Disease analysis method based on multi-target sensing array

Abstract

The invention relates to the technical field of disease analysis and prediction, and discloses a disease analysis method based on a multi-target sensor array, which comprises the following steps: deploying a biosensor array, immersing the biosensor array into a biological sample to be detected, collecting biomarker parameters in the biological sample to be detected, pre-processing the collected biomarker parameters, and analyzing the pre-processed biomarker parameters in the biological sample to be detected. The method comprises the steps of preprocessing biomarker parameters, analyzing the preprocessed biomarker parameters through a biological parameter detection model, outputting the analyzed biomarker parameters to the biological parameter detection model through a disease monitoring model, performing disease prediction, and optimizing the biological parameter detection model and the disease monitoring model through performance indexes. A health warning threshold value is set according to personal signs, safety and privacy protection is set according to disease privacy levels, and data encryption is set; the high-precision acquisition of biomarker parameters and the precise prediction of the disease probability are realized through the biosensor.

Description

Technical Field

[0001] The invention relates to the technical field of disease analysis and prediction, and discloses a disease analysis method based on a multi-target sensor array. Background Art

[0002] Technology based on multi-target sensor arrays has extensive and important applications in the field of disease detection. This technology uses multiple sensors to detect different biomarkers at the same time, which can provide more comprehensive and accurate disease diagnosis. However, multi-target sensor array technology still faces many challenges in its application. Different sensors may have different sensitivities to the same marker. This difference may cause some sensors to be unable to detect low concentrations of markers, thereby affecting the overall test results. Sensors have cross-reactions to different biomarkers, that is, some sensors may react to multiple markers at the same time, resulting in a decrease in the accuracy of the test results. Multi-target sensor arrays generate a large amount of data, which requires efficient data processing and fusion algorithms to extract useful information from them. Multi-target sensor array technology has broad application potential in disease prediction, but it still faces the difficulties of sensitivity and specificity, data processing and analysis, and biocompatibility and safety. Therefore, a disease analysis method based on multi-target sensor arrays is urgently needed.

[0003] For example, the existing Chinese patent with the publication number CN115979735A discloses a human urine odor detection device with disease detection function, which is used to detect volatile gases in human urine and can quickly and non-invasively diagnose diseases. Among them, it includes a central processing and control unit, a urine collection and storage unit, a gas transmission unit, a urine odor signal acquisition unit, a user information storage unit, a data processing unit, a disease prediction algorithm unit, a user disease risk status display unit, a wireless communication unit, a database and a device terminal unit. The user discharges urine into the urine collection and storage unit, and the gas transmission unit transmits the gas to the urine odor signal acquisition unit, and the electronic nose sensor array collects data and sends it to the central processing and control unit. The data processing unit processes the data, and the disease prediction algorithm unit evaluates the disease risk status, and then the disease risk status display unit displays it respectively. The wireless communication unit sends the data to the database, and the device terminal can obtain the data on the database in real time.

[0004] However, the above patents have the following problems: the sensitivity and selectivity of the sensor cannot be guaranteed, especially in low concentration conditions, the sensor may not be able to accurately detect specific biomarkers, there is a lack of compensation units, the data processing is relatively simple, and it is impossible to effectively extract useful biomarkers from a large amount of data; the algorithm is mainly implemented through hardware circuits, and the processing accuracy is much lower than software control, resulting in low data processing accuracy, cumbersome subsequent maintenance, and the inability to achieve automatic prediction effects. Summary of the invention

[0005] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0006] In order to solve the above technical problems, the main purpose of the present invention is to provide a disease analysis method based on a multi-target sensor array, comprising:

[0007] S1. deploying a biosensor array and immersing it in a biological sample to be tested to collect biomarker parameters in the biological sample to be tested;

[0008] S2. preprocessing the collected biomarker parameters, and analyzing the preprocessed biomarker parameters through a biological parameter detection model;

[0009] S3, using the disease monitoring model to analyze the biomarker parameters output by the biological parameter detection model to predict the disease;

[0010] S4. Optimize biological parameter detection models and disease monitoring models through performance indicators, and set health warning thresholds based on personal signs;

[0011] S5. Set security and privacy protection according to the disease privacy level and set data encryption;

[0012] S6. Complete the disease prediction of biological samples and store them in the database. The biomarker parameters and predicted diseases are used as training sets for biological parameter detection models and disease monitoring models.

