Adverse drug reaction monitoring, management and analysis method and device
Through smartphone applications and wearable devices, patients' medication information and health data are collected in real time, and machine learning algorithms are used to build a model to predict adverse drug reactions, which solves the timeliness and accuracy of traditional drug adverse reaction monitoring methods, and achieves more efficient monitoring and higher drug safety.
Patent Information
- Application Number
- CN202411795732.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-09
- Publication Date
- 2025-03-18
AI Technical Summary
传统药品不良反应监测方法存在报告率低、信息不全面、时效性差的问题，难以实现及时、准确的监测和管理。
Through smartphone applications and wearable devices, patients' medication information and health data are collected in real time, and data cleaning, standardization and machine learning algorithms are used to build models to predict adverse drug reactions and generate alerts and analysis results in real time.
It improves the monitoring efficiency and analysis accuracy of adverse drug reactions, reduces the incidence of adverse drug reactions, and improves the safety of patients' medication.
Smart Images
Figure CN119650099A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of medical monitoring, and in particular relates to a method and device for monitoring, managing and analyzing adverse drug reactions. Background Art
[0002] Adverse drug reaction (ADR) refers to the adverse reaction that is unrelated to the purpose of medication or unexpected when a qualified drug is used in normal dosage. The occurrence of adverse drug reactions not only brings pain to patients, but may even endanger their lives, and also increases medical costs and social burdens. Therefore, it is of great significance to monitor and manage adverse drug reactions in a timely and accurate manner.
[0003] Traditional methods of monitoring adverse drug reactions mainly rely on spontaneous reports from medical institutions and patients, which have problems such as low reporting rate, incomplete information, and poor timeliness. With the development of information technology and the Internet of Things, it has become possible to use smartphone applications (APPs) and wearable devices to collect patients' medication information and health data in real time, providing a new means for monitoring adverse drug reactions. Summary of the invention
[0004] The purpose of the present invention is to improve the data collection efficiency and analysis accuracy, ultimately reduce the incidence of adverse drug reactions, and improve the patient's medication safety.
[0005] The technical solution of the present invention is as follows:
[0006] A method for monitoring and analyzing adverse drug reactions comprises the following steps:
[0007] Collecting patients’ medication information and health data through smartphone apps and wearable devices;
[0008] Periodically uploading the health data and medication information to a cloud database for storage;
[0009] Processing the health data and medication information using data cleaning and standardization steps to eliminate outliers and missing values, and then extracting features related to adverse drug reactions;
[0010] Using machine learning algorithms, we train models on the extracted features, build and train models for predicting adverse drug reactions, and identify the probability of adverse drug reactions.
[0011] A threshold for the incidence of adverse reactions is set. When the incidence of adverse reactions of a drug exceeds the threshold, an alarm is generated in real time and the analysis results are fed back to patients and medical staff.
[0012] Furthermore, the medication information includes the name of the drug, dosage, frequency of use and method of use; the health data includes but is not limited to records of heart rate, blood pressure, body temperature, weight and subjective discomfort symptoms.
[0013] Furthermore, the health data and medication information are uploaded every 15 minutes.
[0014] Furthermore, the model for predicting adverse drug reactions is constructed and trained as follows:
[0015] First, features related to adverse drug reactions were extracted for data standardization and coding;
[0016] Based on domain knowledge and statistical methods, we select features that are highly correlated with adverse drug reactions, and use correlation analysis to screen out features that contribute to adverse reaction prediction;
[0017] Select Random Forest as the main algorithm and set the hyperparameters; including the number of trees, maximum depth, and minimum number of sample splits;
[0018] Finally, the training set data is used to train the random forest model. During the training process, the random forest establishes a decision tree through multiple random sampling to learn the distribution and association patterns of different features in the data. In the model training stage, cross-validation is used to verify the performance of the model in the training set, and the parameters are adjusted to obtain the optimal model.
[0019] Furthermore, the calculation formula for the probability of occurrence of adverse drug reactions is:
[0020]
[0021] Among them, N ADR is the number of adverse reactions reported, N total Total number of people taking medication
[0022] Furthermore, the feedback results include the type, severity and recommended treatment measures of the adverse reaction, and are presented in the form of a visual report.
