, the grading thresholds are set. For example, the low-risk threshold is 2, the medium-risk threshold is 5, and the high-risk threshold is 8. The adjusted risk prediction curve is graded. If the risk probability of a certain patient is 85, which is higher than the preset high-risk threshold, the system automatically generates a high-risk warning signal and combines the decision tree algorithm to output and classify the screening results. For example, when the risk probability is greater than 8 and the age is greater than 65 years old, the output classification is "It is recommended that high-risk groups be intervened immediately". Throughout the process, the use of machine learning models and preset thresholds realizes the automatic grading of risk probabilities and the intelligent classification of screening results, ensuring the accuracy and operability of the analysis results.

[0043] The technical solution of this embodiment analyzes historical sample data, including pregnant women's age, weight, ethnic background, pregnancy history records, serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness, etc., to explore the distribution law of chromosomal abnormality risks. The principal component analysis method is used to standardize the detection data, and the clustering algorithm is used to group the characteristics of pregnant women to determine the characteristic offset coefficient. Combining factors such as fetal structural abnormalities, chromosomal abnormalities, and family genetic history, a personalized index reference interval is established. By fusing individual characteristics and ultrasonic measurement data, a personalized risk probability distribution is generated. Finally, the technical solution of this embodiment calculates the risk value and generates a dynamically adjusted risk prediction curve based on the real-time detection data and the personalized reference interval, providing an accurate basis for the clinical decision-making of Down syndrome risk assessment.

[0044] Embodiment 2 On the basis of Embodiment 1, the present invention provides an embodiment of a Down syndrome screening method based on multiple key indicators. The method includes the following steps: S1. Obtain the data of pregnant women's age, weight difference, ethnic background, pregnancy history records, serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness from the historical sample database, and explore the distribution law of chromosomal abnormality risks under the combination of maternal age influence and ethnic differences through statistical analysis to obtain a preliminary risk distribution characteristic set; S2. Map the serum test data and ultrasonic measurement data to a unified feature space to obtain a standardized index vector set, and perform offset correction processing on the standardized index vector set. Combining fetal structural abnormalities, chromosomal abnormalities, family genetic history, and gene mutations, determine the corrected personalized index reference interval set; S3. Extract the distribution characteristics of serum PAPP-A concentration and free β-hCG concentration corresponding to maternal age effect, weight difference and gestational age dependence from the corrected personalized index reference interval, fuse individual characteristics with ultrasound measurement data through a model fusion method based on conditional probability, and obtain a personalized risk probability distribution for chromosomal abnormalities; S4. According to the personalized risk probability distribution, the serum PAPP-A concentration, free β-hCG concentration and nuchal translucency thickness of the pregnant woman are obtained from the real-time detection device. If the test value exceeds the personalized reference interval, the risk calculation value is calculated in combination with the influence of maternal age and family genetic records to obtain the Down syndrome risk assessment curve; S5. According to the changing trend of the risk calculation value in the Down syndrome risk assessment curve, the linear interpolation method is used to fit the dynamic monitoring data within consecutive gestational weeks, determine the dynamic distribution law of the risk probability under gestational week dependence, and obtain the dynamically adjusted risk prediction curve; S6. Use the grading threshold determined based on clinical research data to grade the risk probability in the dynamically adjusted risk prediction curve. If the risk probability is higher than the preset threshold, a high-risk warning signal is generated to determine the output classification of the screening result.

[0045] It should be noted that the beneficial effects of the Down syndrome screening method based on multiple key indicators provided in the above embodiment are the same as the beneficial effects of the Down syndrome screening system 100 based on multiple key indicators, which will not be repeated here.

[0046] Example 3 Based on Example 1, the present invention further includes a psychological assessment module, an intervention decision module and an effect feedback module.