[0013] As a preferred solution of the disease analysis method based on the multi-target sensor array of the present invention, wherein:

[0014] The biosensor array includes a plurality of biosensors forming an array;

[0015] The biosensor comprises: a biosensitive gate, an organic semiconductor channel, a drain electrode, a source electrode and a substrate;

[0016] A fixed voltage V is applied between the source electrode and the drain electrode. D , a drain current I is applied between the source electrode and the drain electrode D , if the biosensitive gate is applied with a voltage V G , and does not change the fixed voltage V D , changing the drain current I D , then the ions in the electrolyte solution can reversibly enter the organic semiconductor channel and change the conductivity in the organic semiconductor channel;

[0017] By combining the biosensors into an array and immersing them in a biological sample to be tested, corresponding biomarker parameters, including biomarker names and concentration parameters, can be collected.

[0018] As a preferred solution of the disease analysis method based on the multi-target sensor array of the present invention, wherein:

[0019] The preprocessing includes data cleaning, feature extraction and data standardization;

[0020] The biomarker parameters collected by the biosensor array are detected and predicted by the bioparameter detection model. The calculation expression of the bioparameter detection model is as follows:

[0021] G(Q)=W T *Q T-1 +γ T

[0022] Among them, G(Q) is the biomarker prediction value output by the biological parameter detection model, W T is the weight matrix of the Tth layer, Q T-1 is the biomarker parameter output by the T-1th layer, γ T is the bias of the Tth layer.

[0023] As a preferred solution of the disease analysis method based on the multi-target sensor array of the present invention, wherein:

[0024] The training set of the disease monitoring model includes biomarkers detected and predicted by the biological parameter detection model, known healthy and diseased sample data, and disease parameters corresponding to the biomarkers in the database;

[0025] The disease monitoring model is essentially a logistic regression, and the output of the logistic regression is a probability value between 0 and 1, indicating the possibility of illness. The calculation expression of the disease monitoring model is as follows:

[0026]

[0027] Where P(Y=1|X) is the probability of disease under given biomarker parameter X, α 0 is the probability intercept term, α 1 is the regression coefficient of the first biomarker parameter, α 2 is the regression coefficient of the second biomarker parameter, α N is the regression coefficient of the Nth biomarker parameter, X 1 is the first biomarker parameter, X 2 is the second biomarker parameter, X N is the Nth biomarker parameter;

[0028] The classifier is used to maximize the interval between the diseased and the non-disease, and the classifier is used to iterate the interval between the diseased and the non-disease affected by each biomarker parameter. The classifier calculation expression is as follows:

[0029]

[0030] Among them, f(s) is the interval between the classifier output affecting the diseased and non-disease, a z is the probability factor of disease caused by abnormality of the zth biomarker parameter, b is bias compensation, P z is the probability that the abnormality of the zth biomarker parameter leads to illness, K(s i ,s 0 ) is the kernel function used to calculate s i With s 0 The inner product of the sample in the high-dimensional feature space, s 0 is the standard normal biomarker parameter, s i is the i-th group of biomarker parameters, z is the serial number of the biomarker parameter type, m is the total number of biomarker parameter types, and sign() is a function that returns an integer variable.

[0031] As a preferred solution of the disease analysis method based on the multi-target sensor array of the present invention, wherein:

[0032] The performance indicators include cross entropy and correction factors;

[0033] The cross entropy controls the steepness of the loss function by setting an adjustment factor, and the loss function is used to minimize the absolute difference between the predicted value and the actual value, and is weighted by an exponential function;

[0034] To compensate for lost values ​​and correct predicted values, a correction factor is applied after the disease monitoring model prediction, which is calculated based on historical errors or deviations to reduce the prediction error.

