A paliperidone therapeutic effect prediction method and system based on plasma proteomics

Abstract

The present application relates to the field of medical informatics, and provide a paliperidone efficacy prediction method and system based on plasma proteomics, the method comprises the following steps: collecting the basic data of the target patient; selecting the protein features through the basic data to obtain the related proteins corresponding to the paliperidone efficacy; training the neural network based on the related proteins and the baseline data in the basic data to obtain the prediction model; predicting the paliperidone efficacy of the patient to be predicted through the prediction model to obtain the prediction result. The present application can predict the treatment response of schizophrenic patients to paliperidone, and significantly improve the objectivity of paliperidone use.

Description

Technical Field

[0001] This invention relates to the field of medical informatics technology, and in particular to a method and system for predicting the efficacy of paliperidone based on plasma proteomics. Background Technology

[0002] Schizophrenia (SCZ) is a severe mental disorder with complex causes and a high rate of disability, imposing a heavy economic and health burden on patients. Antipsychotics are the main treatment for schizophrenia, but their efficacy varies significantly among individuals. Approximately 30% of patients do not respond well to existing antipsychotics, necessitating repeated changes in medication in clinical practice to select the most suitable antipsychotic. Paliperidone, as a relatively new antipsychotic, still exhibits significant individual variability in efficacy, with approximately 40% of patients showing an unsatisfactory response.

[0003] Although there are currently gene testing kits for antipsychotic drugs on the market, these products mostly focus on genes related to drug metabolism, and their prediction of drug efficacy is limited and the accuracy is poor. In addition, there is still a lack of biomarkers for the efficacy of paliperidone, making it difficult to provide patients with personalized auxiliary decision-making before medication. Summary of the Invention

[0004] This invention provides a method and system for predicting the efficacy of paliperidone based on plasma proteomics, in order to overcome the shortcomings of the prior art.

[0005] This invention provides a method for predicting the efficacy of paliperidone based on plasma proteomics, comprising:

[0006] S1: Collect basic data of the target patient;

[0007] S2: Feature selection of proteins is performed using the aforementioned basic data to obtain relevant proteins corresponding to the therapeutic effects of paliperidone;

[0008] S3: Based on the relevant proteins and the baseline data in the basic data, train the neural network to obtain a prediction model;

[0009] S4: The predictive model is used to predict the efficacy of paliperidone for the patients to be predicted, and the prediction results are obtained.

[0010] According to the present invention, a method for predicting the efficacy of paliperidone based on plasma proteomics is provided, wherein the target patient in step S1 is a patient with schizophrenia, and the basic data includes: baseline information, paliperidone treatment records, and relative expression levels of plasma proteins.

[0011] According to the method for predicting the efficacy of paliperidone based on plasma proteomics provided by the present invention, step S2 further includes:

[0012] S21: Extract the target variable from the aforementioned basic data;

[0013] S22: Preprocess the target variable to obtain preprocessed features;

[0014] S23: The preprocessed features are input into the L1 regularized LASSO regression model to perform feature selection and obtain the relevant proteins corresponding to the efficacy of paliperidone.

[0015] According to the method for predicting the efficacy of paliperidone based on plasma proteomics provided by the present invention, in step S23, the expression of the objective function of the LASSO regression model is:

[0016]

[0017] in, These are the parameters of the LASSO regression model. This represents the feature vector index value corresponding to the input preprocessed features. This represents the total number of feature vectors corresponding to the input preprocessed features. For the first For the input preprocessed features, For the first The labels of the preprocessed features are the input. Let be the hypothesis function for Logistic regression. For L1 regularization hyperparameters, It is the sum of the absolute values ​​of the L1 norms of all parameter terms.

