Anti-tumor treatment adverse reaction risk early warning method and system

Abstract

The invention discloses an anti-tumor treatment adverse reaction risk early warning method and system, and relates to the technical field of medical information processing, and the method comprises the steps: obtaining the physiological index data of a patient, and obtaining a short-term risk prediction result based on the time sequence change characteristics of the physiological index data of a deep neural network; obtaining clinical feature data of the patient, constructing a susceptibility assessment model, and analyzing correlation features of the clinical feature data through the susceptibility assessment model to obtain a baseline risk assessment result; and according to the short-term risk prediction result and the baseline risk assessment result, calculating a comprehensive risk score, and when the comprehensive risk score exceeds an early warning threshold of the corresponding adverse reaction type, triggering an early warning signal. According to the method, a double-layer risk assessment system is constructed by combining instant physiological index changes and clinical features of patients, the technical problems that an existing early warning method is insufficient in consideration of individual differences of the patients and poor in early warning timeliness are solved, and the accuracy and timeliness of early warning of adverse reactions of anti-tumor treatment are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical information processing, and in particular to a method and system for early warning of risks of adverse reactions to anti-tumor treatment. Background Art

[0002] In the field of anti-tumor treatment, adverse reaction monitoring and risk warning of patients have always been the core topics of clinical medicine. Traditional adverse reaction monitoring methods mainly rely on threshold judgment and empirical evaluation of single physiological indicators. Although this method is simple to operate, it is difficult to fully reflect the coordinated change characteristics of multiple organ systems in the human body. In recent years, with the in-depth application of machine learning technology in the medical field, risk warning models based on multi-source data fusion have gradually become a research hotspot. This type of model constructs a risk assessment framework by integrating multi-dimensional information such as clinical feature data, laboratory test results, and treatment parameters. However, existing risk warning models often use simple feature combinations or linear models for risk assessment, lack in-depth analysis of complex correlation patterns between clinical features, and are difficult to accurately capture the interactive effects between different organ systems.

[0003] The current risk warning system for adverse reactions to anti-tumor treatment has the following technical problems: First, most of the existing monitoring methods use a single indicator threshold to make judgments, ignoring the coordinated change characteristics of multiple indicators in the same organ system, resulting in insufficient sensitivity and specificity of the warning; second, the existing risk assessment model does not conduct in-depth enough correlation analysis of clinical characteristics, fails to fully consider the interactions and temporal dependencies between different characteristics, and affects the accuracy of risk assessment; third, the traditional warning mechanism lacks quantitative analysis of the risk propagation characteristics of multiple organ systems and cannot effectively predict potential chain reactions; finally, the existing technology generally has a contradiction between response speed and accuracy, making it difficult to achieve rapid response while ensuring the accuracy of the warning. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method and system for early warning of the risk of adverse reactions to anti-tumor treatment, which can solve the problems mentioned in the background technology.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for early warning of the risk of adverse reactions to anti-tumor treatment, comprising: obtaining physiological indicator data of a patient, constructing an immediate risk assessment model, analyzing the time series variation characteristics of the physiological indicator data through the immediate risk assessment model, and obtaining a short-term risk prediction result; Acquiring clinical characteristic data of the patient, constructing a susceptibility assessment model, analyzing the correlation characteristics of the clinical characteristic data through the susceptibility assessment model, and obtaining a baseline risk assessment result; A comprehensive risk score is calculated based on the short-term risk prediction result and the baseline risk assessment result, and a warning signal is triggered when the comprehensive risk score exceeds a warning threshold of a corresponding adverse reaction type.

[0007] As a preferred embodiment of the anti-tumor treatment adverse reaction risk early warning method of the present invention, wherein: the physiological index data includes physical sign monitoring data, organ function index, immune function index, hematopoietic function index, and metabolic level index; Acquiring the patient's physiological indicator data, building an immediate risk assessment model, and analyzing the temporal variation characteristics of the physiological indicator data through the immediate risk assessment model to obtain a short-term risk prediction result, including the following steps: The physiological index data are grouped according to the organ system classification of the WHO adverse reaction grading standard, and the co-variability of each group of indicators is calculated; if the dispersion of a group of indicators exceeds the mean level of the group of indicators, and the change range of two or more indicators in the group of indicators exceeds the standard deviation range of their baseline values, then the group of indicators is determined as a key monitoring category, otherwise, the group of indicators is determined as a routine monitoring category; The physiological indicator data are input into the immediate risk assessment model. If there is an indicator group determined as a key monitoring category, the immediate risk assessment model will give priority to analyzing the time series correlation characteristics of each indicator determined as a key monitoring category and output the short-term risk prediction result. If the physiological indicator data all belong to the conventional monitoring category, the short-term risk prediction result will be output after analyzing the combined characteristics of all category indicators.

