Risk detection method and system for adverse event in drug clinical trial

Abstract

A risk detection method and system for an adverse event in a drug clinical trial. The method comprises: constructing a clinical biological model on the basis of a physical health status, physiological parameters, and a historical examination report of a subject; recording investigational drug information administered to the subject, and generating an investigational drug database; by means of a physiological monitoring device and a health monitoring system, performing real-time monitoring on physiological function parameters and biological indicators of the subject after administration of the investigational drug, to obtain investigational drug data; screening and optimizing the investigational drug data on the basis of a genetic algorithm to obtain investigational drug response characteristics; determining a risk score on the basis of the investigational drug database and the investigational drug response characteristics; and determining a different risk level on the basis of the risk score, and performing risk warning on the basis of the risk level. By screening drug response characteristics and calculating risk scores, the method improves the efficiency of adverse event monitoring, enhances the data analysis capability, and ensures the scientific nature and accuracy of results.

Description

A method and system for detecting adverse events in drug clinical trials Technical Field

[0001] This invention relates to the field of medical technology, and in particular to a method and system for risk detection of adverse events in drug clinical trials. Background Technology

[0002] Adverse events (AEs) refer to all adverse medical events that occur after a subject receives an investigational drug. These events can manifest as symptoms, signs, illnesses, or abnormal laboratory tests. AEs are one of the indicators for evaluating the safety of drug clinical trials. Whether the duration, severity, and relationship to the investigational drug and its dosage of AEs are correctly collected and assessed directly affects the safety evaluation of the investigational drug, and consequently, the scientific validity and reliability of the trial results, as well as the safety of the drug after it is marketed. With the continuous strengthening of national supervision of drug clinical trials and the drug review and approval system, the level of drug clinical trials in my country has greatly improved. However, many problems still exist in the collection and assessment of AEs, affecting the quality of drug clinical trials and the safety of subjects.

[0003] Currently, many clinical trials employ traditional monitoring methods, collecting participants' health data through regular physical examinations and questionnaires. While these methods provide some information, they often suffer from problems such as delayed data collection, low processing efficiency, neglect of individual differences, and inaccurate risk assessment. Furthermore, they fail to adequately consider individual differences among participants, resulting in risk assessment results that lack specificity and accuracy. Therefore, designing a risk detection method and system for adverse events in drug clinical trials is essential. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for risk detection of adverse events in drug clinical trials, which improves the efficiency of adverse event monitoring, enhances data analysis capabilities, and ensures the scientific validity and accuracy of the results by screening drug response characteristics and calculating risk scores.

[0005] To achieve the above objectives, the present invention provides the following solution:

[0006] A method for detecting adverse events in drug clinical trials includes the following steps:

[0007] A clinical biological model was constructed based on the subjects' physical health status, physical parameters, and historical examination reports.

[0008] Record the drug information received by the subjects and generate a drug database;

[0009] Real-time monitoring of the subjects' physical function parameters and biological indicators after drug trials is conducted using physiological monitoring equipment and health monitoring systems to obtain drug trial data.

[0010] Genetic algorithms are used to screen and optimize drug test data to obtain drug response characteristics;

[0011] Risk scores are determined based on drug databases and drug response characteristics;

[0012] Different risk levels are determined based on risk scores, and risk warnings are issued based on these risk levels.

[0013] Optionally, a clinical biological model may be constructed based on the subject's health status, physical parameters, and historical examination reports, including:

[0014] Collect body parameters; body parameters include: height, weight, body mass index, body temperature, blood data, and metabolic indicators;

[0015] Physical health status is obtained through physical examination; physical health status is obtained by physicians through a comprehensive assessment of the subject's physical examination report, lifestyle habits, and mental health.

[0016] Clinical biological models were constructed using regression analysis based on physical health status, physical parameters, and historical examination reports.

[0017] Optionally, the drug database includes: drug name, dosage, route of administration, and time of administration; the route of administration includes: oral, injection, and topical; the time of administration includes: the time point of the drug trial and the time interval between the drug trial time points.