[0047] The psychological assessment module collects physiological data such as ECG signals in real time through wearable devices (such as smart bracelets), and extracts comprehensive features reflecting the psychological state by combining the questionnaire feedback submitted by the mobile terminal. The system's built-in assessment algorithm combines physiological signal features (such as heart rate variability patterns) with standardized psychological scores to generate a dynamic psychological stress index. When the index exceeds the set threshold, the graded intervention mechanism is triggered.

[0048] The intervention decision module selects intervention measures of corresponding intensity from the strategy library based on the combination of risk level and psychological stress characteristics. For high-risk related cases, the system automatically connects to online experts for video counseling; for moderately stressed patients, meditation training courses and patient community entrances are pushed. If the system detects that the patient has not completed the intervention task on time, a reminder notification will be sent to the preset family contact.

[0049] The effect feedback module evaluates the effectiveness of the intervention measures by continuously tracking the physiological signal change curve and the patient's daily mood log. When the improvement trend does not meet the expectations, the system automatically upgrades the intensity of the intervention plan, for example, changing from autonomous training to manual intervention. When both the screening risk and psychological stress exceed the standard, the genetic counseling connection unit generates an interactive report containing medical explanations and schedules a remote consultation with a professional genetic counselor.

[0050] The above description is only a preferred embodiment of the present invention and an explanation of the applied technical principles. The above embodiments are exemplary and should not be construed as limitations of the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A Down syndrome screening system based on multiple key indicators, characterized in that: include: The first preprocessing module is used to obtain the data of maternal age, weight difference, ethnic background, pregnancy history, serum PAPP-A concentration, free β-hCG concentration and nuchal translucency thickness from the historical sample database, and to mine the distribution law of chromosomal abnormality risk under the combination of maternal age and ethnic differences through statistical analysis to obtain a preliminary risk distribution feature set; The second preprocessing module is used to map the serum test data and ultrasound measurement data to a unified feature space to obtain a standardized indicator vector set, and to perform offset correction on the standardized indicator vector set, and to determine a corrected personalized indicator reference interval set in combination with fetal structural abnormalities, chromosomal abnormalities, family genetic disease history, and gene mutations; The third preprocessing module is used to extract the distribution characteristics of serum PAPP-A concentration and free β-hCG concentration corresponding to maternal age effect, weight difference and gestational age dependence from the corrected personalized index reference interval, and fuse individual characteristics with ultrasound measurement data through a model fusion method based on conditional probability to obtain a personalized risk probability distribution for chromosomal abnormalities; The first screening module is used to obtain the serum PAPP-A concentration, free β-hCG concentration and nuchal translucency thickness of pregnant women from the real-time detection equipment according to the personalized risk probability distribution. If the test value exceeds the personalized reference interval, the risk calculation value is calculated in combination with the influence of maternal age and family genetic records to obtain the Down syndrome risk assessment curve; The second screening module is used to fit the dynamic monitoring data within consecutive gestational weeks by linear interpolation method through the changing trend of the risk calculation value in the Down syndrome risk assessment curve, determine the dynamic distribution law of the risk probability under gestational week dependence, and obtain the dynamically adjusted risk prediction curve; The third screening module is used to grade the risk probability in the dynamically adjusted risk prediction curve using a grading threshold determined based on clinical research data. If the risk probability is higher than the preset threshold, a high-risk warning signal is generated to determine the output classification of the screening result.

2. The Down syndrome screening system based on multiple key indicators according to claim 1, characterized in that: The first preprocessing module is specifically used for: Obtain historical sample data including maternal age, weight difference, ethnic background, pregnancy history, serum PAPP-A concentration, free β-hCG concentration, and nuchal translucency thickness to construct the original data set; Calculate the correlation strength between maternal age and chromosomal abnormalities in the original data set using statistical analysis methods to obtain the distribution characteristics of the influence of maternal age; Performing cluster analysis on the ethnic background and chromosomal abnormalities in the original data set to determine the distribution pattern of ethnic differences; If the distribution characteristics of the maternal age effect exceed a preset threshold, the risk adjustment factor is calculated in combination with the serum PAPP-A concentration and the free β-hCG concentration to obtain an adjusted maternal age risk distribution; By integrating the nuchal translucency thickness and the adjusted maternal age risk distribution through regression analysis, a comprehensive risk distribution characteristic is determined; Obtaining cross-statistical results of the comprehensive risk distribution characteristics and the pregnancy history records to determine the multidimensional distribution law of chromosomal abnormalities; Generate feature set data according to the multidimensional distribution law and output a preliminary risk distribution feature set.