[0008] As a preferred solution of the anti-tumor treatment adverse reaction risk warning method of the present invention, the calculation of the collaborative variability includes the following steps: Obtain the values ​​of each indicator in the same organ system within three consecutive sampling periods and calculate the standard deviation of each indicator; The mean of the standard deviations of all indicators was calculated, and the covariance was obtained by dividing the mean of the standard deviations by the overall mean of the indicators within the organ system; The time series correlation feature is obtained by calculating the mutual correlation coefficient between the indicators, wherein the synchronization is determined based on the mutual correlation coefficient of zero time delay, and the hysteresis is determined based on the time delay corresponding to the maximum mutual correlation coefficient; The combined features are obtained based on hierarchical clustering analysis between indicators of different organ systems, and a feature combination relationship matrix is ​​constructed by calculating the Euclidean distance between indicators; The instant risk assessment model adopts a dual-branch structure, which includes a time series analysis branch and a feature fusion branch; wherein the time series analysis branch adopts a multi-layer LSTM network structure, and each layer processes change patterns at different time scales; the feature fusion branch adopts an attention mechanism to adjust the weights of indicators of different organ systems.

[0009] As a preferred embodiment of the anti-tumor treatment adverse reaction risk warning method of the present invention, the clinical characteristic data includes laboratory test data, imaging test data, gene detection data, medication regimen parameters, and treatment cycle indicators; Acquiring clinical characteristic data of the patient, building a susceptibility assessment model, analyzing the correlation characteristics of the clinical characteristic data through the susceptibility assessment model, and obtaining a baseline risk assessment result, including the following steps: A feature association network is constructed based on the clinical feature data, and the association strength between clinical features in the feature association network is calculated; wherein the association strength is the conditional mutual information of the clinical feature set; if the conditional mutual information of the clinical feature set is greater than the marginal mutual information of its constituent features, and the mutual information entropy between the organ system corresponding to the clinical feature set and the physiological indicator data is greater than the mean value of the mutual information entropy, then the clinical feature set is marked as a core feature set, otherwise it is marked as a common feature set; The feature association network is input into the susceptibility assessment model. If there is a clinical feature set marked as a core feature set, the susceptibility assessment model calculates the conditional probability distribution of the core feature set according to the Bayesian network and outputs the baseline risk assessment result. If the clinical feature sets are all common feature sets, the baseline risk assessment result is calculated based on the feature propagation algorithm of the graph neural network.

[0010] As a preferred solution of the anti-tumor treatment adverse reaction risk warning method of the present invention, the feature association network is constructed by the following steps: The clinical characteristic data are grouped according to the organ system classification of the WHO adverse reaction grading standard to obtain an initial set of clinical characteristics; Calculating the conditional mutual information and marginal mutual information within the initial set of clinical features to generate nodes of the feature association network; The connection relationship is determined based on the mutual information entropy between the nodes; if the clinical features corresponding to the two nodes belong to the same organ system, the conditional mutual information between the nodes is used as the connection weight; if the clinical features corresponding to the two nodes belong to different organ systems, a connection is established when the mutual information entropy between the nodes exceeds the average mutual information entropy within the respective organ systems, and the excess value is used as the connection weight; the updating of the feature association network is achieved by calculating the rate of change of the connection weight and iteratively propagating it between the nodes.

[0011] As a preferred embodiment of the method for early warning of the risk of adverse reactions to anti-tumor treatment described in the present invention, a comprehensive risk score is calculated based on the short-term risk prediction result and the baseline risk assessment result, including: Establishing a risk fusion matrix according to the organ system classification, and calculating risk weights of different time scales based on the short-term risk prediction results and the baseline risk assessment results; Applying the risk weight to the risk fusion matrix to calculate a system risk score; if the short-term risk prediction result and the baseline risk assessment result correspond to the same organ system, the diagonal term of the system risk score is calculated by a coupling coefficient; if they correspond to different organ systems, the non-diagonal term of the system risk score is calculated by a risk transfer coefficient; The system risk score is obtained by normalizing the risk fusion matrix using an eigenvalue decomposition method; The comprehensive risk score is calculated as follows: Constructing a risk feature vector, organizing the short-term risk prediction result and the baseline risk assessment result into a matrix form according to the corresponding relationship between organ systems; Calculate the eigenvalue and eigenvector of the risk eigenvector, and select the eigenvector corresponding to the maximum eigenvalue as the main risk direction; Taking the projection of the risk feature vector in the main risk direction as the reference value of the comprehensive risk score; The reference value of the comprehensive risk score is corrected based on the coupling coefficient and the risk transfer coefficient.

[0012] As a preferred solution of the anti-tumor treatment adverse reaction risk warning method of the present invention, wherein: the risk weight is adjusted by a time decay factor; If there is a predicted value of the key monitoring category in the short-term risk prediction result, the inverse of the exponential function of the predicted value is used as the time decay factor to update the diagonal elements of the risk fusion matrix; If there is an evaluation value of the core feature set in the baseline risk evaluation result, the exponential function inverse of the evaluation value is used as the time decay factor to update the non-diagonal elements of the risk fusion matrix; The calculation period of the time attenuation factor is inversely proportional to the occurrence period of the corresponding adverse reaction.

