Prediction method and device for indications of traditional Chinese medicine prescription

Abstract

The invention relates to the technical field of computers, in particular to a traditional Chinese medicine prescription indication prediction method, which comprises the steps of collecting a data set of a relationship between an existing traditional Chinese medicine prescription and a corresponding indication, preprocessing the data set, and training a corresponding prediction model based on the preprocessed data set; obtaining to-be-predicted new traditional Chinese medicine prescription data, inputting the preprocessed new traditional Chinese medicine prescription data into the trained prediction model, obtaining a corresponding prediction result, and generating a corresponding traditional Chinese medicine prescription treatment scheme based on the prediction result; the method comprises the following steps: in a process of treating a patient with indications based on a traditional Chinese medicine prescription treatment scheme, obtaining latest biomics data of the corresponding patient, monitoring the patient to obtain a corresponding treatment effect, and adjusting the traditional Chinese medicine prescription treatment scheme based on the latest biomics data and the treatment effect. According to the scheme, the prediction precision of the indications of the traditional Chinese medicine prescription is improved.

Description

Technical Field

[0001] The present application relates to the technical field of computers, and in particular to a method for predicting indications of traditional Chinese medicine prescriptions. Background Art

[0002] In recent years, with the development of bio-omics technology, more and more biological information has been recorded, and these data provide rich resources for studying the mechanism of action of Chinese medicine prescriptions; traditionally, the application of Chinese medicine prescriptions mainly relies on the experience and clinical observation of Chinese medicine practitioners, and lacks scientific quantitative analysis methods.

[0003] With the advancement of artificial intelligence technology, especially the development of deep learning algorithms, people have begun to try to apply these technologies to the research of traditional Chinese medicine prescriptions to achieve more accurate predictions of drug efficacy. However, due to the limited understanding of the complex mechanism of action of traditional Chinese medicine prescriptions, the prediction accuracy of traditional Chinese medicine prescription indications is not very high, and this situation needs to be further improved. Summary of the invention

[0004] In order to improve the prediction accuracy of TCM prescription indications, the present application provides a TCM prescription indication prediction method and device.

[0005] In the first aspect, the present application provides a method for predicting the indications of a traditional Chinese medicine prescription, which adopts the following technical solution:

[0006] A method for predicting the indications of a Chinese medicine prescription comprises: collecting a data set of the relationship between an existing Chinese medicine prescription and a corresponding indication, preprocessing the data set, and training a corresponding prediction model based on the preprocessed data set; obtaining new Chinese medicine prescription data to be predicted, inputting the preprocessed new Chinese medicine prescription data into the trained prediction model, obtaining a corresponding prediction result, and generating a corresponding Chinese medicine prescription treatment plan based on the prediction result, wherein the prediction model is used to predict the indications of the Chinese medicine prescription for a specific disease; in the process of treating a patient with an indication based on the Chinese medicine prescription treatment plan, obtaining the latest bio-omics data of the corresponding patient, monitoring the patient to obtain the corresponding treatment effect, and adjusting the Chinese medicine prescription treatment plan based on the latest bio-omics data and the treatment effect, wherein the bio-omics data includes genomics, transcriptomics, and proteomics.

[0007] By adopting the above technical solutions, by building and using predictive models to analyze the association between Chinese medicine prescriptions and specific diseases, it is not only possible to provide patients with personalized treatment recommendations, but also to guide Chinese medicine developers to design new Chinese medicine prescriptions based on the latest bio-omics data and actual efficacy feedback, thereby significantly accelerating the research and development process of Chinese medicine and improving the efficiency and accuracy of new drug discovery. This method combines traditional Chinese medicine theory with modern bioinformatics technology, aiming to promote the modernization of Chinese medicine and promote the birth of safer and more effective new drugs.

[0008] Optionally, in the process of obtaining the corresponding prediction results, the method also includes: based on the prediction results, designing a laboratory test plan for the new Chinese medicine prescription to evaluate the actual performance of the new Chinese medicine prescription, the laboratory test plan includes pharmacology testing, toxicology testing, and bioavailability testing; performing the experiment corresponding to the laboratory test plan, and obtaining corresponding experimental data during the experiment, wherein the experimental data includes detailed experimental process and observation results; comparing the experimental data with the prediction results of the prediction model to obtain the corresponding experimental comparison results, and optimizing the prediction model according to the experimental comparison results. If the error between the experimental data and the prediction results of the prediction model reaches an acceptable range, guiding the development of the new Chinese medicine prescription based on the prediction results.

[0009] By adopting the above technical solutions, the prediction model is trained by collecting and analyzing the data between Chinese medicine prescriptions and indications, and the laboratory test scheme (such as pharmacology, toxicology and bioavailability tests) is designed based on the model prediction results to evaluate the actual performance of the new Chinese medicine prescription. By comparing and analyzing the experimental data with the model prediction results, it is not only possible to continuously optimize the accuracy of the prediction model, but also to provide scientific basis for Chinese medicine researchers and developers, guiding them to design new Chinese medicine prescriptions more effectively, thereby accelerating the research and development process of Chinese medicine and improving the efficiency and success rate of new drug discovery. This method combines computational prediction with experimental evidence, aiming to promote innovation and development in the process of modernization of Chinese medicine.