3. The Down syndrome screening system based on multiple key indicators according to claim 1, characterized in that: The second preprocessing module is specifically used for: Acquire serum test data and ultrasound measurement data, process the serum test data and ultrasound measurement data using principal component analysis, map the serum test data and ultrasound measurement data to a unified feature space, and obtain a standardized indicator vector set; For the serum test data, the variance of PAPP-A concentration and free β-hCG concentration is calculated to determine the range of measurement deviation caused by equipment differences; If the variance of the PAPP-A concentration or the free β-hCG concentration exceeds a preset threshold, normalizing the serum test data to obtain an adjusted concentration data set; According to the adjusted concentration data set and the ultrasonic measurement data, a principal component analysis method is used to extract main characteristic components to obtain a characteristic matrix after dimension reduction; Calculate the distribution distance of each sample in the feature space through the feature matrix, judge the measurement consistency between samples, and obtain the consistency judgment result; According to the consistency judgment result, the samples are grouped using the K-means clustering algorithm to obtain a classified vector set; For the classified vector set, the standardized index mean of each group is calculated to obtain a standardized index vector set.

4. The Down syndrome screening system based on multiple key indicators according to claim 1, characterized in that: The second preprocessing module is specifically used for: Obtaining a standardized indicator vector set, and performing offset correction processing on the standardized indicator vector set using a least square method to obtain a corrected preliminary indicator set; Acquiring fetal structural abnormality data, extracting abnormal feature values ​​from the fetal structural abnormality data, and fusing the abnormal feature values ​​with the preliminary indicator set to obtain a first fused indicator set; Acquire chromosome abnormality data, extract variation feature values ​​from the chromosome abnormality data, and perform weighted processing on the variation feature values ​​and the first fusion index set to obtain a second fusion index set; Acquiring family genetic disease history data, extracting risk factors from the family genetic disease history data, and adjusting the second fusion indicator set by using a linear regression method to obtain a third fusion indicator set; Acquire gene mutation data, extract mutation site information from the gene mutation data, determine whether the mutation site exceeds a preset threshold, and if so, perform weighted calibration on the third fusion indicator set to obtain a fourth fusion indicator set; A personalized reference interval set is determined according to the fourth fusion indicator set, and the degree of deviation of abnormal indicators is determined by comparing the fourth fusion indicator set with the personalized reference interval set to obtain a final corrected personalized indicator set.

5. The Down syndrome screening system based on multiple key indicators according to claim 1, characterized in that: The third preprocessing module is specifically used for: Obtaining a corrected reference interval set, wherein the reference interval set includes serum PAPP-A concentration and free β-hCG concentration data corresponding to maternal age effect, weight difference and gestational age dependence; Based on the reference interval set, the distribution characteristics of serum PAPP-A and free β-hCG were extracted by statistical analysis to determine their correlation parameters with maternal age effects, weight differences, and gestational age dependence; Obtaining individual characteristics and ultrasound measurement data, integrating the distribution characteristics of the serum PAPP-A and free β-hCG through preprocessing, and obtaining a standardized feature set; A conditional probability model is used to fuse the standardized feature set to calculate the joint probability distribution of individual characteristics and ultrasound measurements with chromosomal abnormalities; Updating the conditional probability in the joint probability distribution by using a Bayesian method to obtain a preliminary risk probability distribution for chromosomal abnormalities; Determine whether the preliminary risk probability exceeds a preset threshold, and if so, adjust the fusion parameters of the conditional probability model in combination with the distribution characteristics to obtain a final personalized risk probability distribution; According to the final personalized risk probability distribution, the personalized risk probability distribution characteristics of the chromosomal abnormality are obtained.