[0013] To further solve the above technical problems, the present invention provides the following technical solutions: an anti-tumor treatment adverse reaction risk warning system, comprising: a collection module, for acquiring physiological index data and clinical characteristic data of patients; A processing module, used to construct an immediate risk assessment model and a susceptibility assessment model, analyze the temporal variation characteristics of the physiological indicator data through the immediate risk assessment model to obtain a short-term risk prediction result, and analyze the correlation characteristics of the clinical characteristic data through the susceptibility assessment model to obtain a baseline risk assessment result; The early warning module is used to calculate a comprehensive risk score based on the short-term risk prediction result and the baseline risk assessment result, and trigger a warning signal when the comprehensive risk score exceeds the early warning threshold of the corresponding adverse reaction type.

[0014] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the steps of the above-mentioned anti-tumor treatment adverse reaction risk warning method when executing the computer program.

[0015] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the above-mentioned anti-tumor treatment adverse reaction risk warning method are implemented.

[0016] Beneficial effects of the present invention: First, the present invention overcomes the limitation that the traditional single-indicator monitoring method is difficult to reflect the overall functional state of the organ system by introducing the concept of collaborative variability and a dual-channel analysis mechanism, and realizes the accurate identification of the coordinated changes of multiple indicators, thereby improving the response speed to high-risk situations. Secondly, the present invention constructs a feature association network based on organ system classification, and through differentiated connection strategies and a two-layer architecture design, it realizes the intelligent identification of complex association patterns between clinical features, and solves the problem of insufficient feature interaction analysis in existing risk warning models. Finally, the present invention adopts a risk scoring method based on time attenuation, and by introducing coupling coefficients and risk transfer coefficients, it establishes a quantitative assessment mechanism for risk transmission among multiple organ systems, breaking through the technical bottleneck that traditional early warning systems are difficult to predict chain reactions. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0018] Figure 1 This is a schematic diagram of the overall process of a method for early warning of risk of adverse reactions to anti-tumor treatment proposed by the present invention; Figure 2 This is a diagram of computer equipment in a method for early warning of risks of adverse reactions to anti-tumor treatment proposed by the present invention. DETAILED DESCRIPTION

[0019] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0020] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0021] Example 1, reference Figure 1 , which is an embodiment of the present invention, provides a method for early warning of the risk of adverse reactions to anti-tumor treatment.

[0022] Figure 1 The overall process diagram of an anti-tumor treatment adverse reaction risk warning method is shown, including the following steps: S1: Obtain the patient's physiological indicator data, build an immediate risk assessment model, analyze the temporal variation characteristics of the physiological indicator data through the immediate risk assessment model, and obtain short-term risk prediction results.

[0023] Specifically, physiological indicator data include physical sign monitoring data, organ function indicators, immune function indicators, hematopoietic function indicators, and metabolic level indicators.

[0024] S1.1: Group the physiological indicator data according to the organ system classification of the WHO adverse reaction grading standard, and calculate the co-variability of each group of indicators, where the co-variability represents the consistency of changes in various indicators within the same organ system; if the degree of dispersion of a group of indicators exceeds the mean level of the group of indicators, and there are two or more indicators in the group whose change range exceeds the standard deviation range of their baseline values, then the group of indicators will be determined as a key monitoring category, otherwise the group of indicators will be determined as a routine monitoring category.

[0025] S1.2: Input the physiological indicator data into the immediate risk assessment model. If there is an indicator group identified as a key monitoring category, the immediate risk assessment model will give priority to analyzing the temporal correlation characteristics of each indicator under this category and output the short-term risk prediction results. If the physiological indicator data all belong to the routine monitoring category, the combined characteristics of all category indicators will be analyzed and the short-term risk prediction results will be output.

[0026] Specifically, the calculation of co-variability includes the following steps: first, obtain the values ​​of various indicators in the same organ system in three consecutive sampling periods, calculate the standard deviation of each indicator, then calculate the mean of the standard deviations of all indicators, and finally divide the mean of the standard deviations by the overall mean of the organ system indicators to obtain the co-variability.

[0027] The time series correlation characteristics are obtained by calculating the mutual correlation coefficients between indicators, where synchronization is determined based on the mutual correlation coefficient of zero time delay, and hysteresis is determined based on the time delay corresponding to the maximum mutual correlation coefficient. The calculation process of the mutual correlation coefficient involves two time series indicator data. Taking liver function monitoring as an example, when analyzing the time series correlation characteristics of alanine aminotransferase ALT and aspartate aminotransferase AST: the synchronization calculation process is to first obtain the sequence data of the two indicators at the same time point, standardize the data at each time point, and eliminate the dimension effect. Then calculate the mutual correlation coefficient of zero time delay (that is, without considering time offset), specifically, multiply and sum the values ​​of the two standardized sequences at the corresponding time points, and then divide them by the product of their respective standard deviations. The mutual correlation coefficient reflects the degree of consistency of the changes of two indicators at the same time point. For example, if the zero-delay mutual correlation coefficient of ALT and AST is close to 1, it indicates that the two indicators rise or fall synchronously, which often indicates the occurrence of acute liver injury. The hysteresis calculation process is to gradually move one of the indicator sequences on the time axis (it can be moved forward or backward), calculate the correlation coefficient at each offset position, and find the time offset corresponding to the maximum correlation coefficient. This offset reflects the hysteresis relationship between the two indicators. For example, if it is found that the change in bilirubin level reaches its peak 24 hours later than the transaminase level, this time delay feature can be used to predict the progression of liver damage. This has important guiding significance for the timely adjustment of clinical intervention measures.