[0010] Optionally, in the process of training the prediction model, the method also includes: obtaining a set of biomics data, associating each biomics data with a corresponding label, wherein the biomics data includes but is not limited to genomics, transcriptomics, and proteomics, and the labels include valid, invalid, and partially valid; extracting feature data based on the biomics data set, and dividing the feature data corresponding to the biomics data set into training set data and test set data; selecting a deep neural network as a model, training the model using the training set data, and adjusting the model parameters to minimize the loss function, using the test set data to verify the generalization ability of the model, and determining the hyperparameter combination of the optimal model to determine the corresponding prediction model.

[0011] By adopting the above technical solutions, it is ensured that the predictive model can learn effective features from a large amount of bio-omics data, and after proper training and verification, it can accurately predict the indications of Chinese medicine prescriptions for specific diseases, thereby guiding the design of more accurate treatment plans.

[0012] Optionally, in the process of extracting feature data based on a bio-omics data set, the method also includes: classifying all extracted feature data, and using statistical tests to calculate type importance data for each type of feature data; sorting all types of feature data based on the type importance data, and selecting feature types with top rankings; and determining the selected feature types as training feature types and dividing them into training set data and test set data.

[0013] By adopting the above technical solutions, it can be ensured that the prediction model can not only accurately predict the indications of Chinese herbal medicines for specific diseases, but also improve the efficiency and interpretability of the model by optimizing feature selection, thereby providing a more scientific basis for personalized treatment.

[0014] Optionally, after generating the corresponding Chinese medicine prescription treatment plan, the method also includes: using an AR development tool to identify the patient's imaging data, and locating the patient's key organ positions and tissue positions based on the imaging data; superimposing virtual information on the patient's body parts based on the key organ positions and tissue positions; using AR technology to display the change data of Chinese medicine ingredients in the body over time, and generating corresponding real-time feedback based on the change data, wherein the change data includes color changes and dynamic paths.

[0015] By adopting the above technical solutions, the effectiveness and accuracy of traditional Chinese medicine treatment plans can be improved, and the impact of biomics data on individual health status can be better understood and interpreted. Through the combination of AR technology and biomics data analysis, more personalized medical services can be achieved.

[0016] Optionally, the method also includes: acquiring and collecting the patient's genomic data through a gene chip, and preprocessing the genomic data, wherein the genomic data includes whole genome sequencing, whole exome sequencing, and target region sequencing, and the preprocessing includes quality control, variation detection, and annotation; using a machine learning algorithm to construct an evaluation model, the evaluation model is trained based on the patient's genetic variation information, clinical data, and demographic data as input features, and the evaluation model is trained and verified by a cross-validation method; the genomic data is input into the evaluation model to predict the results and generate a genetic risk assessment report, and the patient's treatment plan is adjusted based on the genetic risk assessment report, wherein the genetic risk assessment report contains the patient's genetic susceptibility to specific drugs, potential adverse reaction risks, and corresponding recommendations.

[0017] By adopting the above technical solutions, it is possible to provide precise medical advice based on individual genetic information, thereby optimizing treatment strategies, reducing unnecessary drug use, and improving treatment success rates.

[0018] Optionally, the method also includes: periodically triggering a retraining process of the prediction model, during which advanced optimization algorithms are used to adjust the parameters of the prediction model; after each prediction model update, an independent validation set is used to evaluate the performance of the updated prediction model, and if the updated prediction model performs better than the prediction model before the update, the updated prediction model is adopted; otherwise, the system will continue to use the prediction model before the update.

[0019] By adopting the above technical solution, the retraining and evaluation of the prediction model is periodically triggered, which not only ensures the latest status of the model, but also improves the reliability and practicality of the model, thereby improving the performance of the overall system and user experience.

[0020] In a second aspect, the present application provides a device for predicting indications of a traditional Chinese medicine prescription, comprising:

[0021] A data set collection module is used to collect data sets of the relationship between existing Chinese medicine prescriptions and corresponding indications, preprocess the data sets, and train the corresponding prediction model based on the preprocessed data sets;

[0022] A prediction result acquisition module is used to acquire new Chinese medicine prescription data to be predicted, input the pre-processed new Chinese medicine prescription data into the trained prediction model to obtain the corresponding prediction results, and generate the corresponding Chinese medicine prescription treatment plan based on the prediction results, wherein the prediction model is used to predict the indications of the Chinese medicine prescription for a specific disease;

[0023] The Chinese medicine prescription treatment plan adjustment module obtains the latest biomics data of the corresponding patients during the treatment of patients with indications based on the Chinese medicine prescription treatment plan, monitors the patients to obtain the corresponding treatment effects, and adjusts the Chinese medicine prescription treatment plan based on the latest biomics data and treatment effects, among which the biomics data includes genomics, transcriptomics, and proteomics.

[0024] In a third aspect, the present application provides a method for predicting the indications of a traditional Chinese medicine prescription, using the following technical solution:

[0025] A method for predicting indications of a traditional Chinese medicine prescription comprises a processor in which a program of any one of the above-mentioned methods for predicting indications of a traditional Chinese medicine prescription is run.

[0026] In a fourth aspect, the present application provides a storage medium, which adopts the following technical solution:

[0027] A storage medium stores a program for predicting the indications of a traditional Chinese medicine prescription according to any one of the above.