[0018] Optionally, the subjects' physical function parameters and biological indicators are monitored in real time using physiological monitoring equipment and a health monitoring system to obtain drug trial data. Specifically, after the drug trial, the subjects wear physiological monitoring equipment, and the physiological monitoring equipment collects data on the subjects' physical function parameters and biological indicators. The collected data is then integrated through the health monitoring system to obtain drug trial data.

[0019] Optionally, the drug test data can be screened and optimized based on a genetic algorithm to obtain drug response characteristics, including:

[0020] Preprocessing operations are performed on the trial data to obtain preprocessed data; the preprocessing operations include: data cleaning, outlier detection, data standardization, and feature selection.

[0021] A set of individuals is randomly generated, with each individual representing a preprocessed data point;

[0022] Individual characteristics are evaluated based on the fitness function;

[0023] Individuals with fitness higher than a preset threshold are selected as parents based on the tournament selection method;

[0024] New individuals are obtained by genetic recombination of the parent generation through multi-point crossover and gene alteration.

[0025] The new individuals are merged with the existing individuals to obtain an updated population;

[0026] The population is iteratively updated using a fitness function until the fitness converges to a preset convergence threshold, thus obtaining the selection features.

[0027] By removing influencing factors from the screening characteristics through correlation analysis and LASSO regression, the drug response characteristics were obtained.

[0028] Optionally, the outlier detection steps include:

[0029] Calculate the first and third quartiles of the data from different drug trials;

[0030] The interquartile range is obtained from the first quartile and the third quartile;

[0031] The upper and lower boundaries of the anomaly are determined based on the interquartile range.

[0032] Drug test data that exceeds the upper or lower boundary of an anomaly are defined as suspected outliers;

[0033] The suspected abnormal values ​​are confirmed by a physician, and the actual abnormal values ​​are obtained.

[0034] Optionally, the fitness function can be expressed as: Where F(x) is the fitness value of individual x, E(x) is the efficacy score of individual x, S(x) is the side effect score of individual x, C(x) is the individual difference score of individual x, D(x) is the tolerance score of individual x, α is the adjustment coefficient, ω1, ω2 and ω3 are all weighting coefficients, and ω1+ω2+ω3=1.

[0035] Optionally, the risk score can be calculated using the following formula: Where Q is the risk score, and A is the risk rating. n Here is the nth physical data point of the subject, T is the end time of observation, P(t) is the subject's status score at time t, R is the drug risk coefficient, and M... m Let m be the characteristic of the drug.

[0036] Optionally, different risk levels can be defined based on risk scores, and risk warnings can be issued based on these risk levels, including:

[0037] When the risk score is less than 5, the risk level is determined to be the first risk level;

[0038] When the risk score is greater than or equal to 5 and less than 10, the risk level is determined to be the second risk level.

[0039] When the risk score is greater than or equal to 10 and less than 15, the risk level is determined to be the third risk level.

[0040] When the risk score is greater than or equal to 15 and less than 20, the risk level is determined to be the fourth risk level.

[0041] When the risk score is greater than or equal to 20, the risk level is determined to be the fifth risk level.

[0042] A risk detection system for adverse events in drug clinical trials includes:

[0043] The human information module is used to construct clinical biological models based on the subject's physical health status, physical parameters, and historical examination reports.

[0044] The drug trial information module is used to record the drug trial information received by the subjects and generate a drug trial database;

[0045] The monitoring module is used to monitor the physical function parameters and biomarkers of the subjects after drug testing in real time based on physiological monitoring equipment and health monitoring system, and to obtain drug testing data;

[0046] The AE recognition module is used to screen and optimize drug test data based on genetic algorithms to obtain drug response characteristics;

[0047] The AE determination module is used to determine the risk score based on the drug database and drug response characteristics.

[0048] The AE early warning module is used to classify different risk levels based on risk scores and to issue risk warnings based on the risk levels.