[0035] As a preferred solution of the disease analysis method based on the multi-target sensor array of the present invention, wherein:

[0036] The calculation expression of the cross entropy is as follows:

[0037]

[0038] in, is the cross entropy difference, which is used to reflect the loss between the actual value monitored by the biological parameter detection model and the output predicted value. log is the logarithmic function, a is the adjustment factor, y is the actual value monitored by the biological parameter detection model, is the predicted value output by the biological parameter detection model, n is the actual value data group monitored by the biological parameter detection model output, and i is the label sequence number of the data group.

[0039] As a preferred solution of the disease analysis method based on the multi-target sensor array of the present invention, wherein:

[0040] The sensor array data collection loss suppression method process is as follows:

[0041] S2001, establishing a temperature parameter gradient matrix of the time domain signal at the kth moment, and taking the standard temperature parameter matrix as a reference matrix;

[0042] S2002, performing iterative calculation using the temperature parameter gradient matrix of the time domain signal at the kth moment and the standard temperature parameter matrix;

[0043] S2003, calculating an overfitting factor through the temperature parameter gradient matrix of the time domain signal at the kth moment, the standard temperature parameter matrix, the temperature parameter gradient matrix rank conversion of the time domain signal at the kth moment, and the standard temperature parameter matrix rank conversion;

[0044] S2004, correcting the overfitting part by the overfitting factor, ensuring that the temperature causes a large loss of data collected by the sensor array at the k+1th moment, and preventing the iterative calculation of the temperature parameter gradient matrix and the standard temperature parameter matrix of the time domain signal from deviating from the actual situation;

[0045] S2005. Project the temperature parameter gradient matrix and the standard temperature parameter matrix at the kth moment to ensure that the temperature parameter matrix is ​​the best reference matrix as a scalar.

[0046] As a preferred solution of the disease analysis method based on the multi-target sensor array of the present invention, wherein:

[0047] The process of disease analysis, diagnosis and prediction is as follows:

[0048] S1001. Collect a data set containing biomarker parameters and corresponding disease states, process missing values ​​and abnormal values ​​of biomarker parameters, and perform standardization or normalization, as well as possible feature selection or dimensionality reduction;

[0049] S1002, selecting the biomarker parameters that are most useful for predicting the disease, establishing a disease monitoring model, and calculating the probability of disease under given biomarker parameters by maximizing whether the disease occurs;

[0050] S1003. If the calculated disease probability is greater than the preset probability critical value, the patient is likely to be ill and further examination is performed; if the calculated disease probability is less than the preset probability critical value, the patient is likely not to be ill and the biomarker parameters are output;

[0051] S1004. Analyze the probability of the patient suffering from the disease by combining the patient's historical examination data;

[0052] S1005. Complete the patient's superficial analysis and prediction of the disease.

[0053] As a preferred solution of the disease analysis method based on the multi-target sensor array of the present invention, wherein:

[0054] The database is used to store the prediction results together with the biomarker parameters in the database;

[0055] Data reuse is used to use this data as part of the dataset for future model training to continuously improve the accuracy of the model.

[0056] Beneficial effects of the present invention:

[0057] Multi-target sensor arrays can detect multiple biomarkers simultaneously, improve the accuracy of disease prediction, identify early signs of disease, provide opportunities for early intervention, and improve treatment outcomes. Artificial intelligence algorithms can process complex data sets, reveal underlying disease patterns and risk factors, and enable real-time monitoring of health status, timely adjustment of predictions and intervention measures. Through multiple data sources and advanced algorithms, the probability of misdiagnosis and missed diagnosis can be reduced, the reliability of diagnosis can be improved, the sensor array acquisition data loss suppression can be increased, and the accuracy of signals acquired by sensors can be significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0059] Figure 1 A system composition diagram of a disease analysis method based on a multi-target sensor array according to the present invention;

[0060] Figure 2 A flowchart of the disease analysis method based on a multi-target sensor array of the present invention;

[0061] Figure 3 The present invention is a flow chart of a method for suppressing loss of sensor array data acquisition in a disease analysis method based on a multi-target sensor array. DETAILED DESCRIPTION

[0062] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0063] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0064] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0065] Embodiment 1:

[0066] like Figure 1 As shown, the disease analysis method based on the multi-target sensor array includes:

[0067] S1. deploying a biosensor array and immersing it in a biological sample to be tested to collect biomarker parameters in the biological sample to be tested;

[0068] Wherein, the biosensor array includes a plurality of biosensors forming an array to achieve simultaneous detection of multiple biomarkers;

[0069] Further, the biosensor includes: a biosensitive gate, an organic semiconductor channel, a drain electrode, a source electrode and a substrate;

[0070] Furthermore, the biosensitive gate comprises a gate and a biosensitive layer arranged relatively to each other, and an electrode modification layer arranged between the gate and the biosensitive layer, wherein the surface of the biosensitive layer is modified with an immune response structure;

[0071] The organic semiconductor channel is placed between the drain electrode and the source electrode, connecting the drain electrode and the source electrode;

[0072] The biosensitive gate is placed on the top layer and connected to the device below through the electrolyte solution. A fixed voltage V is applied between the source electrode and the drain electrode. D , a drain current I is applied between the source electrode and the drain electrode D , if the biosensitive gate is applied with a voltage V G , and does not change the fixed voltage V D , changing the drain current I D , then the ions in the electrolyte solution can reversibly enter the organic semiconductor channel and change the conductivity in the organic semiconductor channel;

[0073] By combining the biosensors into an array and immersing them in the biological sample to be tested, the corresponding biomarker parameters can be collected, including biomarker name, concentration and other parameters.

[0074] Furthermore, the sensor array is optimized through loss suppression method.

[0075] S2. preprocessing the collected biomarker parameters, and analyzing the preprocessed biomarker parameters through a biological parameter detection model;

[0076] Among them, preprocessing includes data cleaning, feature extraction and data standardization;

[0077] Data cleaning is used to remove invalid or erroneous data points, such as outliers caused by sensor failures;

[0078] Feature extraction is used to extract useful features from raw signals, such as concentration changes, signal intensity, etc.;

[0079] Data standardization is used to standardize the extracted features so that they meet the requirements of the machine learning model.

[0080] The biological parameter detection model includes selecting biomarkers as features, further screening features by sorting correlation features, extracting trend features and periodic features if it is a time series, and establishing a biological parameter detection model through time series. If it is a non-time series, meaningful features are extracted through signal intensity, including statistical features such as maximum, minimum, mean, and variance of signal intensity;

[0081] Furthermore, by classifying the biomarker categories, the biomarker parameters collected by the biosensor array are detected and predicted through the biological parameter detection model. The calculation expression of the biological parameter detection model is as follows:

[0082] G(Q)=W T *Q T-1 +γ T

[0083] Among them, G(Q) is the biomarker prediction value output by the biological parameter detection model, W T is the weight matrix of the Tth layer, Q T-1 is the biomarker parameter output by the T-1th layer, γ T is the bias of the Tth layer;

[0084] Biomarker parameters can be represented as a matrix Q, where each row represents a sample and each column represents the measured value of a biomarker. A biological parameter detection model G(Q) is constructed, which can predict the category of biomarkers or certain key parameters based on these input data.

[0085] The biological parameter detection model G(Q) is a complex nonlinear function, which can be implemented through a neural network. The input layer of the neural network receives the sensor data Q, the output layer outputs the predicted category or parameter value, and the middle layer (hidden layer) is responsible for learning the nonlinear characteristics of the data.

[0086] The neural network of the biological parameter detection model includes an input layer, a hidden layer, an output layer, and a loss function;

[0087] The size of the input layer matches the dimensionality of the biosensor array, i.e., each neuron corresponds to a measurement of a biomarker.

[0088] The hidden layer contains multiple neurons, each of which receives weighted input from the input layer or other hidden layers and produces output through an activation function (such as ReLU, sigmoid, or tanh). These outputs will serve as input to the next layer.

[0089] The size of the output layer depends on the specific requirements of the prediction task. If the task is classification, the size of the output layer is equal to the number of categories, and each neuron corresponds to the predicted probability of a category. If the task is regression, the size of the output layer is equal to the number of parameters that need to be predicted.