[0018] According to the method for predicting paliperidone efficacy based on plasma proteomics provided by the present invention, the relevant proteins in step S23 include:

[0019] Selenin P, iduxose-2-sulfatase, α-mannosidase 2x, alanine-tRNA-editing protein, phosphatidylinositol polysaccharide 1, collagen α1 chain, kallikrein inhibitor, leucine-rich repeat neuron protein 4, SPON2 protein, angiopoietin-related protein 6, protein kinase C-binding protein, protein 2 containing ankyrin repeats and BTB domain, and sialic acid-binding immunoglobulin-like lectin 14.

[0020] According to the method for predicting the efficacy of paliperidone based on plasma proteomics provided by the present invention, step S3 further includes:

[0021] S31: Input the relevant protein and the baseline data in the basic data, and perform a linear transformation and combination on the relevant protein and the baseline data in the basic data to obtain a linear output;

[0022] S32: Perform a nonlinear transformation on the linear output to obtain a probability output;

[0023] S33: Construct a loss function based on the probability output;

[0024] S34: Optimize the model parameters of the neural network according to the loss function to obtain the prediction model.

[0025] According to the method for predicting paliperidone efficacy based on plasma proteomics provided by the present invention, in step S34, when optimizing the model parameters of the neural network, the expression for updating the parameters is:

[0026]

[0027] in, For the first The model parameters before the update, For the first The updated model parameters, For learning rate, This is the loss function.

[0028] According to the method for predicting the efficacy of paliperidone based on plasma proteomics provided by the present invention, step S4 further includes:

[0029] S41: Collect the prediction data of the patients to be predicted;

[0030] S42: Standardize the data to be predicted to obtain standardized data to be predicted;

[0031] S43: Input the standardized data to be predicted into the prediction model to obtain the prediction result; the prediction result is the probability of the patient responding to paliperidone treatment.

[0032] According to the present invention, a method for predicting the efficacy of paliperidone based on plasma proteomics is provided, wherein the data to be predicted in step S41 includes: the baseline PANSS score of the patient to be predicted and the relative expression level of plasma proteins of the patient to be predicted.

[0033] This invention also provides a paliperidone efficacy prediction system based on plasma proteomics, comprising:

[0034] Data Acquisition Module: Used to collect basic data from the target patient;

[0035] Selection module: used to perform feature selection on proteins based on the basic data to obtain relevant proteins corresponding to the efficacy of paliperidone;

[0036] Training module: used to train the neural network based on the relevant proteins and baseline data in the basic data to obtain a prediction model;

[0037] The prediction module is configured to use the prediction model trained by the training module to predict the efficacy of paliperidone for patients to be predicted and obtain prediction results.

[0038] This invention provides a method and system for predicting the efficacy of paliperidone based on plasma proteomics. By integrating high-throughput mass spectrometry plasma proteomics data with clinical indicators, it achieves accurate prediction of paliperidone treatment response in schizophrenia patients. This invention exhibits excellent predictive performance and good generalization ability, effectively solving the problem of repeated medication changes caused by traditional empirical medication, and providing a scientific basis for individualized clinical treatment decisions.

[0039] This invention, based on the introduction of the LASSO feature selection algorithm, enables the model to efficiently screen biomarkers closely related to the efficacy of paliperidone in a high-dimensional feature space containing thousands of plasma proteins. This not only significantly improves the efficiency of variable selection and the interpretability of the model, but also effectively reduces the interference of multicollinearity on the prediction results, ensuring the stability and reliability of the model. Secondly, the automatic feature selection characteristic of the L1 regularization term enables the model to automatically identify the protein biomarkers with the most predictive value, avoiding the bias of human subjective judgment and preventing the occurrence of model overfitting. In addition, the Logistic regression model combined with cross-validation and grid search parameter optimization strategies ensures the generalization performance of the model on different datasets, and provides clinicians with intuitive treatment response prediction results in the form of probability output.