[0049] According to specific embodiments provided by the present invention, the following technical effects are disclosed: The method for risk detection of adverse events in drug clinical trials provided by the present invention includes: constructing a clinical biological model based on the subject's physical health status, physical parameters, and historical examination reports; recording the drug information received by the subject and generating a drug database; real-time monitoring of the subject's physical function parameters and biological indicators after drug testing using physiological monitoring equipment and a health monitoring system to obtain drug data; filtering and optimizing the drug data based on a genetic algorithm to obtain drug response characteristics; determining a risk score based on the drug database and drug response characteristics; classifying different risk levels based on the risk score and issuing risk warnings based on the risk levels. The present invention significantly improves the efficiency of adverse event monitoring, can promptly detect potential risks, enhances data analysis capabilities, and ensures the scientific validity and accuracy of the results. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 is a flowchart of the adverse event risk detection method of the present invention;

[0052] Figure 2 is a flowchart of the clinical biological model construction process of the present invention;

[0053] Figure 3 is a flowchart of the screening test data of the present invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0056] As shown in Figure 1, this invention provides a method for risk detection of adverse events in drug clinical trials, comprising the following steps:

[0057] Step 100: Construct a clinical biological model based on the subject's health status, physical parameters, and historical examination reports; the specific steps are shown in Figure 2, including:

[0058] Step 101: Collect body parameters; body parameters include: height, weight, body mass index, body temperature, blood data, and metabolic indicators;

[0059] Specifically, this embodiment uses a height measuring device to measure precise height data, a weighing scale to measure weight, a body composition analyzer to calculate body mass index (BMI) and analyze body composition, such as body fat percentage and muscle mass, and a digital thermometer to quickly and accurately measure body temperature. A blood analyzer is also used to comprehensively analyze blood samples, detecting complete blood count and biochemical indicators such as blood glucose and cholesterol, and a metabolic monitoring device is used to measure the subject's metabolic rate and other physiological indicators.

[0060] Step 102: Obtain physical health status through physical examination; physical health status is obtained by physicians through a comprehensive assessment of the subject's physical examination report, lifestyle habits, and mental health;

[0061] Specifically, physicians will carefully review the physical examination report, focusing on the subject's key health indicators. They will also collect information on the subject's lifestyle habits, dietary habits, and exercise frequency through questionnaires or one-on-one interviews. Regarding mental health, standardized psychological assessment tools, such as the Personality Depression Scale (PHQ-9) and the Self-Rating Anxiety Scale (GAD-7), will be used to assess the subject's mental health status. Finally, all this information will be compiled to form an overview of the subject's health status, followed by a comprehensive evaluation.

[0062] Step 103: Based on physical health status, physical parameters, and historical examination reports, construct a clinical biological model using regression analysis.

[0063] Specifically, this embodiment uses ridge regression to construct a clinical biological model. Ridge regression is an improvement on linear regression. The core idea of ​​linear regression is to assume a linear relationship between the dependent variable and one or more independent variables, and to fit a straight line (or hyperplane) using the least squares method to minimize the difference between the predicted and true values ​​(sum of squared residuals). Ridge regression, by adding an L2 regularization term, limits the complexity of the model and prevents overfitting. It also adjusts the size of the penalized regression coefficients to make the parameter estimates more stable, making it more suitable for large datasets with highly correlated features. Finally, visualization tools (such as Matplotlib or Seaborn) are used to display the results, providing a more intuitive understanding of the data and model output. If the model's performance is unsatisfactory, the accuracy and effectiveness of the model can be improved by adjusting the parameters or selecting different input features.

[0064] Step 200: Record the drug information received by the subjects and generate a drug database;

[0065] Specifically, the drug trial database includes: drug name, dosage, route of administration, and time of administration; the route of administration includes: oral, injection, and topical; the time of administration includes: the time point of the trial and the time interval between the time points of the trial.

[0066] Furthermore, this embodiment uses a relational database management system (such as MySQL or PostgreSQL) to construct a trial drug database based on drug name, dosage, route of administration, and administration time. This trial drug database supports queries by subject ID, drug name, administration time, and other criteria, facilitating quick access to relevant information by medical personnel. It also provides a data export function, allowing trial drug information to be exported as CSV or Excel documents for further analysis and report generation.