[0090] The loss function is used to measure the difference between the model's predicted value and the actual value. For classification tasks, commonly used loss functions include cross entropy loss; for regression tasks, commonly used loss functions include mean squared error (MSE).

[0091] Optimization algorithms (such as gradient descent, Adam, etc.) are used to update the weights of the neural network in order to minimize the loss function.

[0092] S3, using the disease monitoring model to analyze the biomarker parameters output by the biological parameter detection model to predict the disease;

[0093] The training set of the disease monitoring model includes biomarkers detected and predicted by the biological parameter detection model, known healthy and diseased sample data, and disease parameters corresponding to biomarkers in the database;

[0094] Training disease monitoring models to predict whether or not a person has a disease or the likelihood of having a disease from biomarker parameters;

[0095] Model application: The data analyzed by the biological parameter detection model is input into the disease monitoring model to obtain the prediction results.

[0096] Convert the prediction results into a readable form, such as "higher risk of a certain disease, lower risk of a certain disease", etc.

[0097] Furthermore, the disease monitoring model is essentially a logistic regression, and the output of the logistic regression is a probability value between 0 and 1, indicating the possibility of illness. The calculation expression of the disease monitoring model is as follows:

[0098]

[0099] Where P(Y=1|X) is the probability of disease under given biomarker parameter X, α 0 is the probability intercept term, α 1 is the regression coefficient of the first biomarker parameter, α 2 is the regression coefficient of the second biomarker parameter, α N is the regression coefficient of the Nth biomarker parameter, X 1 is the first biomarker parameter, X 2 is the second biomarker parameter, X N is the Nth biomarker parameter;

[0100] The classifier is used to maximize the interval between the diseased and the non-disease. That is, the probability of disease cannot determine whether one is completely diseased or completely healthy. Therefore, an interval is needed to determine the degree of disease and the degree of non-disease (health). Therefore, the classifier is used to iterate the interval between the diseased and non-disease affected by each biomarker parameter. The classifier calculation expression is as follows:

[0101]

[0102] Among them, f(s) is the interval between the classifier output affecting the diseased and non-disease, a z is the probability factor of disease caused by abnormality of the zth biomarker parameter, b is bias compensation, P z is the probability that the abnormality of the zth biomarker parameter leads to illness, K(s i ,s 0 ) is the kernel function used to calculate s i With s 0 The inner product of the sample in the high-dimensional feature space, s 0 is the standard normal biomarker parameter, s i is the i-th group of biomarker parameters, z is the serial number of the biomarker parameter type, m is the total number of biomarker parameter types, and sign() is a function that returns an integer variable;

[0103] Furthermore, the kernel function where σ is the variance of the biomarker parameter and the standard biomarker parameter, and l is the length scale parameter.

[0104] S4. Optimize biological parameter detection models and disease monitoring models through performance indicators, and set health warning thresholds based on personal signs;

[0105] The cross entropy is used to reflect the loss value between the biomarker parameters predicted by the biological parameter detection model and the actual biomarker parameters, and the loss value is compensated to correct and optimize the predicted value of the biological parameter detection model;

[0106] Furthermore, the cross entropy controls the steepness of the loss function by setting an adjustment factor. This loss function attempts to minimize the absolute difference between the predicted value and the actual value, and is weighted by an exponential function. The calculation expression is as follows:

[0107]

[0108] in, is the cross entropy difference, which is used to reflect the loss between the actual value monitored by the biological parameter detection model and the output predicted value. log is the logarithmic function, a is the adjustment factor, y is the actual value monitored by the biological parameter detection model, is the predicted value output by the biological parameter detection model, n is the actual value data group output by the biological parameter detection model, and i is the label number of the data group;

[0109] The disease monitoring model is continuously optimized through loss function regression, the model is retrained regularly with new data, and correction factors are output based on the evaluation results to adjust the disease monitoring model parameters.

[0110] Furthermore, in order to compensate for the lost value and correct the predicted value, a correction factor can be applied after the disease surveillance model prediction, and the correction factor can be calculated based on historical errors or deviations, thereby gradually reducing the prediction error;

[0111] The correction factor calculation expression is as follows:

[0112]

[0113] Where C is the correction factor, t is the time step, and j is the time series number;

[0114]

[0115] Among them, y curr is the optimized predicted value.