[0040] The application of plasma proteomics data provides rich biological information for predicting the efficacy of drugs for schizophrenia. Compared with traditional drug metabolism gene-based detection methods, this invention can more comprehensively reflect the patient's biological state and drug action mechanism, significantly improving prediction accuracy. Practical application of the model can rapidly identify high-risk patients with poor paliperidone response in the early stages of treatment, avoiding symptom deterioration and waste of medical resources during ineffective treatment. It also provides scientific guidance for adjusting subsequent treatment plans and individualized management, ultimately improving patient prognosis and overall clinical benefits. The predictive model of this invention breaks through the limitations of the trial-and-error model in schizophrenia treatment, laying a solid foundation for achieving precision medicine and individualized treatment, and has significant clinical application value and social benefits. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0042] Figure 1 This is a schematic diagram of a method for predicting the efficacy of paliperidone based on plasma proteomics, provided in an embodiment of the present invention.

[0043] Figure 2 A schematic diagram of a paliperidone efficacy prediction system based on plasma proteomics provided in an embodiment of the present invention;

[0044] Figure 3 A schematic diagram of the ROC curve of the paliperidone efficacy prediction model provided in this embodiment of the invention. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, embodiments of this invention, and should not be construed as limiting the invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention. In the description of this invention, it should be understood that the terminology used is for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0046] The embodiments of the present invention are described below with reference to the figures.

[0047] like Figure 1 As shown, this invention provides a method for predicting the efficacy of paliperidone based on plasma proteomics, comprising:

[0048] S1: Collect basic data of the target patient.

[0049] Furthermore, the target patients were those with schizophrenia, and the screening criteria strictly followed the DSM-IV diagnostic criteria for diagnosis. Specific screening criteria included: age range 18-45 years, first-episode, medication-naïve patients or patients with a chronic disease course currently experiencing an acute exacerbation of schizophrenia, and a baseline PANSS (Positive and Negative Syndrome Scale) total score greater than or equal to 60. Exclusion criteria included: pregnant or lactating women or women planning pregnancy, patients with contraindications to paliperidone treatment, patients with unstable physical illness, prolonged QTc interval, decompensated congestive heart failure, or complete left bundle branch block. Based on these screening criteria, a total of 208 patients were ultimately included, comprising 188 identified schizophrenia patients and 20 independently validated schizophrenia patients.

[0050] In step S1, the target patient is a patient with schizophrenia, and the basic data includes: baseline information, paliperidone treatment records, and relative expression levels of plasma proteins.

[0051] Baseline data collection encompassed the patient's demographic characteristics and clinical assessment indicators. Demographic characteristics primarily included patient age, recorded in years as a continuous variable. Clinical assessment indicators included the baseline PANSS score, which comprises three parts: a positive symptom scale, a negative symptom scale, and a general psychopathology scale. The total score ranges from 30 to 210 points, with higher scores indicating more severe symptoms. During data collection, trained physicians from the relevant department conducted assessments before the patient began paliperidone treatment to ensure standardization and accuracy. Baseline data was stored numerically; age data was recorded directly as integers or decimals, and the baseline PANSS score was recorded as an integer.

[0052] Data collection for paliperidone treatment included pre- and post-treatment clinical assessments and efficacy determination. Patients underwent a repeat PANSS assessment at the end of the sixth week after receiving paliperidone treatment, obtaining their six-week PANSS score. Efficacy determination was based on the PANSS score reduction rate at the end of the sixth week, calculated as: PANSS reduction rate = (baseline PANSS score - six-week PANSS score) / (baseline PANSS score - 30) × 100%. Based on the PANSS reduction rate, patients were divided into two treatment response groups: patients with a PANSS reduction rate greater than or equal to 50% were defined as having a good response to paliperidone treatment and were labeled 1; patients with a PANSS reduction rate less than 50% were defined as having a poor response to paliperidone treatment and were labeled 0. This binary labeling method provided clear labeled data for subsequent supervised learning of the machine learning model.