[0028] In summary, the present application includes at least one of the following beneficial technical effects: by constructing and using a predictive model to analyze the association between Chinese medicine prescriptions and specific diseases, it can not only provide patients with personalized treatment recommendations, but also guide Chinese medicine developers to design new Chinese medicine prescriptions based on the latest bio-omics data and actual efficacy feedback, thereby significantly accelerating the research and development process of Chinese medicine and improving the efficiency and accuracy of new drug discovery. This method combines traditional Chinese medicine theory with modern bioinformatics technology, aiming to promote the modernization of Chinese medicine and promote the birth of safer and more effective new drugs. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is a flow chart of a method for predicting indications of a traditional Chinese medicine prescription according to an exemplary embodiment;

[0030] Figure 2 It is a structural block diagram of a device for predicting indications of a traditional Chinese medicine prescription according to an exemplary embodiment. DETAILED DESCRIPTION

[0031] Embodiments of the present application are described in detail below, examples of which are shown in the accompanying drawings.

[0032] In the description of this specification, the description with reference to the terms "certain embodiments", "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiments or examples are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0033] The present application embodiment discloses a method for predicting the indications of a traditional Chinese medicine prescription, referring to Figure 1 ,include:

[0034] S100, collecting a data set of the relationship between existing Chinese medicine prescriptions and corresponding indications, preprocessing the data set, and training a corresponding prediction model based on the preprocessed data set.

[0035] Among them, we collect existing literature, clinical trial reports, historical cases, etc. to obtain information about different Chinese medicine prescriptions and their reported or proven effective disease indications. These data may come from traditional medical books, modern scientific research results, electronic health records, etc., which can provide basic data support for subsequent model training.

[0036] In addition, preprocessing work includes but is not limited to cleaning (removing duplicates, correcting errors), standardization (unifying measurement units, normalizing terminology), feature engineering (extracting key variables such as drug ingredients, dosage, administration method, etc.) and data segmentation (dividing training sets and test sets).

[0037] By using a trained prediction model, which is trained based on a large number of known data sets, and taking the pre-processed bio-omics data as input, the prediction model is used to obtain the prediction results about which Chinese medicine prescription is most suitable for this patient.

[0038] S110, obtaining new Chinese medicine prescription data to be predicted, inputting the preprocessed new Chinese medicine prescription data into the trained prediction model, obtaining corresponding prediction results, and generating corresponding Chinese medicine prescription treatment plans based on the prediction results, wherein the prediction model is used to predict the indications of Chinese medicine prescriptions for specific diseases.

[0039] When a new Chinese medicine formula is available, the relevant information must first be organized in a previously set standard format, and then input into the trained prediction system. It should be noted here that when generating prediction results, the inner product between the drug node vector and the disease node vector can be calculated, and then after a certain conversion operation, the association probability between the drug (Chinese medicine or Chinese medicine formula) and the disease can be obtained. This probability value will be used as part of the prediction model output to predict the indications of Chinese medicine formulas for specific diseases.

[0040] The inner product operation can quantify the similarity between two vectors. Here, the drug node vector represents the characteristics of Chinese medicine or Chinese medicine formula, while the disease node vector reflects the state of the disease. The larger the inner product, the stronger the correlation between the two. The result obtained by the inner product may be a continuous range of values, which is usually mapped to the [0,1] interval through a conversion operation (such as Sigmoid function or other probability conversion function) and expressed as a probability value. The higher the probability value, the stronger the indication association between the Chinese medicine formula and the specific disease.

[0041] The final probability value will be used as an output of the model to determine whether the Chinese medicine prescription is suitable for treating a specific disease. A high probability value means that the Chinese medicine prescription may be effective in treating the disease. Doctors can decide whether to use a certain Chinese medicine prescription for treatment based on the predicted association probability value, and it can also be used as a reference for formulating personalized treatment plans.

[0042] Finally, based on the prediction results, a suitable Chinese medicine prescription treatment plan is generated for the patient, which not only improves the targeted treatment, but also reduces unnecessary attempts and improves treatment efficiency.

[0043] S120, in the process of treating patients with indications based on the Chinese medicine prescription treatment plan, the latest biomics data of the corresponding patients is obtained, and the patients are monitored to obtain the corresponding treatment effects, and the Chinese medicine prescription treatment plan is adjusted based on the latest biomics data and the treatment effects.

[0044] Among them, bioinformatics data include but are not limited to genomics, transcriptomics, and proteomics; the preprocessing stage usually includes steps such as data cleaning (removing outliers, filling missing values, etc.), standardization or normalization (making different types of measurement data comparable).

[0045] By accurately capturing the patient's biomarkers, we can provide a basis for personalized treatment, thereby ensuring the data quality of subsequent analysis and improving the accuracy and reliability of the predictive model.

[0046] At different stages of treatment, biological changes during treatment can be reflected by regularly collecting patients' bio-omics data; at the same time, the treatment effect can be evaluated by monitoring patients' treatment responses through clinical observation or other means; and necessary adjustments can be made to the treatment plan based on the newly collected data and feedback on the treatment effect, thereby achieving dynamic optimization of the treatment plan, ensuring timely updating of treatment strategies, and helping to identify problems and adjust plans in a timely manner, which can better meet patients' treatment needs and achieve better treatment effects.

[0047] It should be pointed out here that in the embodiment of the present application, a heterogeneous network can be constructed by combining the known components of a single Chinese medicine, small molecule-target action data, target-disease association data, and drug-disease association data in the Fufang Cloud database. Specifically, a heterogeneous network is constructed, including but not limited to Chinese medicine component nodes, small molecule-target action nodes, target-disease association nodes, etc., and a comprehensive network is formed by calculating the similarity matrix between each node. This network contains different types of nodes and various types of connections between them, which are represented as an adjacency matrix, and the associated subnetworks of each node are further decomposed.