[0023] The present invention also provides a drug adverse reaction monitoring management and analysis device, comprising:
[0024] Data collection module for collecting patients’ medication information and health data in real time through smartphone applications and wearable devices.
[0025] A data storage module is used to upload the collected data to a cloud database. The data storage module uses an AES-256 encryption algorithm to encrypt the data;
[0026] Data analysis module, which performs real-time analysis on uploaded data and identifies patterns of adverse drug reactions;
[0027] The feedback module is used to feed back the analysis results to patients and medical staff in real time, and the feedback module has the functions of generating alarms, notifications and visual reports.
[0028] Furthermore, the data acquisition module has a data input interface and supports manual and automatic data recording.
[0029] Furthermore, the data analysis module includes:
[0030] Data cleaning function: automatically identify and process outliers and missing values;
[0031] Probability calculation function, used to calculate the probability of occurrence of adverse drug reactions.
[0032] Furthermore, the device also includes a user management module for managing user registration, login and permission settings, and supporting multiple user roles.
[0033] Compared with the prior art, the present invention has the following advantages:
[0034] The present invention combines smartphone applications with wearable devices to achieve real-time monitoring and automated feedback of adverse drug reactions, with the advantages of personalized medicine, data integration and analysis, and privacy protection. The system can accurately identify adverse reaction patterns, generate alarms in a timely manner, and provide visual reports, thereby improving the safety of patient medication and the decision-making efficiency of medical staff. In addition, the system is user-friendly, easy to operate, and has the ability to continuously improve, which can be expanded to the monitoring and management of other diseases. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] The accompanying drawings generally illustrate various embodiments by way of example and not limitation, and together with the description and claims, serve to illustrate the embodiments of the invention. Where appropriate, the same reference numerals are used throughout the drawings to refer to the same or similar parts. Such embodiments are illustrative and are not intended to be exhaustive or exclusive embodiments of the present apparatus or method.
[0036] Figure 1 A schematic diagram of the method flow of the present invention is shown;
[0037] Figure 2 A module schematic diagram of the present invention is shown. DETAILED DESCRIPTION
[0038] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0039] The embodiment of the present invention provides a method for monitoring, managing and analyzing adverse drug reactions, comprising:
[0040] Data Collection
[0041] The data collection of this method mainly relies on smartphone applications and wearable devices. Smartphone applications are designed to facilitate patients to enter medication information, including drug name, dosage, frequency of use, and method of use. At the same time, the application can also receive health data from wearable devices, such as heart rate, blood pressure, body temperature, weight, and subjective discomfort symptoms. These data are transmitted to smartphones in real time through wireless communication technologies such as Bluetooth and Wi-Fi, and then preliminarily processed and packaged by smartphone applications.
[0042] Drug name: Patients need to accurately enter the name of the drug they are taking, such as "aspirin", "cefradine", etc.
[0043] Dosage: Patients need to enter the dosage of the drug taken each time, such as "50 mg", "1 tablet", etc., and can select the dosage unit.
[0044] Frequency of medication: Patients need to select the frequency of medication, such as "once a day", "twice a day", etc., and can choose the specific time of medication.
[0045] Medication method: Patients need to choose the method of medication, such as "oral", "topical", etc.
[0046] For health data, wearable devices monitor patients' physiological indicators in real time through built-in sensors, such as heart rate, blood pressure, body temperature, etc., and transmit these data to smartphone applications through wireless communication technology. In addition, patients can also manually record their weight and subjective discomfort symptoms such as headaches, nausea, vomiting, etc. through the application.
[0047] Heart rate: Wearable devices monitor the patient's heart rate in real time through a photoplethysmography (PPG) sensor, with data in beats per minute.
[0048] Blood Pressure: Wearable devices monitor the patient's blood pressure in real time through a blood pressure sensor. The data is in mmHg, including systolic and diastolic pressure.
[0049] Body temperature: Wearable devices monitor the patient's body temperature in real time through temperature sensors, and the data is in °C.
[0050] Weight: Patients can manually enter their weight in kg via the smartphone app.