[0053] Plasma protein relative expression levels were obtained using high-throughput mass spectrometry, an analytical method based on the separation and detection of ions by their mass-to-charge ratio. This method can accurately quantify the relative abundance of proteins in plasma. Plasma protein relative expression levels reflect the relative abundance of proteins in plasma and are expressed as unitless relative values.

[0054] Specifically, the relative expression levels of plasma proteins were obtained by collecting peripheral blood samples from patients at baseline. Blood samples were collected using EDTA anticoagulant tubes, and after centrifugation to separate the plasma, proteins were extracted and analyzed by mass spectrometry. The mass spectrometry analysis covered 1466 plasma proteins, and the relative expression level of each protein was calculated using peak area integral values. After quality control and standardization, the final protein expression profile data was generated.

[0055] After data collection, a data matrix containing 1467 variables was established, including 1466 variables representing the relative expression levels of plasma proteins and 1 variable representing the baseline PANSS score. Each patient sample corresponds to one row of data, and each variable corresponds to one column of data, forming a complete data matrix of 208×1467.

[0056] S2: Feature selection of proteins is performed using the aforementioned basic data to obtain the relevant proteins corresponding to the therapeutic effects of paliperidone.

[0057] Step S2 further includes:

[0058] S21: Extract the target variable from the aforementioned basic data.

[0059] Furthermore, the target variable comprises two parts: feature variables and label variables. Feature variables are the independent variables used to predict the model input, containing the relative expression levels of 1466 plasma proteins and the baseline PANSS score. The relative expression level data of plasma proteins are stored in the form of a numerical matrix with dimensions of 188×1466, where 188 represents the number of patients found in the sample, and 1466 represents the number of plasma protein types detected. Each matrix element represents the relative expression level of a specific protein in a specific patient, and the specific value depends on the peak area integral value of the mass spectrometry detection. The baseline PANSS score, as a separate feature variable, is stored as a one-dimensional array with a length of 188, and each element corresponds to a patient's total baseline PANSS score. The label variable refers to the target output predicted by the model, namely, the patient's treatment response to paliperidone. Label variables are determined by calculating the PANSS score reduction rate for each patient. This calculated PANSS score reduction rate is compared to a 50% threshold. When the PANSS score reduction rate is ≥50%, the label variable is assigned a value of 1, indicating a good treatment response; when the PANSS score reduction rate is <50%, the label variable is assigned a value of 0, indicating a poor treatment response. Label variables are stored as a one-dimensional binary array with a length of 188, and each element has a value of either 0 or 1.

[0060] During data extraction, the raw data file of relative plasma protein expression levels was first read, and an index mapping relationship was established according to protein name and patient number to ensure data accuracy and completeness. Then, baseline PANSS score data was read and matched and merged with the protein data according to patient number. Next, the PANSS score data at the end of the sixth week was read, and the PANSS reduction rate for each patient was calculated. Binary label variables were generated based on a 50% threshold. Finally, a complete dataset containing feature variables and label variables was formed; the feature variable matrix has a dimension of 188×1467, and the label variable array has a length of 188.

[0061] S22: Preprocess the target variable to obtain preprocessed features.

[0062] Further, the preprocessing of the extracted target variables includes two main steps: missing value handling and feature standardization. Missing value handling involves imputing missing values ​​in the relative expression levels of plasma proteins, specifically using median imputation, where the median value of a specific variable is replaced with the median value of that variable across all non-missing samples. Feature standardization involves standardizing the plasma protein relative expression levels after missing value imputation. The purpose of standardization is to eliminate the influence of dimensional differences between different proteins, using the Z-score standardization method. The baseline PANSS score is not standardized and retains its original score format. After preprocessing, a preprocessed feature matrix containing 1467 features is formed, with a matrix dimension of 188×1467.

[0063] S23: The preprocessed features are input into the L1 regularized LASSO regression model to perform feature selection and obtain the relevant proteins corresponding to the efficacy of paliperidone.