[0028] The combined features are obtained based on the hierarchical cluster analysis between the indicators of different organ systems, and the feature combination relationship matrix is ​​constructed by calculating the Euclidean distance between the indicators. The above mechanism adopts differentiated analysis strategies for different risk levels, which ensures the accuracy of early warning while improving the response speed to high-risk situations.

[0029] Furthermore, the instant risk assessment model is a risk prediction model based on deep learning, which is specially designed for the characteristics of adverse reactions of anti-tumor treatment. The model adopts a dual-branch structure: a time series analysis branch and a feature fusion branch. The time series analysis branch adopts a multi-layer LSTM network structure, and each layer processes the change pattern of different time scales. Taking liver function indicators as an example, the first layer of LSTM captures acute changes at the hour level (such as the rapid increase of transaminase), the second layer captures the cumulative effect at the day level (such as the gradual change of bilirubin), and the third layer focuses on the long-term trend at the weekly level (such as the slow decline of albumin levels). This multi-scale time series analysis can monitor acute adverse reactions and chronic cumulative damage at the same time. The feature fusion branch adopts an attention mechanism to dynamically adjust the weights of indicators of different organ systems. For example, when an abnormal cardiac function indicator is detected, the model will automatically increase the attention to the renal function indicator, because cardiac and renal functions are often closely related. This dynamic weight adjustment between features enables the model to capture the interactive effects between organ systems. The output layer of the model is designed with a hierarchical prediction mechanism: for key monitoring categories, the probability prediction value of each specific adverse reaction type is directly output; for routine monitoring categories, a comprehensive risk score is output. This hierarchical output mechanism ensures timely response to high-risk situations while maintaining continuous assessment of overall risks.

[0030] In S1.1, the present invention selects the WHO adverse reaction classification standard as the basis for indicator grouping, which is different from the traditional single indicator monitoring, reflecting the present invention's in-depth understanding of clinical practice. Through the classification of organ systems, the correlation between different physiological indicators is established, making the monitoring more in line with the overall physiological characteristics of the human body. The introduced concept of co-variability not only avoids the problem of direct comparison of indicators of different dimensions by calculating the ratio of the standard deviation to the mean, but also reflects the overall change trend of each indicator in the same organ system, which is of great significance for the early identification of organ dysfunction. Especially in anti-tumor treatment, adverse drug reactions are often first manifested as coordinated changes in multiple indicators within a certain organ system, rather than significant abnormalities in a single indicator.

[0031] The dual-channel analysis mechanism in S1.2 solves the contradiction between response speed and accuracy in traditional early warning methods. The time series correlation characteristics of key monitoring categories are analyzed first, and the synchronization and lag between indicators are calculated by the mutual correlation coefficient, so that early signs of adverse reactions can be captured in time. This analysis method is particularly suitable for the field of anti-tumor treatment, because some serious adverse reactions (such as myocardial damage, liver function damage, etc.) often manifest as a chain reaction of multiple related indicators. For conventional monitoring categories, hierarchical clustering analysis is used to construct a feature combination relationship matrix, and the degree of correlation between indicators of different organ systems is measured by Euclidean distance, which not only ensures the comprehensiveness of the early warning, but also avoids the waste of computing resources.

[0032] From the perspective of solving technical problems, the present invention solves the problem that traditional single-indicator monitoring cannot reflect the overall functional status of the organ system through the calculation method of coordinated variability. The dual-channel analysis mechanism overcomes the defect of the existing early warning system in insufficient recognition of coordinated changes in multiple indicators. Especially in clinical practice, this organ system-based group monitoring method can effectively reduce false positive warnings and improve the accuracy of warnings. For example, in the monitoring of liver function damage caused by chemotherapy drugs, traditional methods often rely on the threshold judgment of a single indicator (such as transaminase levels), while the present invention can detect potential liver damage risks earlier by analyzing the coordinated changes of liver function-related indicators.

[0033] S2: Obtain the patient's clinical characteristic data, build a susceptibility assessment model, analyze the correlation characteristics of the clinical characteristic data through the susceptibility assessment model, and obtain the baseline risk assessment results.

[0034] Specifically, clinical characteristic data include laboratory test data, imaging test data, genetic testing data, medication regimen parameters, and treatment cycle indicators.

[0035] S2.1: Construct a feature association network based on clinical feature data and calculate the correlation strength between clinical features in the feature association network.

[0036] Among them, the correlation strength is the conditional mutual information of the clinical feature set; if the conditional mutual information of the clinical feature set is greater than the marginal mutual information of its constituent features, and the mutual information entropy of the organ system and physiological indicator data corresponding to the clinical feature set is greater than the mean mutual information entropy, then the clinical feature set is marked as a core feature set, otherwise it is marked as a common feature set.