[0048] It should be pointed out here that in the process of obtaining the corresponding prediction results, the method also includes:

[0049] S131, based on the prediction results, design a laboratory test plan for the new Chinese medicine formula to evaluate the actual performance of the new Chinese medicine formula.

[0050] Among them, after obtaining the predicted results of the indications of the new Chinese medicine prescription, a series of laboratory test plans are designed based on this information. This includes but is not limited to pharmacology testing (evaluating the mechanism and effect of the drug), toxicology testing (studying the safety of the drug and its potential side effects), and bioavailability testing (investigating the absorption, distribution, metabolism and excretion of the drug in the body). Factors such as the selection of experimental animals, the method of administration, and the dosage range need to be considered during the design. This ensures that the hypotheses proposed by the predictive model are verified through scientific and rigorous methods, providing solid experimental evidence support for the practical application of the new prescription.

[0051] S132, executing an experiment corresponding to the laboratory test plan, and obtaining corresponding experimental data during the experiment.

[0052] The experimental work is carried out according to the designed test plan. For pharmacological testing, it may involve the use of specific disease prediction models to observe the therapeutic effect of drugs on diseases; toxicological testing requires recording the toxic reactions caused by drugs at different doses; and bioavailability testing usually uses technical means such as labeled compounds or radioactive isotope tracking methods. During the experiment, each step of the operation process, the observed phenomena and the final results must be recorded in detail. Obtaining first-hand information that directly reflects the performance of new Chinese medicine prescriptions is a key link in testing the degree of consistency between theoretical predictions and actual conditions.

[0053] S133, comparing the experimental data with the prediction results of the prediction model, obtaining the corresponding experimental comparison results, optimizing the prediction model according to the experimental comparison results, and if the error between the experimental data and the prediction results of the prediction model reaches an acceptable range, guiding the development of new Chinese medicine prescriptions based on the prediction results.

[0054] The data obtained from the experiment is compared with the results previously generated by the prediction model. This step involves the application of statistical methods, such as calculating the degree of difference between the two and determining the error range. If a significant deviation is found, it is necessary to further explore the cause. It may be that the prediction model parameters are improperly set, the input features are inaccurately selected, or there are unknown variables that affect the experimental results. Comparative analysis can verify the effectiveness and reliability of the prediction model, and it is also an important basis for adjusting and optimizing the prediction model.

[0055] In addition, when there is a large gap between the experimental data and the prediction model prediction, it is necessary to re-examine the entire modeling process, including whether the data preprocessing stage is reasonable, whether the feature selection is appropriate, whether the algorithm itself has defects, etc., and make corresponding adjustments accordingly. For example, it may be necessary to add new feature variables, change the prediction model structure, or adopt different training strategies. By continuously improving the prediction model, the accuracy of its prediction of future new prescription indications can be improved, thereby better serving actual research and development needs.

[0056] Finally, once the experimental results are highly consistent with the predictions of the prediction model after multiple iterations, it means that the prediction model has a high degree of credibility. At this time, the information output by the prediction model can be used to guide the research and development of new Chinese medicine prescriptions, such as selecting appropriate ingredient combinations, setting reasonable dosage levels, etc. This can accelerate the development of new drugs, reduce research and development costs, and ensure the safety and effectiveness of the final product.

[0057] Based on the execution steps from S131 to S133, by combining the prediction model with systematic laboratory testing, it can not only effectively improve the accuracy of the prediction of TCM prescription indications, but also provide scientific guidance for new drug research and development. It emphasizes the importance of a closed-loop process from theoretical prediction to practical verification and feedback correction, which is helpful to promote the modernization of traditional Chinese medicine and promote more efficient and accurate new drug creation.

[0058] It should be noted here that the prediction model used in the embodiment of the present application is trained in advance, and the specific training process includes:

[0059] S001, obtaining a set of bio-omics data, and associating each bio-omics data with a corresponding label.

[0060] Among them, bioinformatics data includes but is not limited to genomics, transcriptomics, and proteomics. Labels include effective, ineffective, and partially effective. The above labels are usually defined based on clinical trial results or expert experience, indicating the effectiveness of a certain treatment for the patient.

[0061] By collecting multiple types of bioinformatics data from patients and assigning a label to each piece of bioinformatics data, we ensure that labeled datasets can be used for supervised learning tasks, providing target criteria for model training and helping to distinguish which biomarkers are associated with treatment response.

[0062] S002, extracting feature data based on the bio-omics data set, and dividing the feature data corresponding to the bio-omics data set into training set data and test set data.

[0063] Among them, the extracted feature data is randomly divided into two parts, one part is used as a training set for model training, and the other part is used as a test set for evaluating model performance.

[0064] S003, select a deep neural network as the model, use the training set data to train the model, and adjust the model parameters to minimize the loss function, use the test set data to verify the generalization ability of the model, and determine the hyperparameter combination of the optimal model, and determine the corresponding prediction model.

[0065] Among them, deep neural networks are selected as prediction models because they can handle complex nonlinear relationships and are suitable for large-scale data sets. The model parameters are iteratively updated using the training set data to continuously improve the performance of the model on the training set, that is, to minimize the loss function. It should be pointed out that the parameter adjustment here includes adjusting hyperparameters such as the learning rate, number of layers, and number of nodes to optimize model performance. Parameter adjustment helps to find the best prediction model configuration, thereby improving prediction accuracy.