[0067] It should be noted that recording the subjects' drug trial information and generating a drug trial database enables systematic and standardized management. This not only helps the clinical trials proceed smoothly, but also provides important data support for subsequent drug efficacy evaluation, thereby ensuring the accuracy and completeness of the data.

[0068] Step 300: Real-time monitoring of the subjects' physical function parameters and bioindicators after drug testing using physiological monitoring equipment and health monitoring systems to obtain drug testing data;

[0069] Specifically, after the drug trial, the subjects wear physiological monitoring devices to collect data on their physical function parameters and biological indicators. The collected data are then integrated through a health monitoring system to obtain the drug trial data.

[0070] Furthermore, in this embodiment, the Huawei Band 6 is used as a heart rate, step count, and sleep monitoring device; the Canon HEM-7121 blood pressure monitor is used to monitor the subject's blood pressure; the Ora Ring 2 is used to track the subject's sleep and activity; and the Dexcom G6 continuous glucose monitor is used to monitor the subject's blood glucose level in real time.

[0071] Furthermore, the health monitoring system in this embodiment aggregates data from various monitoring devices via cloud technology. First, the data is cleaned and standardized to ensure comparability across different data types within the same timescale. Then, the collected physiological parameters, such as heart rate, blood pressure, blood sugar, and sleep cycles, are stored in a secure database through a data integration platform for easy subsequent analysis. Finally, various charts and reports are generated to display the subject's physiological trends and changes in biometrics.

[0072] It should be noted that the complete data collection and integration process not only provides information on changes in the subjects' physical functions after drug trials, but also enables real-time monitoring and analysis of drug effects and adverse reactions, providing important evidence for further clinical decision-making and research.

[0073] Step 400: Optimize and filter the drug test data using a genetic algorithm to obtain drug response characteristics; the specific steps are shown in Figure 3, including:

[0074] Step 401: Perform preprocessing operations on the drug trial data to obtain preprocessed data; the preprocessing operations include: data cleaning, outlier detection, data standardization, and feature selection;

[0075] Specifically, the steps for outlier detection include:

[0076] Calculate the first quartile Q1 and the third quartile Q3 of different drug trial data. Quartiles are calculated by sorting the data in ascending order and finding the corresponding positions, where Q1 is the value at the 25th percentile of the sorted data, and Q3 is the value at the 75th percentile of the sorted data.

[0077] The interquartile range (IQR) is obtained from the first and third quartiles; the formula for calculating the interquartile range is: IQR = Q3 - Q1;

[0078] The upper boundary QS and lower boundary QX of the anomaly are determined based on the interquartile range; the calculation formulas for QS and QX are respectively: QX=Q3+1.5*IQR; QX=Q1-1.5*IQR;

[0079] Drug test data that exceeds the upper or lower boundary of an anomaly are defined as suspected outliers;

[0080] The suspected abnormal values ​​are confirmed by a physician, and the actual abnormal values ​​are obtained.

[0081] Step 402: Randomly generate a set of individuals, each representing a preprocessed data; these individuals can be seen as encapsulations of different feature subsets, and the random generation process ensures the diversity of individuals.

[0082] Step 403: Evaluate the characteristics of individuals based on the fitness function;

[0083] Specifically, the expression for the fitness function is:

[0084] Where F(x) is the fitness value of individual x, E(x) is the efficacy score of individual x, S(x) is the side effect score of individual x, C(x) is the individual difference score of individual x, D(x) is the tolerance score of individual x, α is the adjustment coefficient, ω1, ω2 and ω3 are all weighting coefficients, and ω1+ω2+ω3=1.