6. The Down syndrome screening system based on multiple key indicators according to claim 1, characterized in that: The second screening module is specifically used for: Obtain initial calculated values ​​of risk assessment within consecutive gestational weeks; The linear interpolation method is used to perform data fitting on the initial calculated value to obtain a smoothed variation trend; According to the changing trend, the distribution law characteristics of the risk probability under gestational age dependence are determined; If the distribution law characteristic deviates from a preset threshold, dynamically adjusting the risk probability to obtain an adjusted risk probability value; Draw a prediction curve according to the adjusted risk probability value; Obtaining time series data of the prediction curve to determine the changing trend under gestational age dependence; According to the change trend, a dynamic monitoring update strategy is generated to obtain an optimized risk assessment result.

7. The Down syndrome screening system based on multiple key indicators according to claim 1, characterized in that: Also includes: The psychological assessment module is used to obtain the patient's physiological signal data and psychological assessment data in real time when the third screening module generates a high-risk warning signal, generate a multi-dimensional psychological state feature vector based on the risk level, and determine whether to trigger a psychological intervention signal; The psychological assessment module is specifically used for: Collect patients' real-time physiological signals through wearable devices and extract characteristic parameters related to the autonomic nervous system; Synchronously obtain the psychological state feedback information submitted by the patient and convert it into standardized psychological assessment parameters; Comprehensively analyzing the physiological characteristic parameters, psychological assessment parameters and current risk level to generate a psychological stress index; When the psychological stress index exceeds a preset range, a psychological intervention signal is triggered and a state vector containing risk-related features is generated.

8. The Down syndrome screening system based on multiple key indicators according to claim 7, characterized in that: Also includes: An intervention decision module, used for matching a corresponding intervention plan from a multi-level intervention strategy library according to the state vector; The intervention decision module is specifically used for: Establish a multi-level intervention strategy that includes immediate consultation, self-training, and professional referral; Generate matching parameters according to multiple dimensional features of the state vector and dynamically select the type of intervention strategy; When it is detected that the psychological stress index is strongly correlated with the high risk level, real-time interactive intervention by professionals will be initiated as a priority; If the psychological stress index is within the adjustable range, customized training content and community support resources will be pushed to the patient terminal; When it is monitored that the intervention response does not meet expectations, the family collaborative intervention mechanism is automatically triggered.

9. The Down syndrome screening system based on multiple key indicators according to claim 1, characterized in that: Also includes: The effect feedback module is used to update the psychological state assessment model through dynamic monitoring of physiological signals and interactive feedback data; The effect feedback module is specifically used for: Continuously collect data on physiological signal change trends during the intervention implementation cycle; Obtain subjective experience feedback from patients through an interactive interface and extract emotional state characteristics; Conduct correlation trend analysis on physiological and psychological data, and automatically increase the level of intervention intensity when the improvement rate is detected to be lower than expected; The updated feature data are input into the adaptive model to optimize the psychological stress assessment parameters for the next stage.

10. The Down syndrome screening system based on multiple key indicators according to claim 9, characterized in that: The effect feedback module also includes a genetic counseling connection unit, which automatically generates a visual report integrating the screening results and the psychological assessment when the high-risk prompt signal and the psychological intervention effect data meet the preset association conditions, and starts the genetic counseling service channel; The genetic counseling connection unit is specifically used for: Conduct spatiotemporal correlation analysis between the dynamic risk prediction curve and the trend of psychological stress changes; When a sustained positive correlation between risk level and psychological stress is detected, a priority consultation channel is triggered; The complete assessment record is transmitted to the designated genetic counseling terminal via a secure encrypted protocol.

Images