[0066] It should be pointed out here that in the model training process in the embodiment of the present application, attention mechanisms are introduced into the model architecture, including subnet attention mechanisms, layer attention mechanisms, and graph attention mechanisms, which are used to capture key features in biological relationships; rich vector representations of various types of nodes are generated through node embedding blocks, topological subgraph embedding blocks, graph attention blocks, and layer attention blocks. These vectors not only contain the inherent characteristics of the nodes, but also reflect the position of the nodes in the network, the degree of influence of their neighboring nodes, and the important features emphasized by the attention mechanism.

[0067] In addition, the test set is used to verify whether the prediction model is overfitting the training data, and the selection of the optimal model ensures that the prediction model performs best on new data.

[0068] In the embodiment of the present application, in the process of extracting feature data based on the bio-omics data set, the following steps are also performed:

[0069] S0021, classify all the extracted feature data, and use statistical tests to calculate the type importance data of each type of feature data.

[0070] First, all the extracted feature data are classified according to their types. For example, the features can be divided into gene expression data, protein expression data, metabolite data, etc., and statistical test methods (such as ANOVA, t-test, chi-square test, etc.) are used to calculate the importance of each type of feature data. Type importance data can help identify which types of feature data are more strongly correlated with treatment effects, thereby screening out the feature types that have the greatest impact on the prediction results to improve the efficiency and accuracy of the model.

[0071] S0022, sort all types of feature data based on type importance data, and select the feature types with the highest ranking.

[0072] All types of feature data are sorted using the calculated type importance data. In the embodiment of the present application, feature types with higher importance scores are ranked higher, and a portion of the top-ranked feature types are selected from the sorted feature types as the final feature types.

[0073] By sorting and selecting all types of feature data, the number of features can be further reduced, retaining only the most important feature types, which can simplify the model and improve the model's predictive ability.

[0074] S0023, the selected feature types are determined to be training feature types and divided into training set data and test set data.

[0075] By dividing the data set of a certain training feature type into training set data and test set data, for example, 70% of the data set can be used as a training set and 30% as a test set. The training set is used to train the model so that the model can learn the patterns in the data, while the test set is used to evaluate the generalization ability of the model to ensure that the model performs well on unseen data.

[0076] By executing steps S0021 to S0023, it can be ensured that the prediction model finally constructed can not only accurately predict the indications of Chinese herbal medicines for specific diseases, but also maintain high stability and generalization ability in practical applications.

[0077] It should be pointed out that after generating the corresponding Chinese medicine prescription treatment plan, the system first performs treatment simulation based on the Chinese medicine prescription treatment plan, which specifically includes the following processes:

[0078] S111, use AR development tools to identify the patient's imaging data and locate the patient's key organs and tissues based on the imaging data.

[0079] Among them, the image data is processed by algorithms to identify and mark the positions of the patient's key organs (such as the heart, liver, lungs, etc.) and tissues (such as tumor tissue, inflammatory areas, etc.). By accurately identifying and locating the positions of key organs and tissues, it can provide accurate reference points for subsequent virtual information overlay, which helps doctors better understand the patient's physical condition, especially in situations where precise operations are required, such as surgical planning.

[0080] S112, superimposing virtual information on the patient's body parts based on the key organ positions and tissue positions.

[0081] Among them, AR technology is used to superimpose virtual information on the patient's real-time view. For example, the location of key organs is marked with specific colors or icons. This can not only ensure that the virtual information accurately corresponds to the actual organ position, but also help doctors more intuitively observe the location of key organs and tissues, thereby making more accurate diagnoses or formulating more precise treatment plans, especially in complex surgical environments.

[0082] S113, using AR technology to display the change data of Chinese medicine ingredients in the body over time, and generating corresponding real-time feedback based on the change data, wherein the change data includes color change and dynamic path.

[0083] Among them, AR technology is used to display in real time the distribution of Chinese medicine ingredients in the patient's body and the changing trends over time. The display can include color changes (such as different colors represent different concentration levels) and dynamic paths (showing how the ingredients move), thereby generating corresponding feedback information based on the changing data of the Chinese medicine ingredients. The real-time display of changes in Chinese medicine ingredients helps doctors monitor the efficacy of drugs and adjust medication plans in a timely manner.

[0084] By executing the above steps S111 to S113, not only the accuracy and efficiency of doctors in the diagnosis and treatment process can be improved, but also the patient's sense of participation and treatment experience can be enhanced.

[0085] In addition, in the embodiment of the present application, the following steps are also performed:

[0086] S121, the gene chip acquires and collects the patient's genomic data and pre-processes the genomic data.

[0087] Among them, gene chip technology is used to obtain the patient's genomic data. Gene chips can simultaneously detect information on thousands of gene sites, including whole genome sequencing (WGS), whole exome sequencing (WES) and target region sequencing (TGS), etc. The obtained raw genomic data is subjected to a series of processing, including but not limited to: quality control (QC): remove low-quality data points or samples to ensure the data quality of subsequent analysis; variation detection: identify variations in the genome, such as single nucleotide polymorphisms (SNPs), insertions / deletions (INDELs), etc.; annotation: compare the detected variation with the information in the known database to understand its functional significance and possible impact. Variation detection and annotation help to understand the significance of the variation and lay the foundation for further genetic risk assessment.