[0037] S2.2: The feature association network is input into the susceptibility assessment model. If there is a clinical feature set marked as a core feature set, the susceptibility assessment model calculates the conditional probability distribution of the core feature set based on the Bayesian network and outputs the baseline risk assessment result. If the clinical feature sets are all common feature sets, the baseline risk assessment result is calculated based on the feature propagation algorithm of the graph neural network.

[0038] Specifically, the susceptibility assessment model is a dual-channel risk analysis model based on graph neural networks and Bayesian reasoning, which is specially designed for the individualized characteristics of adverse reactions to anti-tumor treatment. The model adopts a two-layer architecture: feature representation layer and risk reasoning layer.

[0039] The feature representation layer consists of two parallel branches: 1. Medical knowledge encoding branch: Map clinical feature data to a predefined medical knowledge graph. Each feature node contains three types of attribute information: basic feature value, time series change trend, and association rule strength. The basic feature value reflects the current state of the indicator, the time series change trend captures the dynamic characteristics of the indicator, and the association rule strength represents the interactive relationship with other features.

[0040] 2. Feature association analysis branch: Analyze the interaction patterns between features based on the feature association network. For the core feature set, the Bayesian network is used to calculate the conditional probability distribution and establish the causal relationship chain between features; for the common feature set, the combined representation of features is extracted through the message passing mechanism of the graph neural network.

[0041] The risk inference layer combines the outputs of the two branches to achieve accurate assessment of risk levels: 1. For the core feature set, the posterior probabilities of different adverse reactions are calculated through Bayesian reasoning, and verified and explained based on the medical knowledge graph.

[0042] 2. For common feature sets, the attention mechanism of graph neural networks is used to dynamically adjust the weights of feature nodes according to their importance in the knowledge graph.

[0043] Specifically, the feature association network is constructed through the following steps: grouping the clinical feature data according to the organ system classification of the WHO adverse reaction grading standard to obtain an initial set of clinical features, calculating the conditional mutual information and marginal mutual information within the initial set of clinical features, and generating nodes of the feature association network; determining the connection relationship based on the mutual information entropy between nodes, wherein, if the clinical features corresponding to two nodes belong to the same organ system, the conditional mutual information between the nodes is used as the connection weight; if the clinical features corresponding to the two nodes belong to different organ systems, a connection is established when the mutual information entropy between the nodes exceeds the average mutual information entropy within each organ system, and the excess value is used as the connection weight; the update of the feature association network is achieved by calculating the rate of change of the connection weight and iteratively propagating it between nodes.

[0044] Conditional mutual information is obtained by calculating the probabilistic dependency between the clinical feature set and the adverse reaction label, and marginal mutual information is the probabilistic dependency between a single feature in the clinical feature set and the adverse reaction label; the conditional probability distribution in the Bayesian network is obtained by the maximum likelihood estimation method, and the feature propagation algorithm of the graph neural network uses a message passing mechanism to update the node features of the feature association network; the mutual information entropy is calculated based on the co-variability of the clinical feature set and the physiological indicator data. This analysis method based on information theory and probabilistic graph models constructs a quantitative relationship between clinical feature data and physiological indicator data. For example, when evaluating the risk of bone marrow suppression of chemotherapy drugs, the mutual information entropy of genetic testing data and hematopoietic function indicators reflects the patient's susceptibility to bone marrow suppression; when predicting the cardiac toxicity of targeted therapy, the topological relationship between functional indicators and imaging features is captured by the feature propagation algorithm, combined with the co-variability of cardiac function indicators, to provide a more accurate risk assessment.

[0045] The present invention solves the technical problem that the risk warning model in the prior art is insufficient in analyzing the correlation of clinical features. The prior art usually uses a simple feature combination or linear model for risk assessment, which is difficult to capture complex feature interaction patterns. The present invention constructs a feature association network and designs a differentiated connection strategy based on organ system classification, so that the model can adaptively establish the correlation strength between features. For example, when evaluating chemotherapy-related cardiotoxicity, traditional methods often consider electrocardiogram changes or myocardial enzyme spectrum changes alone, while ignoring the interactive effects between these indicators and drug dosage and gene polymorphism. The feature association network of the present invention establishes a complete feature association evaluation system by calculating the conditional mutual information of features within the same organ system and the mutual information entropy of cross-system features. This mechanism shows unique advantages in practical applications: when it is detected that certain feature combinations have strong predictive power, the model will automatically increase the weight of the relevant features, thereby improving the accuracy of the warning. In particular, when predicting chronic cumulative toxicity, the present invention can capture the potential laws of the evolution of clinical features over time through the feature association network, which is difficult to achieve by simply relying on feature combinations or statistical analysis. In addition, the differential analysis strategy of the present invention (Bayesian network is used for the core feature set and graph neural network is used for the common feature set) achieves a good balance between computational efficiency and early warning timeliness, which is of great significance in clinical practice that requires rapid response.