[0064] In step S23, the expression for the objective function of the LASSO regression model is:

[0065]

[0066] in, These are the parameters of the LASSO regression model. This represents the feature vector index value corresponding to the input preprocessed features. This represents the total number of feature vectors corresponding to the input preprocessed features. For the first For the input preprocessed features, For the first The labels of the preprocessed features are the input. Let be the hypothesis function for Logistic regression. For L1 regularization hyperparameters, It is the sum of the absolute values ​​of the L1 norms of all parameter terms.

[0067] In step S23, the preprocessed features are input into the LASSO regression model for feature selection. The LASSO regression model of this invention is a machine learning algorithm that introduces an L1 regularization term based on traditional Logistic regression. The L1 regularization term is a penalty term applied to the sum of the absolute values ​​of the model parameters. The objective function of the LASSO regression model is shown above. The feature selection mechanism of the LASSO regression model is based on the sparsity-induced properties of the L1 regularization term. When the regularization coefficient λ is sufficiently large, some regression coefficients will shrink to 0, and the corresponding feature variables are automatically excluded, thus achieving feature selection. The final selected features include the aforementioned 13 paliperidone efficacy-related proteins. After feature selection, a final feature set containing 13 protein features and one baseline PANSS score feature is formed.

[0068] The relevant proteins mentioned in step S23 include:

[0069] Selenin P, iduxose-2-sulfatase, α-mannosidase 2x, alanine-tRNA-editing protein, phosphatidylinositol polysaccharide 1, collagen α1 chain, kallikrein inhibitor, leucine-rich repeat neuron protein 4, SPON2 protein, angiopoietin-related protein 6, protein kinase C-binding protein, protein 2 containing ankyrin repeats and BTB domain, and sialic acid-binding immunoglobulin-like lectin 14.

[0070] Specifically, the proteins obtained through screening include: selenoprotein P, iduronate 2-sulfatase, alpha-mannosidase 2x, alanyl-tRNA editing protein Aarsd1, phosphatidylinositol glycan 1, collagen alpha-1, kallistatin, leucine-rich repeat neuronal protein 4, Spondin-2, angiopoietin-related protein 6, protein kinase C-binding protein, and ankyrin repeat and BTB / POZ domain-containing protein. 2) Sialic acid-binding Ig-like lectin 14, a total of 13 protein indicators.

[0071] S3: Based on the relevant proteins and baseline data in the basic data, train the neural network to obtain a prediction model.

[0072] Preferably, the neural network is a Logistic regression prediction model.

[0073] Step S3 further includes:

[0074] S31: Input the relevant protein and the baseline data in the basic data, and perform a linear transformation and combination on the relevant protein and the baseline data in the basic data to obtain a linear output.

[0075] Furthermore, the input data includes 13 paliperidone efficacy-related protein features obtained through LASSO feature selection and the baseline PANSS score. During the linear transformation combination process, the 14 input features (13 protein features plus 1 baseline PANSS score) are combined to form a feature vector, and the linear output is calculated through matrix operations. During initialization, the weight parameters are usually set to small random values, and the bias term is set to 0.

[0076] S32: Perform a nonlinear transformation on the linear output to obtain a probability output.

[0077] In step S32, the present invention performs a nonlinear transformation on the linear output obtained in step S31, using the Sigmoid function to map it to the (0,1) interval to achieve a probability output. The output value obtained after the Sigmoid function transformation represents the probability that the patient sample belongs to the category of good response to paliperidone treatment. When the probability value is greater than 0.5, it predicts that the patient has a good response to paliperidone treatment; when the probability value is less than or equal to 0.5, it predicts that the patient has a poor response to paliperidone treatment.

[0078] S33: Construct a loss function based on the probability output.

