Intelligent adverse drug reaction prediction and early warning system and method thereof
By establishing an intelligent prediction and early warning system based on hospital information module, using logistic regression and Bayesian network algorithms, combining individual patient data, monitoring symptoms after medication, setting risk levels and thresholds, the limitations of drug adverse reaction prediction and early warning in the existing technology are solved, and more accurate prediction and timely early warning are achieved.
Patent Information
- Application Number
- CN202510098482.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-22
- Publication Date
- 2025-05-23
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing intelligent prediction and early warning system for drug adverse reactions have limitations in actual use. For example, animal experimental results cannot fully predict human responses, and it is difficult to accurately predict the association of adverse reactions in individual patients.
By establishing an intelligent prediction and early warning system based on hospital information module, using logistic regression algorithm and Bayesian network algorithm, combining individual patients' medical records, medication and examination data, monitoring the symptoms after medication, setting a risk level and setting corresponding thresholds to issue early warnings.
The system can more accurately predict the probability of adverse drug reactions, and by setting risk levels and thresholds, promptly issuing early warnings, improving the predictive and early warning capabilities of adverse drug reactions.
Smart Images
Figure CN120032833A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of intelligent medical information processing, and in particular to an intelligent prediction and early warning system for adverse drug reactions and a method thereof. Background Art
[0002] The existing intelligent prediction and early warning systems and methods for adverse drug reactions still have the following drawbacks in actual use:
[0003] Adverse drug reactions, in a broad sense, include harmful reactions caused by drug quality problems or improper use of drugs, including drug side effects, toxic effects (toxic reactions), sequelae (aftereffects), allergic reactions, idiosyncratic reactions, superinfection caused by anti-infective drugs, dependence, and carcinogenic, teratogenic, and mutagenic effects. Some of these adverse reactions occur when the drug itself contains impurities or is used improperly, while some can occur when the quality inspection is qualified and the clinical usage and dosage are normal. Therefore, it is necessary to predict the adverse drug reactions. The methods mainly include preclinical pharmacology models, in vitro drug metabolism and transport studies, genomic and proteomic methods, and predictability studies of animal models. However, although these methods can help predict adverse drug reactions to a certain extent, there are still some challenges and limitations. For example, the results of animal experiments may not fully predict human reactions, because differences between species may lead to different results.
[0004] When drugs are tested for adverse reactions in individual patients, in order to accurately predict the correlation between adverse reactions between drugs and individual patients, intelligent prediction of adverse drug reactions is also required, as well as intelligent early warning of adverse drug reactions. Summary of the invention
[0005] The purpose of the present invention is to provide an intelligent prediction and early warning system for adverse drug reactions and a method thereof to solve the problems existing in the prior art.
[0006] In order to achieve the above-mentioned object, the present invention provides the following technical solutions: an intelligent prediction and early warning system for adverse drug reactions and a method thereof, comprising extracting medical record data of individual patients based on a hospital information module, the medical record data covering the basic information, past medical history and allergy history of individual patients, and obtaining individual patient information data;
[0007] Extract the medication data of individual patients, which covers the details of the current and past medications of individual patients, and obtain the medication data of individual patients;
[0008] Extract the examination data of individual patients. The examination data covers the blood routine, urine routine, liver and kidney function index, coagulation function and other laboratory test results collected regularly by individual patients. The examination data can dynamically reflect the changes in the patient's body function during the medication process, and obtain the examination data of individual patients;
[0009] Establishing an intelligent prediction module to monitor various symptoms of individual patients after taking medication, and combining the individual patient information data, the individual patient medication data and the individual patient examination data to obtain prediction data;
[0010] An intelligent early warning module is established to establish risk levels based on the probability of adverse reactions in the prediction model, and corresponding thresholds are set for the risk levels to obtain early warning detail data.