[0051] Subjective discomfort symptoms: Patients can select or input their subjective discomfort symptoms such as headache, nausea, vomiting, etc. through the smartphone application.
[0052] Data upload and storage
[0053] The smartphone application packages the collected medication information and health data and uploads them to the cloud database every 15 minutes. The cloud database adopts a distributed architecture with high availability and scalability, can store a large amount of patient data, and supports efficient data query and analysis.
[0054] To ensure data security, the cloud database uses the AES-256 encryption algorithm to encrypt data.
[0055] Data cleaning and standardization
[0056] After uploading the data to the cloud database, data cleaning and standardization are required. The purpose of data cleaning is to eliminate outliers and missing values in the data to ensure the accuracy and completeness of the data. Outliers may be caused by equipment failure, data transmission errors, or patient misoperation, such as heart rate exceeding the normal range (such as 150 beats / minute or less than 50 beats / minute). Missing values may be caused by not wearing the device, low battery, or the patient forgetting to record.
[0057] Data standardization is to convert data into a unified format and unit to facilitate subsequent analysis and processing. For example, heart rate data is converted into beats / minute, blood pressure data is converted into mmHg, and weight data is converted into kg.
[0058] Data cleaning and standardization can be achieved by writing specific data cleaning scripts or using existing data cleaning tools. These scripts or tools can automatically identify and process outliers and missing values in the data and convert the data into a unified format and unit.
[0059] Data analysis and identification
[0060] After data cleaning and standardization, the data needs to be analyzed and identified to find patterns and trends of adverse drug reactions. The invention uses machine learning algorithms for data analysis and identification, including random forest algorithms.
[0061] First, the data needs to be feature extracted and selected. Feature extraction is to extract features related to adverse drug reactions from the original data, such as heart rate changes, blood pressure fluctuations, weight changes, etc., and standardize numerical data (such as dosage, heart rate) to make its mean 0 and standard deviation 1 to reduce the impact of different data ranges on model training. One-hot encoding is performed on categorical data (such as drug name, medication method) to convert categorical features into numerical features for easy processing by the random forest algorithm.
[0062] Next, the selected random forest algorithm is used to train the model on the extracted features.
[0063] Based on domain knowledge and statistical methods, select features that are highly correlated with adverse drug reactions. For example, features such as medication dosage, heart rate fluctuations, blood pressure changes, and specific drug combinations may be important factors affecting the occurrence of adverse reactions. Use correlation analysis to screen out the features that contribute most to the prediction of adverse reactions.
[0064] For example, combining drug dosage with factors such as body temperature and ambient temperature to form interactive features can help identify the risk that certain drugs are more likely to cause adverse reactions under special conditions.
[0065] Random forest is selected as the main algorithm. Random forest consists of multiple decision trees. By combining the prediction results of multiple trees, the generalization ability and prediction accuracy of the model are improved. Each tree is generated by random sampling on features and samples, which can effectively avoid overfitting.
[0066] Hyperparameter settings
[0067] Number of trees: Set the number of decision trees, usually between 100 and 500. The more trees there are, the more stable the model will be, but the amount of computation will also increase.
[0068] Maximum depth: Set the maximum depth of each tree to prevent the tree from being too deep and causing overfitting. For healthy data, shallower trees (such as 5-10 layers) are usually selected.
[0069] Minimum number of sample splits: Set the minimum number of samples required for node splitting. Generally, it can be set to 5 or 10 to avoid generating too many nodes.
[0070] The random forest model is trained using the training set data. During the training process, the random forest learns the distribution and association patterns of different features in the data through multiple random sampling and building decision trees.
[0071] During the model training phase, cross-validation (such as 5-fold cross-validation) is used to verify the performance of the model in the training set, and the parameters are adjusted to obtain the optimal model.
[0072] After the model training is completed, the model needs to be verified and evaluated. The purpose of verification and evaluation is to evaluate the prediction performance of the model, including indicators such as accuracy, recall, and F1 score. If the prediction performance of the model meets the expected requirements, the model can be used to predict and identify new data.