[0046] The present invention realizes intelligent recognition of complex association patterns between clinical features through the differentiated connection strategy of feature association network and the hierarchical analysis mechanism based on organ system, which not only improves the accuracy of risk assessment, but also provides doctors with an interpretable basis for risk assessment. This method overcomes the limitations of feature analysis fragmentation and insufficient correlation analysis in the prior art, and plays an important role in individualized risk warning of anti-tumor treatment.

[0047] S3: Calculate the comprehensive risk score based on the short-term risk prediction results and the baseline risk assessment results. When the comprehensive risk score exceeds the warning threshold of the corresponding adverse reaction type, a warning signal is triggered.

[0048] S3.1: Establish a risk fusion matrix according to the organ system classification, and calculate the risk weights of different time scales based on the short-term risk prediction results and baseline risk assessment results.

[0049] The risk weight is adjusted by the time decay factor; if there is a predicted value of the key monitoring category in the short-term risk prediction result, the inverse of the exponential function of the predicted value is used as the time decay factor to update the diagonal elements of the risk fusion matrix; if there is an assessment value of the core feature set in the baseline risk assessment result, the inverse of the exponential function of the assessment value is used as the time decay factor to update the non-diagonal elements of the risk fusion matrix. Among them, the calculation period of the time decay factor is inversely proportional to the occurrence period of the corresponding adverse reaction.

[0050] S3.2: Apply risk weights to the risk fusion matrix and calculate the system risk score; if the short-term risk prediction result and the baseline risk assessment result correspond to the same organ system, the diagonal terms of the system risk score are calculated using the coupling coefficient; if they correspond to different organ systems, the off-diagonal terms of the system risk score are calculated using the risk transfer coefficient.

[0051] Among them, the system risk score is obtained by normalizing the risk fusion matrix through the eigenvalue decomposition method.

[0052] Specifically, the comprehensive risk score is calculated according to the following steps: 1. Construct a risk feature vector and organize the short-term risk prediction results and baseline risk assessment results into a matrix form according to the corresponding relationship between organ systems; 2. Calculate the eigenvalue and eigenvector of the risk eigenvector, and select the eigenvector corresponding to the maximum eigenvalue as the main risk direction; 3. The projection of the risk feature vector in the main risk direction is used as the benchmark value of the comprehensive risk score; 4. Modify the baseline value of the comprehensive risk score based on the coupling coefficient and risk transfer coefficient.

[0053] The coupling coefficient reflects the consistency between the short-term risk and the baseline risk in the same organ system, and is obtained by calculating the mutual information entropy between the risk prediction value and the assessment value; the risk transfer coefficient characterizes the risk propagation characteristics between different organ systems, and is obtained by conditional probability calculation after constructing a Bayesian network based on the temporal correlation pattern of organ function damage in historical case data. For example, when monitoring chemotherapy-related cardiorenal toxicity, if abnormal cardiac function indicators are detected, the model will predict the probability of renal function damage through the risk transfer coefficient, thereby achieving joint early warning of multiple organ systems. This risk scoring method based on feature decomposition not only ensures the interpretability of the scoring results, but also realizes the accurate quantification of complex system risks.

[0054] Further, S3.3: Establishing an adaptive warning threshold system based on historical clinical data, comprising the following steps: 1. Construct an early warning threshold matrix based on the organ system classification of the WHO adverse reaction grading standard. For key monitoring categories, determine the initial threshold based on the diagonal elements of the risk fusion matrix and the time attenuation factor; for routine monitoring categories, determine the initial threshold based on ROC curve analysis; 2. For each type of adverse reaction, the baseline parameters for threshold adjustment are calculated by combining the historical frequency of occurrence, the weighted score of severity, and the coupling coefficient of the corresponding organ system; for adverse reaction types with a core feature set, the threshold is further modified by the risk transfer coefficient; 3. The threshold matrix is ​​updated periodically based on the newly added clinical data. The update rules are as follows: if the false alarm rate of a certain type of adverse reaction in the most recent monitoring cycle exceeds the set value, the warning threshold of this type of adverse reaction is reduced by the time decay factor; if the false alarm rate of a certain type of adverse reaction in the most recent monitoring cycle exceeds the set value, the warning threshold of this type of adverse reaction is increased accordingly. The threshold update cycle is inversely proportional to the occurrence cycle of the corresponding adverse reaction.

[0055] S3.4: Design a hierarchical early warning signal triggering mechanism, including: 1. When the comprehensive risk score exceeds the warning threshold of the corresponding adverse reaction type by different degrees, different levels of warning signals are triggered: For key monitoring categories, when the diagonal item of the system risk score exceeds the warning threshold but does not reach 1.2 times, a level 1 warning is triggered; When the risk exceeds the warning threshold by 1.2 times but does not reach 1.5 times, or when the non-diagonal item of the risk score of an organ system with a risk transfer relationship exceeds the warning threshold, a secondary warning is triggered; When it exceeds the warning threshold by 1.5 times, a level 3 warning is triggered; 2. The information content contained in different levels of warning signals increases in sequence: The first-level warning includes risk scoring, abnormal indicator analysis and co-variability data; The second-level warning adds core feature set analysis results, risk transmission predictions, and preliminary intervention recommendations; The third-level warning further supplements the characteristic correlation network analysis results, detailed medical intervention plans and multi-department collaborative disposal recommendations.