[0116] Furthermore, the health alert threshold settings include personalized settings and dynamic adjustments;

[0117] Furthermore, personalized settings are to set personalized health warning thresholds based on personal historical health data and physiological characteristics.

[0118] Dynamic adjustment means that the warning threshold can be dynamically adjusted to adapt to individual changes as time goes by and new data is input.

[0119] S5. Set security and privacy protection according to the disease privacy level and set data encryption;

[0120] Among them, security and privacy protection include data encryption and access control;

[0121] Furthermore, data encryption encrypts sensitive data using a strong encryption algorithm (such as AES).

[0122] Furthermore, access control ensures that only authorized users can access data, ensuring data security.

[0123] Privacy level settings include disease privacy level settings, user control, and transparency;

[0124] Furthermore, the disease privacy level setting sets the patient's disease disclosure level by disease severity and privacy;

[0125] User controls allow users to control how and to what extent their data is used.

[0126] Transparency ensures that users understand how their data is used.

[0127] S6. Complete the disease prediction of biological samples and store them in the database. The biomarker parameters and predicted diseases are used as training sets for biological parameter detection models and disease monitoring models.

[0128] Furthermore, the database storage is used to store the prediction results together with the biomarker parameters in a database.

[0129] Data reuse is used to use this data as part of the dataset for future model training to continuously improve the accuracy of the model.

[0130] Embodiment 2:

[0131] like Figure 2 As shown, the disease analysis, diagnosis and prediction method process is as follows:

[0132] S1001. Collect a data set containing biomarker parameters and corresponding disease states, process missing values ​​and abnormal values ​​of biomarker parameters, and perform standardization or normalization, as well as possible feature selection or dimensionality reduction;

[0133] S1002, selecting the biomarker parameters that are most useful for predicting the disease, establishing a disease monitoring model, and calculating the probability of disease under given biomarker parameters by maximizing whether the disease occurs;

[0134] Furthermore, the calculation of maximizing whether or not a patient is ill includes the patient's disease type, the patient's disease and test history, and real-time patient life marker parameters;

[0135] S1003. If the calculated disease probability is greater than the preset probability critical value, the patient is likely to be ill and needs further examination; if the calculated disease probability is less than the preset probability critical value, the patient is likely not ill and the biomarker parameters are output;

[0136] S1004. Analyze the probability of the patient suffering from the disease by combining the patient's historical examination data;

[0137] S1005. Complete the patient's superficial analysis and prediction of the disease.

[0138] The data collected by the sensor array is affected by the ambient temperature and the concentration of the liquid to be tested, so the loss of collected data is compensated by cross suppression;

[0139] like Figure 3 As shown, taking the influence of ambient temperature parameters as an example, the process of the sensor array data acquisition loss suppression method is as follows:

[0140] S2001, establishing a temperature parameter gradient matrix of the time domain signal at the kth moment, and taking the standard temperature parameter matrix as a reference matrix;

[0141] S2002, performing iterative calculation using the temperature parameter gradient matrix of the time domain signal at the kth moment and the standard temperature parameter matrix;

[0142] S2003, calculating an overfitting factor through the temperature parameter gradient matrix of the time domain signal at the kth moment, the standard temperature parameter matrix, the temperature parameter gradient matrix rank conversion of the time domain signal at the kth moment, and the standard temperature parameter matrix rank conversion;

[0143] S2004, correcting the overfitting part by the overfitting factor, ensuring that the temperature causes a large loss of data collected by the sensor array at the k+1th moment, and preventing the iterative calculation of the temperature parameter gradient matrix and the standard temperature parameter matrix of the time domain signal from deviating from the actual situation;

[0144] S2005. Project the temperature parameter gradient matrix and the standard temperature parameter matrix at the kth moment to ensure that the temperature parameter matrix is ​​the best reference matrix as a scalar.