[0085] Furthermore, the side effect score represents the severity of adverse drug reactions. Lower side effect scores are converted to values ​​closer to 1 through exponential mapping, thereby reducing the negative impact on fitness. The individual difference score represents individual physiological differences (such as age, gender, weight, etc.), which may affect drug metabolism and response. The tolerability score reflects the subject's tolerance to the drug; better tolerability improves fitness. Different weighting coefficients need to be adjusted according to the actual research objectives; this embodiment does not impose specific limitations. The adjustment coefficient is used to control the degree of influence of the side effect score on fitness; the larger the α, the more significant the impact of side effects on fitness.

[0086] Furthermore, the fitness function smooths the side effect scores through an exponential mapping, ensuring that higher side effect scores significantly inhibit fitness, while lower scores have a weaker impact on fitness, effectively reducing fitness in individuals with high side effect rates. By adding individual difference and tolerability scores to account for the variability in individual responses to the same drug, the practicality and adaptability of the fitness function are further improved. Dynamically adjustable weighting coefficients allow for dynamic adjustments to efficacy, side effects, individual differences, and tolerability based on specific subjects and drug characteristics, ensuring that the fitness function accurately reflects drug performance.

[0087] Step 404: Select individuals with fitness higher than a preset threshold as parents based on tournament selection;

[0088] Specifically, the tournament selection method involves randomly selecting several individuals for comparison and choosing the one with the highest fitness as the parent of the next generation, thus preserving the genetic traits of the most fit individuals.

[0089] Step 405: Gene recombination of the parent generation is performed through multi-point crossover and gene alteration to obtain new individuals;

[0090] Specifically, multi-point crossover refers to randomly selecting multiple crossover points among the parents and exchanging their characteristics to generate new offspring, while gene alteration involves randomly changing some characteristics. Such operations can introduce variation and increase the diversity of the population.

[0091] Step 406: Merge the new individuals with the existing individuals to obtain an updated population; this ensures that the best individuals in the current population are preserved while introducing new individuals, thus maintaining the population's evolutionary potential.

[0092] Step 407: Iteratively update the population using the fitness function until the fitness converges to the preset convergence threshold to obtain the selection features;

[0093] Step 408: Remove influencing factors from the screening features through correlation analysis and LASSO regression to obtain the drug response characteristics.

[0094] Specifically, correlation analysis determines which features have a weak relationship with the target variable by calculating the correlation coefficients between each feature and the target variable (such as the Pearson correlation coefficient), thereby identifying redundant features. LASSO regression is a regularized linear regression method that, by adding an L1 penalty term, causes the coefficients of certain features to drop to zero, effectively selecting important features and removing influencing factors.

[0095] It should be noted that the drug response features optimized using genetic algorithms will have higher reliability and validity. This not only improves the scientific rigor of feature selection but also reduces data dimensionality, lowers model complexity, and enhances the model's generalization ability to new data. Furthermore, it significantly improves the efficiency and accuracy of drug trial data analysis.

[0096] Step 500: Determine the risk score based on the drug database and drug response characteristics;

[0097] Specifically, the formula for calculating the risk score is as follows:

[0098] Where Q is the risk score, and A is the risk rating. n Here is the nth physical data point of the subject, T is the end time of observation, P(t) is the subject's status score at time t, R is the drug risk coefficient, and M... m Let m be the characteristic of the drug.

[0099] Furthermore, exponential operations on physical data represent the non-linear impact of physical data on risk. The definite integral of the natural logarithm of the subject's status score reflects the subject's dynamic health status. Multiplication operations represent the complex relationship between different drug characteristics and adverse events (AEs).

[0100] It should be noted that risk assessment based on drug trial databases and drug response characteristics can effectively integrate and analyze the subjects' physical condition and drug response characteristics, thereby ensuring the scientific rigor and accuracy of the risk assessment. Quantitative formulas are used to represent the impact of different factors on subject risk, making clinical decision-making more explicit and actionable. A dynamic monitoring mechanism is introduced, continuously updating subject status assessments to reflect changes in individual risk in a timely manner, enhancing the foresight and effectiveness of risk management. Furthermore, mathematical models are used to systematically evaluate the nonlinear relationships between risk factors, improving the model's adaptability and interpretability, thereby strengthening the predictive ability for potential risks.