[0056] 3. The early warning signal is sent in a multi-modal manner: system pop-up prompts, push notifications from medical staff’s mobile terminals, sound and light reminders from diagnosis and treatment workstations, etc., to ensure that medical staff can receive early warning information in a timely manner.

[0057] In summary, the present invention has the following advantages: First, by introducing the concept of collaborative variability and a dual-channel analysis mechanism, the present invention overcomes the limitation that the traditional single-indicator monitoring method is difficult to reflect the overall functional state of the organ system, realizes the accurate identification of the coordinated changes of multiple indicators, and improves the response speed to high-risk situations. Secondly, the present invention constructs a feature association network based on organ system classification, and through differentiated connection strategies and a two-layer architecture design, it realizes the intelligent identification of complex association patterns between clinical features, and solves the problem of insufficient feature interaction analysis in existing risk warning models. Finally, the present invention adopts a risk scoring method based on time attenuation, and by introducing coupling coefficients and risk transfer coefficients, it establishes a quantitative assessment mechanism for risk transmission among multiple organ systems, breaking through the technical bottleneck that traditional early warning systems are difficult to predict chain reactions.

[0058] Embodiment 2 is an embodiment of the present invention, which provides an anti-tumor treatment adverse reaction risk warning system, comprising: An acquisition module is used to obtain the patient's physiological index data and clinical characteristic data; A processing module is used to construct an immediate risk assessment model and a susceptibility assessment model, analyze the temporal variation characteristics of physiological indicator data through the immediate risk assessment model to obtain a short-term risk prediction result, and analyze the correlation characteristics of clinical characteristic data through the susceptibility assessment model to obtain a baseline risk assessment result; The early warning module is used to calculate the comprehensive risk score based on the short-term risk prediction results and the baseline risk assessment results. When the comprehensive risk score exceeds the early warning threshold of the corresponding adverse reaction type, the early warning signal is triggered.

[0059] Example 3, reference Figure 2, is an embodiment of the present invention, which is different from the previous embodiment in that: if the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0060] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0061] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0062] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0063] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for early warning of the risk of adverse reactions to anti-tumor treatment, characterized in that: include: Acquire the patient's physiological indicator data, build an immediate risk assessment model, analyze the temporal variation characteristics of the physiological indicator data through the immediate risk assessment model, and obtain a short-term risk prediction result; Acquiring clinical characteristic data of the patient, constructing a susceptibility assessment model, analyzing the correlation characteristics of the clinical characteristic data through the susceptibility assessment model, and obtaining a baseline risk assessment result; A comprehensive risk score is calculated based on the short-term risk prediction result and the baseline risk assessment result, and a warning signal is triggered when the comprehensive risk score exceeds a warning threshold of a corresponding adverse reaction type.

2. The method for early warning of risk of adverse reactions to anti-tumor therapy according to claim 1, characterized in that: The physiological index data include physical sign monitoring data, organ function index, immune function index, hematopoietic function index, and metabolic level index; Acquiring the patient's physiological indicator data, building an immediate risk assessment model, and analyzing the temporal variation characteristics of the physiological indicator data through the immediate risk assessment model to obtain a short-term risk prediction result, including the following steps: The physiological index data are grouped according to the organ system classification of the WHO adverse reaction grading standard, and the co-variability of each group of indicators is calculated; if the dispersion of a group of indicators exceeds the mean level of the group of indicators, and the change range of two or more indicators in the group of indicators exceeds the standard deviation range of their baseline values, then the group of indicators is determined as a key monitoring category, otherwise, the group of indicators is determined as a routine monitoring category; The physiological indicator data are input into the immediate risk assessment model. If there is an indicator group determined as a key monitoring category, the immediate risk assessment model will give priority to analyzing the time series correlation characteristics of each indicator determined as a key monitoring category and output the short-term risk prediction result. If the physiological indicator data all belong to the conventional monitoring category, the short-term risk prediction result will be output after analyzing the combined characteristics of all category indicators.

3. The method for early warning of risk of adverse reactions to anti-tumor therapy according to claim 2, characterized in that: The calculation of the co-variability comprises the following steps: Obtain the values ​​of each indicator in the same organ system within three consecutive sampling periods and calculate the standard deviation of each indicator; The mean of the standard deviations of all indicators was calculated, and the covariance was obtained by dividing the mean of the standard deviations by the overall mean of the indicators within the organ system; The time series correlation feature is obtained by calculating the mutual correlation coefficient between the indicators, wherein the synchronization is determined based on the mutual correlation coefficient of zero time delay, and the hysteresis is determined based on the time delay corresponding to the maximum mutual correlation coefficient; The combined features are obtained based on hierarchical clustering analysis between indicators of different organ systems, and a feature combination relationship matrix is ​​constructed by calculating the Euclidean distance between indicators; The instant risk assessment model adopts a dual-branch structure, which includes a time series analysis branch and a feature fusion branch; wherein the time series analysis branch adopts a multi-layer LSTM network structure, and each layer processes change patterns at different time scales; the feature fusion branch adopts an attention mechanism to adjust the weights of indicators of different organ systems.