[0079] Furthermore, in step S33, based on the probability output obtained in step S32, this invention constructs a log-likelihood loss function (cross-entropy loss function) to measure the error between the model's predicted probability and the true label. Specifically, first, the home function is defined, and combined with the aforementioned nonlinear mapping, the hypothesis function of the model is defined. This is used to describe the mapping relationship between input features and output class probabilities. Then, a loss function is defined based on the hypothesis function. The specific expression of the loss function is as follows:

[0080]

[0081] in, For the defined loss function, For model parameters, The input sample index value, The total number of samples input. For the input of the first One sample, For the first The true label of each sample.

[0082] S34: Optimize the model parameters of the neural network according to the loss function to obtain the prediction model.

[0083] In step S34, the model parameters are optimized using gradient descent to minimize the loss function. In each iteration, the partial derivative of the loss function with respect to each parameter is calculated, and the parameters are updated accordingly. The specific expression is shown below.

[0084] In step S34, when optimizing the model parameters of the neural network, the expression for updating the parameters is:

[0085]

[0086] in, For the first The model parameters before the update, For the first The updated model parameters, For learning rate, This is the loss function.

[0087] Furthermore, the learning rate is used to control the magnitude of each parameter update. The above iterative process continues until the loss function converges to a predetermined threshold or reaches the maximum number of iterations. After the loss function converges, the final trained model parameters are output. and bias At this point, the obtained model is able to perform probability prediction and binary classification on new input samples.

[0088] S4: The predictive model is used to predict the efficacy of paliperidone for the patients to be predicted, and the prediction results are obtained.

[0089] Step S4 further includes:

[0090] S41: Collect the data to be predicted for the patients to be predicted.

[0091] The data to be predicted in step S41 includes: the baseline PANSS score of the patient to be predicted and the relative expression level of plasma proteins of the patient to be predicted.

[0092] The data to be predicted includes two core indicators: the baseline PANSS score of the patients to be predicted and the relative expression levels of plasma proteins. The baseline PANSS score refers to the positive and negative symptom scale scores, which contains 30 items: 7 items for positive symptoms, 7 items for negative symptoms, and 16 items for general psychopathology. Each item is assessed using a 7-point scale (1-7), and the baseline PANSS total score must be greater than or equal to 60. The relative expression level of plasma proteins was collected using non-targeted mass spectrometry to detect the relative expression levels of plasma proteins. Plasma samples were collected from patients in a fasting state, with 5-10 ml of venous blood drawn using EDTA anticoagulant tubes. After standing at room temperature for 30 minutes, the plasma was separated by centrifugation at 3000 rpm for 10 minutes. The plasma samples were frozen at -80°C until detection. The mass spectrometry detection process used liquid chromatography-tandem mass spectrometry. Proteins were digested with trypsin, followed by peptide separation and mass spectrometry detection to obtain the relative expression levels of each protein.

[0093] S42: Standardize the data to be predicted to obtain standardized data to be predicted.

[0094] The standardization process employs Z-score standardization, using the mean and standard deviation calculated from the training set data to standardize the data to be predicted. For plasma protein features, the mean and standard deviation of each protein feature are first calculated from the training set, and then standardized. During standardization, the mean of all feature values ​​is ensured to be 0, and the standard deviation to be 1, eliminating dimensional and numerical range differences between different protein features. The baseline PANSS score is processed using the same standardization method, using the mean and standard deviation of the PANSS scores from the training set to standardize the PANSS scores of the patients to be predicted. The standardized data to be predicted constitute a feature vector, which contains all standardized feature values ​​(protein features + PANSS scores).

[0095] S43: Input the standardized data to be predicted into the prediction model to obtain the prediction result; the prediction result is the probability of the patient responding to paliperidone treatment.