[0011] Furthermore, the intelligent prediction module is established to monitor various symptoms of individual patients after medication, and to obtain prediction data by combining the individual patient information data, the individual patient medication data and the individual patient examination data, specifically including:
[0012] Based on the logistic regression algorithm, the basic formula is: Where X = (X 1 ,X 2 ,…,X n ) represents individual patient information data, individual patient medication data and individual patient examination data, Y represents adverse reaction data, and Y has two values (usually 0 and 1), Y = 0 means no adverse drug reaction occurs, Y = 1 means adverse drug reaction occurs, P = (Y = 1 | X) is given by feature X = (X 1 ,X 2 ,…,X n ), the probability of adverse drug reactions occurring (Y = 1), β 0 ,β 1 ,…,β n are the parameters of the model.
[0013] Furthermore, the intelligent early warning module is established, and based on the probability of occurrence of adverse reactions of the prediction model, the risk level is established, and the corresponding threshold is set for the risk level to obtain early warning details data, specifically including:
[0014] Based on the Bayesian network algorithm, the basic formula is: Among them, A represents adverse drug reaction, B represents drug dosage, C represents patient age, P(B,C|A) represents the joint probability of drug dosage and patient age in the event of adverse reaction, P(A) represents the prior probability of adverse reaction, and P(B,C) is the joint probability of drug dosage and patient age.
[0015] Furthermore, the probability of occurrence of adverse reactions based on the prediction model, establishing a risk level, setting a corresponding threshold for the risk level, and obtaining early warning details data also include:
[0016] Risks are divided into low, medium and high levels, and corresponding risk thresholds are set for each level. When the predicted probability reaches or exceeds the corresponding threshold, an early warning is triggered and detailed early warning data is obtained.
[0017] The intelligent prediction and early warning method for adverse drug reactions includes a hospital information module, an intelligent prediction module, an intelligent early warning module and a central processing module, and the steps are as follows:
[0018] Step 1: Based on the hospital information module, obtain individual patient information data, individual patient medication data, and individual patient examination data;
[0019] Step 2: Based on the intelligent prediction module and the logistic regression algorithm, the prediction data is obtained;
[0020] Step 3: Based on the intelligent warning module and the Bayesian network algorithm, obtain the warning details data;
[0021] Step 4: Based on the central processing module, the warning details data are evaluated and analyzed.
[0022] Compared with the prior art, the drug adverse reaction intelligent prediction and early warning system and method provided by the present invention have the following beneficial effects:
[0023] The intelligent prediction and early warning system and method for adverse drug reactions can handle binary or multi-classification problems through logistic regression algorithm and Bayesian network algorithm. By establishing the relationship between features and whether adverse drug reactions occur (binary classification), when predicting whether a certain drug will cause specific adverse reactions, the patient's basic information, medication status, etc. can be used as input features, and the occurrence of adverse reactions can be used as output (yes / no). After training on a large amount of case data, a logistic regression model is obtained for the adverse reaction prediction of new patients. In the early warning of adverse drug reactions, nodes can represent drug usage, basic information of patients, and adverse reaction events. By constructing a Bayesian network, the probability of adverse reactions can be inferred based on known prior probabilities and conditional probabilities. If this probability exceeds a certain threshold, an early warning can be issued. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.
[0025] Figure 1 It is a schematic diagram of the system flow of the present invention;
[0026] Figure 2 It is a schematic diagram of the method flow of the present invention. DETAILED DESCRIPTION
[0027] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.
[0028] See also Figure 1 , 2 , the intelligent prediction and early warning system for adverse drug reactions and the method thereof, including extracting the medical record data of individual patients based on the hospital information module, the medical record data covering the basic information, past medical history and allergy history of individual patients, and obtaining the information data of individual patients;
[0029] Extract the medication data of individual patients, which covers the details of the current and past medications of individual patients, and obtain the medication data of individual patients;
[0030] Extract the examination data of individual patients. The examination data covers the blood routine, urine routine, liver and kidney function index, coagulation function and other laboratory test results collected regularly by individual patients. The examination data can dynamically reflect the changes in the patient's body function during the medication process, and obtain the examination data of individual patients;
[0031] Establish an intelligent prediction module to monitor the various symptoms of individual patients after taking medication, and obtain prediction data by combining individual patient information data, individual patient medication data and individual patient examination data;
[0032] An intelligent early warning module is established to establish risk levels based on the probability of adverse reactions in the prediction model, and corresponding thresholds are set for the risk levels to obtain early warning detail data.