4. The method for early warning of risk of adverse reactions to anti-tumor therapy according to claim 3, characterized in that: The clinical characteristic data include laboratory test data, imaging test data, genetic test data, medication regimen parameters, and treatment cycle indicators; Acquiring clinical characteristic data of the patient, building a susceptibility assessment model, analyzing the correlation characteristics of the clinical characteristic data through the susceptibility assessment model, and obtaining a baseline risk assessment result, including the following steps: A feature association network is constructed based on the clinical feature data, and the association strength between clinical features in the feature association network is calculated; wherein the association strength is the conditional mutual information of the clinical feature set; if the conditional mutual information of the clinical feature set is greater than the marginal mutual information of its constituent features, and the mutual information entropy between the organ system corresponding to the clinical feature set and the physiological indicator data is greater than the mean value of the mutual information entropy, then the clinical feature set is marked as a core feature set, otherwise it is marked as a common feature set; The feature association network is input into the susceptibility assessment model. If there is a clinical feature set marked as a core feature set, the susceptibility assessment model calculates the conditional probability distribution of the core feature set according to the Bayesian network and outputs the baseline risk assessment result. If the clinical feature sets are all common feature sets, the baseline risk assessment result is calculated based on the feature propagation algorithm of the graph neural network.

5. The method for early warning of risk of adverse reactions to anti-tumor therapy according to claim 4, characterized in that: The feature association network is constructed by the following steps: The clinical characteristic data are grouped according to the organ system classification of the WHO adverse reaction grading standard to obtain an initial set of clinical characteristics; Calculating the conditional mutual information and marginal mutual information within the initial set of clinical features to generate nodes of the feature association network; The connection relationship is determined based on the mutual information entropy between the nodes; if the clinical features corresponding to the two nodes belong to the same organ system, the conditional mutual information between the nodes is used as the connection weight; if the clinical features corresponding to the two nodes belong to different organ systems, a connection is established when the mutual information entropy between the nodes exceeds the average mutual information entropy within the respective organ systems, and the excess value is used as the connection weight; the updating of the feature association network is achieved by calculating the rate of change of the connection weight and iteratively propagating it between the nodes.

6. The method for early warning of risk of adverse reactions to anti-tumor therapy according to claim 5, characterized in that: Calculating a comprehensive risk score based on the short-term risk prediction results and the baseline risk assessment results includes: Establishing a risk fusion matrix according to the organ system classification, and calculating risk weights of different time scales based on the short-term risk prediction results and the baseline risk assessment results; Applying the risk weight to the risk fusion matrix to calculate a system risk score; if the short-term risk prediction result and the baseline risk assessment result correspond to the same organ system, the diagonal term of the system risk score is calculated by a coupling coefficient; if they correspond to different organ systems, the non-diagonal term of the system risk score is calculated by a risk transfer coefficient; The system risk score is obtained by normalizing the risk fusion matrix using an eigenvalue decomposition method; The comprehensive risk score is calculated as follows: Constructing a risk feature vector, organizing the short-term risk prediction result and the baseline risk assessment result into a matrix form according to the corresponding relationship between organ systems; Calculate the eigenvalue and eigenvector of the risk eigenvector, and select the eigenvector corresponding to the maximum eigenvalue as the main risk direction; Taking the projection of the risk feature vector in the main risk direction as the reference value of the comprehensive risk score; The reference value of the comprehensive risk score is corrected based on the coupling coefficient and the risk transfer coefficient.

7. The method for early warning of risk of adverse reactions to anti-tumor therapy according to claim 6, characterized in that: The risk weight is adjusted by a time decay factor; If there is a predicted value of the key monitoring category in the short-term risk prediction result, the inverse of the exponential function of the predicted value is used as the time decay factor to update the diagonal elements of the risk fusion matrix; If there is an evaluation value of the core feature set in the baseline risk evaluation result, the exponential function inverse of the evaluation value is used as the time decay factor to update the non-diagonal elements of the risk fusion matrix; The calculation period of the time attenuation factor is inversely proportional to the occurrence period of the corresponding adverse reaction.

8. An anti-tumor therapy adverse reaction risk warning system, based on the anti-tumor therapy adverse reaction risk warning method according to any one of claims 1 to 7, characterized in that: include, An acquisition module is used to obtain the patient's physiological index data and clinical characteristic data; A processing module, used to construct an immediate risk assessment model and a susceptibility assessment model, analyze the temporal variation characteristics of the physiological indicator data through the immediate risk assessment model to obtain a short-term risk prediction result, and analyze the correlation characteristics of the clinical characteristic data through the susceptibility assessment model to obtain a baseline risk assessment result; The early warning module is used to calculate a comprehensive risk score based on the short-term risk prediction result and the baseline risk assessment result, and trigger a warning signal when the comprehensive risk score exceeds the early warning threshold of the corresponding adverse reaction type.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the anti-tumor treatment adverse reaction risk warning method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the anti-tumor treatment adverse reaction risk warning method described in any one of claims 1 to 7 are implemented.

Images