[0088] This ensures that the genomics data used for analysis is high-quality and accurate, reducing erroneous conclusions due to data quality issues.

[0089] S122, use machine learning algorithms to build an evaluation model, the evaluation model is trained based on the patient's genetic variation information, clinical data and demographic data as input features, and the evaluation model is trained and verified through cross-validation methods.

[0090] Among them, the specific training process of the evaluation model can refer to the training of the above-mentioned prediction model, select appropriate machine learning algorithms, such as logistic regression, random forest, support vector machine, etc., and train the model based on the patient's genetic variation information, clinical data and demographic data as input features. The model is trained and verified by cross-validation method to ensure the generalization ability of the model, that is, the model can perform well on unseen data, avoid overfitting, and make the model more reliable.

[0091] S123, inputting the genomic data into the evaluation model prediction results to generate a genetic risk assessment report, and adjusting the patient's treatment plan based on the genetic risk assessment report.

[0092] Among them, the genetic risk assessment report includes the patient's genetic susceptibility to specific drugs, potential adverse reaction risks and corresponding recommendations. The genetic risk assessment report provides a scientific basis for doctors to help them develop personalized treatment plans based on the patient's specific circumstances. It can also enhance the safety and effectiveness of treatment and reduce the risk of unnecessary side effects.

[0093] In an embodiment of the present application, during the calling and using process of the prediction model, the following steps are also performed: the retraining process of the prediction model is periodically triggered, and during the retraining process, advanced optimization algorithms are used to adjust the parameters of the prediction model; after each prediction model update, an independent validation set is used to evaluate the performance of the updated prediction model. If the performance of the updated prediction model is better than that of the prediction model before the update, the updated prediction model is adopted; otherwise, the system will continue to use the prediction model before the update.

[0094] First, by setting a fixed cycle (such as weekly, monthly or quarterly), the model retraining process is automatically started at the end of this cycle. Before the retraining process, the latest data set needs to be collected, including newly collected genomic data, clinical data, and any relevant treatment outcome data. Similarly, the new data needs to be quality checked, outliers cleaned, and standardized or normalized as needed. By retraining the model regularly, it can be ensured that the model is always based on the latest available data, thereby improving the prediction accuracy.

[0095] Then, according to the characteristics of the model, select a suitable optimization algorithm, such as Adam, RMSprop, or a more advanced adaptive learning rate method, and use the selected optimization algorithm to adjust the parameters of the model to minimize the loss function so that the model better fits the training data. Parameter adjustment enables the model to better learn patterns in the data and improve prediction accuracy.

[0096] Secondly, after each prediction model update, a portion of independent data is separated from the dataset as a validation set. This portion of data does not participate in the training of the model. After the model is updated, the data on the validation set is used to test the performance of the model. The use of an independent validation set can objectively evaluate the generalization ability of the model and prevent the model from overfitting the training data.

[0097] Finally, if the performance of the updated prediction model is better than that of the prediction model before the update, the updated prediction model will be adopted; otherwise, the system will continue to use the prediction model before the update, and compare the performance of the new and old models on the validation set. If the new model is better than the old model in all key indicators, the new model is considered better. If the new model performs better, it will be deployed to the production environment to replace the old model; otherwise, the current model will continue to be used. This mechanism ensures that the system always uses the model with the best performance, thereby improving the stability and reliability of the system.

[0098] Through the above steps, the prediction model can be continuously optimized and updated to ensure that it is always in the best state, thereby improving the accuracy of genetic risk assessment and thus improving the treatment effect of patients. This method not only enhances the dynamic adaptability of the model, but also provides more solid support for clinical decision-making.

[0099] The present application also discloses a device for predicting the indications of a Chinese medicine prescription, referring to Figure 2 ,include:

[0100] The data set collection module 200 is used to collect the data set of the relationship between the existing Chinese medicine prescriptions and the corresponding indications, pre-process the data set, and train the corresponding prediction model based on the pre-processed data set;

[0101] The prediction result acquisition module 210 acquires new Chinese medicine prescription data to be predicted, inputs the pre-processed new Chinese medicine prescription data into the trained prediction model to obtain the corresponding prediction result, and generates the corresponding Chinese medicine prescription treatment plan based on the prediction result, wherein the prediction model is used to predict the indication of the Chinese medicine prescription for a specific disease;

[0102] The Chinese medicine prescription treatment plan adjustment module 220 obtains the latest bio-omics data of the corresponding patient during the treatment of the patient with the indication based on the Chinese medicine prescription treatment plan, monitors the patient to obtain the corresponding treatment effect, and adjusts the Chinese medicine prescription treatment plan based on the latest bio-omics data and the treatment effect, wherein the bio-omics data includes genomics, transcriptomics, and proteomics.

[0103] Optionally, the device also includes but is not limited to:

[0104] A laboratory test program design module, based on the prediction results, is used to design a laboratory test program for the new Chinese medicine prescription to evaluate the actual performance of the new Chinese medicine prescription, and the laboratory test program includes pharmacology test, toxicology test, and bioavailability test;

[0105] An experimental data acquisition module, which executes the experiment corresponding to the laboratory test plan and is used to acquire corresponding experimental data during the experiment, wherein the experimental data includes a detailed experimental process and observation results;

[0106] The experimental comparison result acquisition module compares the experimental data with the prediction results of the prediction model to obtain the corresponding experimental comparison results, and optimizes the prediction model according to the experimental comparison results. If the error between the experimental data and the prediction results of the prediction model reaches an acceptable range, the research and development of new Chinese medicine prescriptions is guided based on the prediction results.