[0145] It is important to note that the construction and arrangement of the present application shown in a plurality of different exemplary embodiments are only exemplary. Although only two embodiments are described in detail in this disclosure, it should be easily understood by the personnel referring to this disclosure that many modifications are possible, for example, the size, scale, structure, shape and ratio of various elements, and parameter values ​​(e.g., temperature, pressure, etc.), installation arrangement, use of materials, color, directional changes, etc., without departing substantially from the novel teachings and advantages of the subject matter described in the application. For example, the element shown as integrally formed can be composed of multiple parts or elements, the position of the element can be inverted or otherwise changed, and the nature or number or position of the discrete element can be changed or changed. Therefore, all such modifications are intended to be included in the scope of the present invention. The order or sequence of any process or method step can be changed or reordered according to the alternative embodiment. Any "device plus function" clause is intended to cover the structure of the execution function described in this article, and is not only structurally equivalent but also equivalent structure. Without departing from the scope of the present invention, other replacements, modifications, changes and omissions can be made in the design, operating conditions and arrangement of the exemplary embodiment. Therefore, the invention is not limited to a specific embodiment, but extends to numerous modifications still falling within the scope of the appended claims.

[0146] Additionally, in order to provide a concise description of exemplary embodiments, all features of an actual embodiment (ie, those features that are not relevant to the best mode presently contemplated for carrying out the invention or those that are not relevant to implementing the invention) may not be described.

[0147] It should be understood that in the development of any actual implementation, as in any engineering or design project, numerous implementation-specific decisions may be made. Such a development effort may be complex and time-consuming, but for those of ordinary skill having the benefit of this disclosure, the development effort will be a routine task of design, fabrication, and production without undue experimentation.

[0148] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A disease analysis method based on a multi-target sensor array, characterized in that: include: S1. deploying a biosensor array and immersing it in a biological sample to be tested to collect biomarker parameters in the biological sample to be tested; S2. preprocessing the collected biomarker parameters, and analyzing the preprocessed biomarker parameters through a biological parameter detection model; S3, using the disease monitoring model to analyze the biomarker parameters output by the biological parameter detection model to predict the disease; S4. Optimize biological parameter detection models and disease monitoring models through performance indicators, and set health warning thresholds based on personal signs; S5. Set security and privacy protection according to the disease privacy level and set data encryption; S6. Complete the disease prediction of biological samples and store them in the database. The biomarker parameters and predicted diseases are used as training sets for biological parameter detection models and disease monitoring models.

2. The disease analysis method based on a multi-target sensor array according to claim 1, characterized in that: The biosensor array includes a plurality of biosensors forming an array; The biosensor comprises: a biosensitive gate, an organic semiconductor channel, a drain electrode, a source electrode and a substrate; A fixed voltage V is applied between the source electrode and the drain electrode. D , a drain current I is applied between the source electrode and the drain electrode D , if the biosensitive gate is applied with a voltage V G , and does not change the fixed voltage V D , changing the drain current I D , then the ions in the electrolyte solution can reversibly enter the organic semiconductor channel and change the conductivity in the organic semiconductor channel; By combining the biosensors into an array and immersing them in a biological sample to be tested, corresponding biomarker parameters, including biomarker names and concentration parameters, can be collected.

3. The disease analysis method based on a multi-target sensor array according to claim 2, characterized in that: The preprocessing includes data cleaning, feature extraction and data standardization; The biomarker parameters collected by the biosensor array are detected and predicted by the bioparameter detection model. The calculation expression of the bioparameter detection model is as follows: G(Q)=W T *Q T-1 +γ T Among them, G(Q) is the predicted value of biomarkers output by the biological parameter detection model, W T is the weight matrix of the Tth layer, Q T-1 is the biomarker parameter output by the T-1th layer, γ T is the bias of the Tth layer.