[0033] Establish an intelligent prediction module to monitor the various symptoms of individual patients after taking medication, and combine individual patient information data, individual patient medication data and individual patient examination data to obtain prediction data, including:
[0034] Based on the logistic regression algorithm, the basic formula is: Where X = (X 1 ,X 2 ,…,X n ) represents individual patient information data, individual patient medication data and individual patient examination data, Y represents adverse reaction data, and Y has two values (usually 0 and 1), Y = 0 means no adverse drug reaction occurs, Y = 1 means adverse drug reaction occurs, P = (Y = 1 | X) is given by feature X = (X 1 ,X 2 ,…,Xn ), the probability of adverse drug reactions occurring (Y = 1), β 0 ,β 1 ,…,β n are the parameters of the model.
[0035] Establish an intelligent early warning module, set up risk levels based on the probability of adverse reactions in the prediction model, set corresponding thresholds for the risk levels, and obtain early warning details data, including:
[0036] Based on the Bayesian network algorithm, the basic formula is: Among them, A represents adverse drug reaction, B represents drug dosage, C represents patient age, P(B, C|A) represents the joint probability of drug dosage and patient age in the event of adverse reaction, P(A) represents the prior probability of adverse reaction, and P(B, C) is the joint probability of drug dosage and patient age.
[0037] Example 1
[0038] Construct a Bayesian network, containing the following nodes and relationships:
[0039] node:
[0040] A represents adverse drug reaction (values: yes, no),
[0041] B represents the drug dosage (values: low, medium, high),
[0042] C represents the patient's age (values: young, middle-aged, old);
[0043] Relationship: Drug dosage and patient age affect the probability of adverse drug reactions;
[0044] Determine the conditional probability table of each node through historical data statistics or expert experience:
[0045] 1. P(B, C|A) (the probability of an adverse drug reaction occurring given the drug dose and patient age)
[0046] B (drug dosage) C (patient age) P(A=Yes|B,C) P(A=None|B,C) Low young 0.1 0.9 Low middle aged 0.15 0.85 Low elderly 0.2 0.8 middle young 0.25 0.75 middle middle aged 0.3 0.7 middle elderly 0.35 0.65 high young 0.4 0.6 high middle aged 0.45 0.55 high elderly 0.5 0.5
[0047] 2. P(B) (prior probability of drug dosage)
[0048] B (drug dosage) P(B) Low 0.3 middle 0.5 high 0.2
[0049] 3. P(C) (prior probability of patient age)
[0050] C (patient age) P(C) young 0.4 middle aged 0.3 elderly 0.3
[0051] Suppose there is a middle-aged patient who uses a medium dose of drugs. Calculate the probability of adverse drug reactions in this patient. According to the chain rule and conditional independence assumption of the Bayesian network, we can calculate:
[0052]
[0053] =0.21
[0054] Therefore, the probability that this middle-aged patient will experience an adverse reaction after using a moderate dose of the drug is 0.09, and the probability that no adverse reaction will occur is 0.21;
[0055] If we set a warning threshold, for example, a warning is issued when the probability of an adverse reaction is greater than 0.1, then in this case the system will issue a warning of adverse drug reactions.
[0056] Based on the probability of adverse reactions in the prediction model, the risk level is established, and the corresponding threshold is set for the risk level to obtain the warning details data, including:
[0057] Risks are divided into low, medium and high levels, and corresponding risk thresholds are set for each level. When the predicted probability reaches or exceeds the corresponding threshold, an early warning is triggered and detailed early warning data is obtained.