[0107] Optionally, the device also includes but is not limited to:

[0108] An association module, which obtains a set of bio-omics data and is used to associate each bio-omics data with a corresponding label, wherein the bio-omics data includes but is not limited to genomics, transcriptomics, and proteomics, and the labels include valid, invalid, and partially valid;

[0109] A feature extraction module is used to extract feature data based on the bio-omics data set, and divide the feature data corresponding to the bio-omics data set into training set data and test set data;

[0110] The prediction model determination module selects a deep neural network as the model, uses the training set data to train the model, adjusts the model parameters to minimize the loss function, uses the test set data to verify the generalization ability of the model, and determines the optimal model hyperparameter combination to determine the corresponding prediction model.

[0111] Optionally, the device also includes but is not limited to:

[0112] A calculation module, which classifies all the extracted feature data and uses statistical tests to calculate type importance data of each type of feature data;

[0113] The sorting module is used to sort all types of feature data based on type importance data and select the top-ranked feature types;

[0114] The data partitioning module is used to divide the selected feature types into training set data and test set data to determine the training feature types.

[0115] Optionally, the device also includes but is not limited to:

[0116] AR recognition module, used to use AR development tools to identify the patient's image data and locate the patient's key organs and tissues based on the image data;

[0117] A virtual information adding module is used to overlay virtual information on the patient's body parts based on the key organ locations and tissue locations;

[0118] The real-time feedback generation module uses AR technology to display the changing data of Chinese medicine ingredients in the body over time, and generates corresponding real-time feedback based on the changing data, where the changing data includes color changes and dynamic paths.

[0119] Optionally, the device also includes but is not limited to:

[0120] Gene chip module: Gene chip is used to obtain and collect genomic data of patients and pre-process the genomic data. Genomic data includes whole genome sequencing, whole exome sequencing and target region sequencing. Pre-processing includes quality control, variation detection and annotation.

[0121] The evaluation model building module uses machine learning algorithms to build the evaluation model. The evaluation model is trained based on the patient's genetic variation information, clinical data, and demographic data as input features. The evaluation model is trained and validated through cross-validation methods.

[0122] The genetic risk assessment report generation module inputs genomic data into the assessment model prediction results to generate a genetic risk assessment report, and adjusts the patient's treatment plan based on the genetic risk assessment report. The genetic risk assessment report contains the patient's genetic susceptibility to specific drugs, potential adverse reaction risks and corresponding recommendations.

[0123] Optionally, the device also includes but is not limited to:

[0124] The retraining process of the prediction model is triggered periodically. During the retraining process, advanced optimization algorithms are used to adjust the parameters of the prediction model. After each prediction model update, an independent validation set is used to evaluate the performance of the updated prediction model. If the performance of the updated prediction model is better than that of the prediction model before the update, the updated prediction model is adopted; otherwise, the system will continue to use the prediction model before the update.

[0125] The embodiment of the present application also discloses a method for predicting indications of a traditional Chinese medicine prescription, comprising a processor, wherein a program for predicting indications of any one of the above-mentioned methods is run in the processor.

[0126] The embodiment of the present application also discloses a storage medium storing a program of any one of the above-mentioned methods for predicting indications of a traditional Chinese medicine prescription.

[0127] In summary, by collecting and utilizing a variety of bio-omics data from patients, we can fully understand the physiological status of patients at the molecular level, which provides high-quality input data for subsequent predictive models, thereby improving the accuracy of predictions.

[0128] Based on each patient's unique bio-omics characteristics, the predictive model can customize and recommend the most effective Chinese medicine prescription, making the treatment plan more targeted and personalized, which helps to improve the treatment effect.

[0129] During the treatment process, by regularly collecting the patient's latest bio-omics data and dynamically adjusting the treatment plan based on the actual treatment effect, problems that arise during the treatment process can be discovered and responded to in a timely manner, ensuring that the treatment plan is always in the optimal state.

[0130] By ranking and selecting the importance of feature data, the model training process is optimized, and the efficiency and explanatory power of the model are improved, so that the model can not only accurately predict indications but also better understand the influencing factors.

[0131] With the help of AR technology, the dynamic changes of Chinese medicine ingredients in the body can be intuitively displayed, providing real-time feedback to help doctors and patients better understand the treatment process.

[0132] Genomic data obtained through gene chips are combined with clinical and demographic data to construct an assessment model to predict genetic risks, thereby providing patients with personalized medical advice and optimizing treatment strategies.

[0133] Periodic model retraining and evaluation ensures that the model can continuously learn from the latest data, maintain its optimal predictive performance, and improve the reliability and practicality of the system.

[0134] In summary, the above scheme can predict the most effective Chinese medicine prescription based on the unique bio-omics characteristics of each patient through multi-dimensional data collection, efficient model training, dynamic treatment adjustment and the application of modern technology, thereby improving the accuracy of indication prediction.

[0135] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for predicting indications of a traditional Chinese medicine prescription, characterized in that: include: Collect the data set of the relationship between the existing Chinese medicine prescriptions and the corresponding indications, preprocess the data set, and train the corresponding prediction model based on the preprocessed data set; Acquire new Chinese medicine prescription data to be predicted, input the preprocessed new Chinese medicine prescription data into the trained prediction model, obtain corresponding prediction results, and generate corresponding Chinese medicine prescription treatment plans based on the prediction results, wherein the prediction model is used to predict the indications of Chinese medicine prescriptions for specific diseases; In the process of treating patients with indications based on Chinese medicine prescriptions, the latest bioomics data of the corresponding patients is obtained, and the patients are monitored to obtain the corresponding treatment effects. The Chinese medicine prescription treatment plan is adjusted based on the latest bioomics data and treatment effects, where the bioomics data includes genomics, transcriptomics, and proteomics.