4. The disease analysis method based on a multi-target sensor array according to claim 3, characterized in that: The training set of the disease monitoring model includes biomarkers detected and predicted by the biological parameter detection model, known healthy and diseased sample data, and disease parameters corresponding to the biomarkers in the database; The disease monitoring model is essentially a logistic regression, and the output of the logistic regression is a probability value between 0 and 1, indicating the possibility of illness. The calculation expression of the disease monitoring model is as follows: Where P(Y=1|X) is the probability of disease under a given biomarker parameter X, α0 is the probability intercept term, α1 is the regression coefficient of the first biomarker parameter, α2 is the regression coefficient of the second biomarker parameter, and α N is the regression coefficient of the Nth biomarker parameter, X1 is the first biomarker parameter, X2 is the second biomarker parameter, X N is the Nth biomarker parameter; The classifier is used to maximize the interval between the diseased and the non-disease, and the classifier is used to iterate the interval between the diseased and the non-disease affected by each biomarker parameter. The classifier calculation expression is as follows: Among them, f(s) is the interval between the classifier output affecting the diseased and non-disease, a z is the probability factor of disease caused by abnormality of the zth biomarker parameter, b is bias compensation, P z is the probability that the abnormality of the zth biomarker parameter leads to illness, K(s i ,s0) is the kernel function used to calculate s i The inner product of the s0 sample in the high-dimensional feature space, s0 is the standard normal biomarker parameter, s i is the i-th group of biomarker parameters, z is the type number of the biomarker parameter, m is the total number of biomarker parameter types, and sign() is a function that returns an integer variable.

5. The disease analysis method based on a multi-target sensor array according to claim 4, characterized in that: The performance indicators include cross entropy and correction factors; The cross entropy controls the steepness of the loss function by setting an adjustment factor, and the loss function is used to minimize the absolute difference between the predicted value and the actual value, and is weighted by an exponential function; To compensate for lost values ​​and correct predicted values, a correction factor is applied after the disease monitoring model prediction, which is calculated based on historical errors or deviations to reduce the prediction error.

6. The disease analysis method based on a multi-target sensor array according to claim 5, characterized in that: The calculation expression of the cross entropy is as follows: in, is the cross entropy difference, which is used to reflect the loss between the actual value monitored by the biological parameter detection model and the output predicted value. log is the logarithmic function, a is the adjustment factor, y is the actual value monitored by the biological parameter detection model, is the predicted value output by the biological parameter detection model, n is the actual value data group monitored by the biological parameter detection model output, and i is the label sequence number of the data group.

7. The disease analysis method based on a multi-target sensor array according to claim 6, characterized in that: The sensor array data collection loss suppression method process is as follows: S2001, establishing a temperature parameter gradient matrix of the time domain signal at the kth moment, and taking the standard temperature parameter matrix as a reference matrix; S2002, performing iterative calculation using the temperature parameter gradient matrix of the time domain signal at the kth moment and the standard temperature parameter matrix; S2003, calculating an overfitting factor through the temperature parameter gradient matrix of the time domain signal at the kth moment, the standard temperature parameter matrix, the temperature parameter gradient matrix rank conversion of the time domain signal at the kth moment, and the standard temperature parameter matrix rank conversion; S2004, correcting the overfitting part by the overfitting factor; S2005. Project the temperature parameter gradient matrix and the standard temperature parameter matrix at the kth moment.

8. The disease analysis method based on a multi-target sensor array according to claim 1, characterized in that: The process of disease analysis, diagnosis and prediction is as follows: S1001. Collect a data set containing biomarker parameters and corresponding disease states, process missing values ​​and abnormal values ​​of the biomarker parameters, and perform standardization or normalization; S1002, selecting the biomarker parameters that are most useful for predicting the disease, establishing a disease monitoring model, and calculating the probability of disease under given biomarker parameters by maximizing whether the disease occurs; S1003. If the calculated disease probability is greater than the preset probability critical value, the patient is likely to be ill and further examination is performed; if the calculated disease probability is less than the preset probability critical value, the patient is likely not to be ill and the biomarker parameters are output; S1004. Analyze the probability of the patient suffering from the disease by combining the patient's historical examination data; S1005. Complete the patient's superficial analysis and prediction of the disease.

9. The disease analysis method based on a multi-target sensor array according to claim 1, characterized in that: The database is used to store the prediction results together with the biomarker parameters in the database; Data reuse is used to use this data as part of the dataset for future model training to continuously improve the accuracy of the model.

Images