[0096] Furthermore, during the prediction process, this invention inputs standardized data to be predicted into a trained Logistic regression prediction model for forward propagation calculation. First, feature values ​​corresponding to 13 efficacy-related proteins selected during training are extracted. These 13 protein feature values ​​are combined with the standardized baseline PANSS score to form a 14-dimensional feature vector, which is then input into the prediction model. The model ultimately outputs a probability value, representing the probability that the patient will respond well to paliperidone treatment, ranging from 0 to 1. A probability value greater than 0.5 predicts a good treatment response, while a probability value less than 0.5 predicts a poor treatment response. The prediction result output includes specific probability values ​​and corresponding binary classification results, providing clinicians with a quantitative reference for paliperidone medication decisions.

[0097] like Figure 2 As shown, the present invention also provides a paliperidone efficacy prediction system based on plasma proteomics, comprising:

[0098] Data Acquisition Module 100: Used to collect basic data of the target patient;

[0099] Selection module 200: used to perform feature selection on proteins based on the basic data to obtain relevant proteins corresponding to the efficacy of paliperidone;

[0100] Training module 300: used to train the neural network based on the relevant proteins and baseline data in the basic data to obtain a prediction model;

[0101] The prediction module 400 is configured to use the prediction model trained by the training module 300 to predict the efficacy of paliperidone for the patient to be predicted and obtain the prediction result.

[0102] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0103] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0104] This invention provides a method and system for predicting the efficacy of paliperidone based on plasma proteomics. Based on plasma proteomics data obtained by high-throughput mass spectrometry, it acquires rich biological state information of schizophrenia patients and organically integrates it with the patients' clinical indicators, significantly improving the objectivity and information content of the data.

[0105] The LASSO-based feature selection method of this invention can efficiently screen biomarkers closely related to drug efficacy in a high-dimensional protein feature space, improving the efficiency of variable selection and model interpretability, and reducing the impact of multicollinearity on model results. The logistic regression prediction model can more efficiently and flexibly predict the probability of paliperidone treatment response. In the model construction process, combined with clinical indicators such as baseline PANSS scores, logistic regression is used and cross-validation and grid search are introduced to optimize parameters, ensuring the model's generalization ability and stability on the training set and independent validation set.

[0106] The probability of response to paliperidone treatment can be predicted using this invention, such as... Figure 3As shown, the prediction model constructed by the method of the present invention can achieve an average AUC of 0.855 on the discovery samples and an AUC of 0.813 on the independent validation samples, demonstrating good prediction performance and generalization.

[0107] This invention is applicable to the rapid identification of high-risk individuals with poor paliperidone response in the early stages of treatment. After the data is entered into the doctor's computer to predict the paliperidone treatment response of schizophrenia patients, it assists in the decision-making of whether to give the patient a subsequent paliperidone treatment plan, providing a scientific basis for subsequent treatment adjustment and individualized management, and helping to improve patient prognosis and overall medical benefits.

[0108] This invention is applicable to the rapid identification of high-risk individuals with poor paliperidone response in the early stages of treatment. By inputting relevant data into a physician's computer, it predicts the treatment response of schizophrenia patients to paliperidone, assists in the decision-making process regarding paliperidone treatment, and provides a scientific basis for subsequent treatment adjustments and individualized management, ultimately contributing to improved patient prognosis and overall clinical benefits.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting the efficacy of paliperidone based on plasma proteomics, characterized in that, include: S1: Collect basic data of the target patient; The target patients are patients with schizophrenia, and the basic data include: baseline information, paliperidone treatment records, and relative expression levels of plasma proteins. S2: Feature selection of proteins is performed using the aforementioned basic data to obtain relevant proteins corresponding to the therapeutic effects of paliperidone; the relevant proteins include: selenoprotein P, iduro-2-sulfatase, α-mannosidase 2x, alanine-tRNA-editing protein, phosphatidylinositol polysaccharide 1, collagen α1 chain, kallikrein inhibitor, leucine-rich repeat neuron protein 4, SPON2 protein, angiopoietin-related protein 6, protein kinase C-binding protein, protein 2 containing ankyrin repeats and BTB domains, and sialic acid-binding immunoglobulin-like lectin 14. S3: Based on the relevant proteins and the baseline data in the basic data, train the neural network to obtain a prediction model; S4: The prediction model is used to predict the efficacy of paliperidone in the patients to be predicted, and the prediction results are obtained. Step S4 further includes: S41: collecting the data to be predicted from the patient to be predicted; S42: standardizing the data to be predicted to obtain standardized data to be predicted; S43: inputting the standardized data to be predicted into the prediction model to obtain the prediction result; the prediction result is the probability of the patient to be predicted responding to paliperidone treatment. The data to be predicted in step S41 includes: the baseline PANSS score of the patient to be predicted and the relative expression level of plasma proteins of the patient to be predicted.