[0101] Step 600: Divide the risk into different levels based on the risk score, and issue risk warnings based on the risk levels. Specifically, this includes:

[0102] When the risk score is less than 5, the risk level is determined to be the first risk level; the first risk level is mild AE, which is manifested as mild symptoms that do not affect the subject's daily activities;

[0103] When the risk score is greater than or equal to 5 and less than 10, the risk level is determined to be the second risk level; the second risk level is moderate AE, which is characterized by obvious symptoms, but the impact is limited and requires continued observation.

[0104] When the risk score is greater than or equal to 10 and less than 15, the risk level is determined to be the third risk level; the third risk level is severe AE, which is manifested by symptoms that seriously affect the subject's life and requires immediate treatment;

[0105] When the risk score is greater than or equal to 15 and less than 20, the risk level is determined to be the fourth risk level; the fourth risk level is life-threatening AE, which manifests as serious adverse reactions that may endanger life and require emergency intervention.

[0106] When the risk score is greater than or equal to 20, the risk level is determined to be the fifth risk level; the fifth risk level is death AE, which is manifested as the patient's death due to an adverse event.

[0107] This invention also provides a risk detection system for adverse events in drug clinical trials, comprising:

[0108] The human information module is used to construct clinical biological models based on the subject's physical health status, physical parameters, and historical examination reports.

[0109] The drug trial information module is used to record the drug trial information received by the subjects and generate a drug trial database;

[0110] The monitoring module is used to monitor the physical function parameters and biomarkers of the subjects after drug testing in real time based on physiological monitoring equipment and health monitoring system, and to obtain drug testing data;

[0111] The AE recognition module is used to screen and optimize drug test data based on genetic algorithms to obtain drug response characteristics;

[0112] The AE determination module is used to determine the risk score based on the drug database and drug response characteristics.

[0113] The AE early warning module is used to classify different risk levels based on risk scores and to issue risk warnings based on the risk levels.

[0114] The beneficial effects of this invention are as follows:

[0115] 1) By using genetic algorithms and correlation analysis, significant drug response characteristics were effectively extracted, thereby simplifying the data, improving model performance, and enhancing the accuracy of clinical decision-making.

[0116] 2) By classifying risks in detail, medical staff can take corresponding measures for different levels of risk, which improves the timeliness and effectiveness of clinical response and ensures the safety of subjects.

[0117] 3) By integrating specific technologies for data analysis and risk assessment, the clinical research process has been optimized, and research efficiency and data processing speed have been improved.

[0118] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0119] Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. Furthermore, those skilled in the art will recognize that, based on the ideas of this invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for detecting risk of adverse events in a drug clinical trial, characterized by, Includes the following steps: A clinical biological model was constructed based on the subjects' physical health status, physical parameters, and historical examination reports. Record the drug trial information received by the subjects and generate a drug trial database; The subjects' physical function parameters and biological indicators were monitored in real time using physiological monitoring equipment and health monitoring systems to obtain drug trial data. The drug test data were screened and optimized using a genetic algorithm to obtain drug response characteristics; Risk scores are determined based on the drug database and the drug response characteristics. Different risk levels are determined based on the risk score, and risk warnings are issued based on the risk levels.

2. The method of claim 1, wherein the method is a method of detecting risk of adverse events in a drug clinical trial. A clinical biological model was constructed based on the subjects' health status, physical parameters, and historical medical reports, including: Collect the body parameters; the body parameters include: height, weight, body mass index, body temperature, blood data, and metabolic indicators; The physical health status is obtained through a physical examination; the physical health status is obtained by a physician through a comprehensive assessment of the subject's physical examination report, lifestyle habits, and mental health. The clinical biological model is constructed using regression analysis based on the physical health status, the physical parameters, and the historical examination reports.

3. The method for detecting the risk of adverse events in drug clinical trials according to claim 1, characterized in that, The drug database includes: drug name, dosage, route of administration, and time of administration; the route of administration includes: oral, injection, and topical application; the time of administration includes: the time point of the drug test and the time interval between the time points of the drug test.