2. The method for predicting the indications of a Chinese medicine prescription according to claim 1, characterized in that: In the process of obtaining the corresponding prediction result, the method further includes: Based on the prediction results, designing a laboratory test plan for the new Chinese medicine formula to evaluate the actual performance of the new Chinese medicine formula, the laboratory test plan including pharmacology test, toxicology test, and bioavailability test; Performing an experiment corresponding to the laboratory test plan, and obtaining corresponding experimental data during the experiment, wherein the experimental data includes a detailed experimental process and observation results; The experimental data are compared with the predicted results of the prediction model to obtain the corresponding experimental comparison results, and the prediction model is optimized according to the experimental comparison results. If the error between the experimental data and the predicted results of the prediction model reaches an acceptable range, the research and development of new Chinese medicine prescriptions is guided based on the prediction results.

3. The method for predicting the indications of a Chinese medicine prescription according to claim 1, characterized in that: In the process of training the prediction model, the method further includes: Obtain a set of bio-omics data, and associate each bio-omics data with a corresponding label, wherein the bio-omics data includes but is not limited to genomics, transcriptomics, and proteomics, and the labels include valid, invalid, and partially valid; Extract feature data based on the bio-omics data set, and divide the feature data corresponding to the bio-omics data set into training set data and test set data; A deep neural network is selected as the model, the training set data is used to train the model, and the model parameters are adjusted to minimize the loss function. The test set data is used to verify the generalization ability of the model, and the hyperparameter combination of the optimal model is determined to determine the corresponding prediction model.

4. The method for predicting the indications of a Chinese medicine prescription according to claim 2, characterized in that: In the process of extracting feature data based on the bio-omics data set, the method also includes: Classify all the extracted feature data and use statistical tests to calculate the type importance data of each type of feature data; Sort all types of feature data based on type importance data and select the top-ranked feature types; The selected feature type determines the training feature type and divides it into training set data and test set data.

5. The method for predicting the indications of a Chinese medicine prescription according to claim 1, characterized in that: For individual patients, The method also includes: Determine the corresponding target individual patient, and obtain the individual bio-omics data corresponding to the target individual patient; Determining corresponding patient condition data based on the individual patient biomics data; A mapping relationship table between symptom data and a Chinese medicine prescription treatment plan is retrieved, based on the patient symptom data and the Chinese medicine prescription treatment plan, wherein the mapping relationship table is generated by the prediction model.

6. The method for predicting the indications of a Chinese medicine prescription according to claim 1, characterized in that: The method also includes: Gene chip acquisition and collection of patients' genomic data, and pre-processing of genomic data, including whole genome sequencing, whole exome sequencing and target region sequencing, pre-processing including quality control, variation detection and annotation; Use machine learning algorithms to build an evaluation model, which is trained based on the patient's genetic variation information, clinical data, and demographic data as input features. The evaluation model is trained and validated using cross-validation methods. The genomic data is input into the evaluation model to predict the results and generate a genetic risk assessment report, based on which the patient's treatment plan is adjusted. The genetic risk assessment report contains the patient's genetic susceptibility to specific drugs, potential adverse reaction risks and corresponding recommendations.

7. The method for predicting the indications of a Chinese medicine prescription according to claim 1, characterized in that: The method also includes: The retraining process of the prediction model is triggered periodically. During the retraining process, advanced optimization algorithms are used to adjust the parameters of the prediction model. After each prediction model update, an independent validation set is used to evaluate the performance of the updated prediction model. If the performance of the updated prediction model is better than that of the prediction model before the update, the updated prediction model is adopted; otherwise, the system will continue to use the prediction model before the update.

8. A device for predicting indications of traditional Chinese medicine prescriptions, characterized in that: include: A data set collection module is used to collect data sets of the relationship between existing Chinese medicine prescriptions and corresponding indications, preprocess the data sets, and train the corresponding prediction model based on the preprocessed data sets; A prediction result acquisition module is used to acquire new Chinese medicine prescription data to be predicted, input the pre-processed new Chinese medicine prescription data into the trained prediction model to obtain the corresponding prediction results, and generate the corresponding Chinese medicine prescription treatment plan based on the prediction results, wherein the prediction model is used to predict the indications of the Chinese medicine prescription for a specific disease; The Chinese medicine prescription treatment plan adjustment module obtains the latest biomics data of the corresponding patients during the treatment of patients with indications based on the Chinese medicine prescription treatment plan, monitors the patients to obtain the corresponding treatment effects, and adjusts the Chinese medicine prescription treatment plan based on the latest biomics data and treatment effects, among which the biomics data includes genomics, transcriptomics, and proteomics.

9. A method for predicting indications of a traditional Chinese medicine prescription, characterized in that: The method comprises a processor, wherein a program of a method for predicting indications of a traditional Chinese medicine prescription according to any one of claims 1 to 7 is run in the processor.

10. A storage medium, characterized in that: A program storing a method for predicting indications of a traditional Chinese medicine prescription according to any one of claims 1 to 7.

Images