2. The method for predicting paliperidone efficacy based on plasma proteomics according to claim 1, characterized in that, Step S2 further includes: S21: Extract the target variable from the aforementioned basic data; S22: Preprocess the target variable to obtain preprocessed features; S23: The preprocessed features are input into the L1 regularized LASSO regression model to perform feature selection and obtain the relevant proteins corresponding to the efficacy of paliperidone.

3. The method for predicting paliperidone efficacy based on plasma proteomics according to claim 2, characterized in that, In step S23, the expression for the objective function of the LASSO regression model is: in, These are the parameters of the LASSO regression model. This represents the feature vector index value corresponding to the input preprocessed features. This represents the total number of feature vectors corresponding to the input preprocessed features. For the first For the input preprocessed features, For the first The labels of the preprocessed features are the input. Let be the hypothesis function for Logistic regression. For L1 regularization hyperparameters, It is the sum of the absolute values ​​of the L1 norms of all parameter terms.

4. The method for predicting paliperidone efficacy based on plasma proteomics according to claim 1, characterized in that, Step S3 further includes: S31: Input the relevant protein and the baseline data in the basic data, and perform a linear transformation and combination on the relevant protein and the baseline data in the basic data to obtain a linear output; S32: Perform a nonlinear transformation on the linear output to obtain a probability output; S33: Construct a loss function based on the probability output; S34: Optimize the model parameters of the neural network according to the loss function to obtain the prediction model.

5. The method for predicting paliperidone efficacy based on plasma proteomics according to claim 4, characterized in that, In step S34, when optimizing the model parameters of the neural network, the expression for updating the parameters is: in, For the first The model parameters before the update, For the first The updated model parameters, For learning rate, This is the loss function.

6. A paliperidone efficacy prediction system based on plasma proteomics, characterized in that, include: Data Acquisition Module: Used to collect basic data from the target patient; The target patients are patients with schizophrenia, and the basic data include: baseline information, paliperidone treatment records, and relative expression levels of plasma proteins. Selection module: used to perform feature selection on proteins based on the basic data to obtain relevant proteins corresponding to the efficacy of paliperidone; the relevant proteins include: selenoprotein P, iduro-2-sulfatase, α-mannosidase 2x, alanine-tRNA-editing protein, phosphatidylinositol polysaccharide 1, collagen α1 chain, kallikrein inhibitor, leucine-rich repeat neuron protein 4, SPON2 protein, angiopoietin-related protein 6, protein kinase C-binding protein, protein 2 containing ankyrin repeats and BTB domains, and sialic acid-binding immunoglobulin-like lectin 14; Training module: used to train the neural network based on the relevant proteins and baseline data in the basic data to obtain a prediction model; A prediction module is configured to use the prediction model trained by the training module to predict the efficacy of paliperidone for the patient to be predicted and obtain the prediction result. The prediction module is further used for: collecting the prediction data of the patient to be predicted; standardizing the prediction data to obtain standardized prediction data; inputting the standardized prediction data into the prediction model to obtain the prediction result; the prediction result is the probability of the patient to be predicted responding to paliperidone treatment; wherein, the prediction data includes: the baseline PANSS scale score of the patient to be predicted and the relative expression level of plasma proteins of the patient to be predicted.

Images