4. The method for risk detection of adverse events in drug clinical trials according to claim 1, characterized in that, The subjects' physical function parameters and biological indicators are monitored in real time by physiological monitoring equipment and health monitoring system to obtain drug trial data. Specifically, the subjects wear the physiological monitoring equipment after the drug trial, and the physiological monitoring equipment collects data on the subjects' physical function parameters and biological indicators. The collected data is then integrated through the health monitoring system to obtain the drug trial data.

5. The method for risk detection of adverse events in drug clinical trials according to claim 1, characterized in that, The drug test data is screened and optimized based on a genetic algorithm to obtain drug response characteristics, including: The drug trial data is preprocessed to obtain preprocessed data; the preprocessing operations include: data cleaning, outlier detection, data standardization, and feature selection. A set of individuals is randomly generated, and each individual represents one piece of preprocessed data. The individual's characteristics are evaluated based on the fitness function; Individuals with fitness higher than a preset threshold are selected as parents based on the tournament selection method; New individuals are obtained by genetic recombination of the parent generation through multi-point crossover and gene alteration. The new individual is merged with the existing individual to obtain an updated population; The population is iteratively updated using the fitness function until the fitness converges to a preset convergence threshold, thereby obtaining the selection features. The drug response characteristics were obtained by removing influencing factors from the screening features through correlation analysis and LASSO regression.

6. The method for risk detection of adverse events in drug clinical trials according to claim 5, characterized in that, The outlier detection steps include: Calculate the first and third quartiles of the different drug test data; The interquartile range is obtained based on the first quartile and the third quartile; The upper and lower boundaries of the anomaly are determined based on the interquartile range. The test data that exceeds the upper or lower boundary of the anomaly is defined as a suspected anomaly. The suspected abnormal values ​​were confirmed by a physician, and the actual abnormal values ​​were obtained.

7. The method for risk detection of adverse events in drug clinical trials according to claim 5, characterized in that, The expression for the fitness function is: Where F(x) is the fitness value of individual x, E(x) is the efficacy score of individual x, S(x) is the side effect score of individual x, C(x) is the individual difference score of individual x, D(x) is the tolerance score of individual x, α is the adjustment coefficient, ω1, ω2 and ω3 are all weighting coefficients, and ω1+ω2+ω3=1.

8. The method for risk detection of adverse events in drug clinical trials according to claim 1, characterized in that, The formula for calculating the risk score is as follows: Where Q is the risk score, and A n Here is the nth physical data point of the subject, T is the end time of observation, P(t) is the subject's status score at time t, R is the drug risk coefficient, and M... m Let m be the characteristic of the drug.

9. The method for risk detection of adverse events in drug clinical trials according to claim 1, characterized in that, Different risk levels are assigned based on the risk score, and risk warnings are issued based on the risk levels, including: When the risk score is less than 5, the risk level is determined to be the first risk level; When the risk score is greater than or equal to 5 and less than 10, the risk level is determined to be the second risk level; When the risk score is greater than or equal to 10 and less than 15, the risk level is determined to be the third risk level; When the risk score is greater than or equal to 15 and less than 20, the risk level is determined to be the fourth risk level. When the risk score is greater than or equal to 20, the risk level is determined to be the fifth risk level.

10. A risk detection system for adverse events in drug clinical trials, characterized in that, include: The human information module is used to construct clinical biological models based on the subject's physical health status, physical parameters, and historical examination reports. The drug testing information module is used to record the drug testing information received by the subjects and generate a drug testing database; The monitoring module is used to monitor the physical function parameters and bioindicators of the subjects after drug testing in real time based on physiological monitoring equipment and health monitoring system, and to obtain drug testing data; The AE recognition module is used to screen and optimize the test drug data based on a genetic algorithm to obtain test drug response characteristics; The AE determination module is used to determine a risk score based on the drug database and the drug response characteristics; The AE early warning module is used to classify different risk levels based on the risk score and to issue risk warnings based on the risk levels.

Images