[0058] The intelligent prediction and early warning method for adverse drug reactions includes a hospital information module, an intelligent prediction module, an intelligent early warning module and a central processing module, and the steps are as follows:
[0059] Step 1: Based on the hospital information module, obtain individual patient information data, individual patient medication data, and individual patient examination data;
[0060] Step 2: Based on the intelligent prediction module and the logistic regression algorithm, the prediction data is obtained;
[0061] Step 3: Based on the intelligent warning module and the Bayesian network algorithm, obtain the warning details data;
[0062] Step 4: Based on the central processing module, the warning details data are evaluated and analyzed.
[0063] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.
Claims
1. Intelligent prediction and early warning system for adverse drug reactions, characterized by: Including extracting individual patient's medical record data based on the hospital information module, the medical record data covers the individual patient's basic information, past medical history and allergy history, etc., to obtain individual patient information data; Extract the medication data of individual patients, which covers the details of the current and past medications of individual patients, and obtain the medication data of individual patients; Extract the examination data of individual patients. The examination data covers the blood routine, urine routine, liver and kidney function index, coagulation function and other laboratory test results collected regularly by individual patients. The examination data can dynamically reflect the changes in the patient's body function during the medication process, and obtain the examination data of individual patients; Establishing an intelligent prediction module to monitor various symptoms of individual patients after taking medication, and combining the individual patient information data, the individual patient medication data and the individual patient examination data to obtain prediction data; An intelligent early warning module is established to establish risk levels based on the probability of adverse reactions in the prediction model, and corresponding thresholds are set for the risk levels to obtain early warning detail data.
2. The intelligent prediction and early warning system for adverse drug reactions according to claim 1, characterized in that: The intelligent prediction module is established to monitor various symptoms of individual patients after medication, and to obtain prediction data by combining the individual patient information data, the individual patient medication data and the individual patient examination data, specifically including: Based on the logistic regression algorithm, the basic formula is: Where X=(X1,X2,…,X n ) represents individual patient information data, individual patient medication data and individual patient examination data, Y represents adverse reaction data, and Y has two values (usually 0 and 1), Y = 0 means no adverse drug reaction occurs, Y = 1 means adverse drug reaction occurs, P = (Y = 1 | X) is given by feature X = (X1, X2, ..., X n ), the probability of adverse drug reactions occurring (Y=1), β0, β1,…, β n are the parameters of the model.
3. The intelligent prediction and early warning system for adverse drug reactions according to claim 1, characterized in that: The intelligent early warning module is established to establish a risk level based on the probability of occurrence of adverse reactions in the prediction model, and to set a corresponding threshold for the risk level to obtain early warning details data, specifically including: Based on the Bayesian network algorithm, the basic formula is: Among them, A represents adverse drug reaction, B represents drug dosage, C represents patient age, P(B,C|A) represents the joint probability of drug dosage and patient age in the event of adverse reaction, P(A) represents the prior probability of adverse reaction, and P(B,C) is the joint probability of drug dosage and patient age.
4. The intelligent prediction and early warning system for adverse drug reactions according to claim 3, characterized in that: The probability of occurrence of adverse reactions based on the prediction model, establishing a risk level, setting a corresponding threshold for the risk level, and obtaining early warning details data also include: Risks are divided into low, medium and high levels, and corresponding risk thresholds are set for each level. When the predicted probability reaches or exceeds the corresponding threshold, an early warning is triggered and detailed early warning data is obtained.
5. Intelligent prediction and early warning method for adverse drug reactions, characterized by It includes hospital information module, intelligent prediction module, intelligent warning module and central processing module. The steps are as follows: Step 1: Based on the hospital information module, obtain individual patient information data, individual patient medication data, and individual patient examination data; Step 2: Based on the intelligent prediction module and the logistic regression algorithm, the prediction data is obtained; Step 3: Based on the intelligent warning module and the Bayesian network algorithm, obtain the warning details data; Step 4: Based on the central processing module, the warning details data are evaluated and analyzed.