[0073] Through data analysis and identification, the probability and pattern of adverse drug reactions can be found. In order to quantify the incidence of adverse reactions, this method uses the following formula for calculation:
[0074]
[0075] Among them, N ADR is the number of adverse reactions reported, N total The total number of people taking medication.
[0076] Finally, the analysis results will be fed back to patients and medical staff in a timely manner. The specific contents are as follows:
[0077] Alert generation:
[0078] When the incidence of adverse reactions to a drug exceeds a set threshold (e.g. 5%), the system will automatically generate an alert, which is pushed to the devices of patients and medical staff through the application.
[0079] Alert information includes:
[0080] Type of adverse reaction (such as headache, nausea, etc.)
[0081] The incidence and severity of adverse reactions (such as mild, moderate, severe)
[0082] Recommended treatment measures (such as reducing medication dosage, changing medications, seeking medical advice, etc.)
[0083] The feedback report generated by the system is displayed in a visual form, including: a time trend graph of adverse reactions, a pie chart or bar chart of adverse reaction types, and a relationship graph between key health indicators (such as heart rate, blood pressure) and adverse reactions.
[0084] Patients can view their health monitoring data and adverse reaction records through the application to increase their understanding of their own health status.
[0085] Establish a patient feedback mechanism to collect patients' evaluations and suggestions on their system experience and continuously optimize system functions.
[0086] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. A method for monitoring, managing and analyzing adverse drug reactions, characterized in that: The following steps are involved: Collecting patients’ medication information and health data through smartphone apps and wearable devices; Periodically uploading the health data and medication information to a cloud database for storage; Processing the health data and medication information using data cleaning and standardization steps to eliminate outliers and missing values, and then extracting features related to adverse drug reactions; Using machine learning algorithms, we train models on the extracted features, build and train models for predicting adverse drug reactions, and identify the probability of adverse drug reactions. A threshold for the incidence of adverse reactions is set. When the incidence of adverse reactions of a drug exceeds the threshold, an alarm is generated in real time and the analysis results are fed back to patients and medical staff.
2. The method for monitoring and analyzing adverse drug reactions according to claim 1, characterized in that: The medication information includes the name of the drug, dosage, frequency of use and method of use; The health data includes but is not limited to records of heart rate, blood pressure, body temperature, weight and subjective discomfort symptoms.
3. The method for monitoring and analyzing adverse drug reactions according to claim 1, characterized in that: The specific steps for building a model to predict adverse drug reactions and training it are as follows: First, features related to adverse drug reactions were extracted for data standardization and coding; Based on domain knowledge and statistical methods, we select features that are highly correlated with adverse drug reactions, and use correlation analysis to screen out features that contribute to adverse reaction prediction; Select Random Forest as the main algorithm and set the hyperparameters; including the number of trees, maximum depth, and minimum number of sample splits; Finally, the training set data is used to train the random forest model. During the training process, the random forest establishes a decision tree through multiple random sampling to learn the distribution and association patterns of different features in the data. In the model training stage, cross-validation is used to verify the performance of the model in the training set, and the parameters are adjusted to obtain the optimal model.
4. The method for monitoring and analyzing adverse drug reactions according to claim 1, characterized in that: The health data and medication information are uploaded every 15 minutes.
5. The method for monitoring and analyzing adverse drug reactions according to claim 1, characterized in that: The calculation formula for the probability of occurrence of the adverse drug reaction is: Among them, N ADR is the number of adverse reactions reported, N total The total number of people taking medication.
6. The method for monitoring and analyzing adverse drug reactions according to claim 1, characterized in that: The feedback results include the type, severity and recommended treatment measures of the adverse reaction, and are presented in the form of a visual report.
7. A drug adverse reaction monitoring management and analysis device, characterized in that: include: Data collection module for collecting patients’ medication information and health data in real time through smartphone applications and wearable devices. A data storage module is used to upload the collected data to a cloud database. The data storage module uses an AES-256 encryption algorithm to encrypt the data; A data analysis module, which processes the health data and medication information using data cleaning and standardization steps, eliminates outliers and missing values, and then extracts features related to adverse drug reactions; Use machine learning algorithms to train models on the extracted features, build models to predict adverse drug reactions, and identify the probability of adverse drug reactions; The feedback module is used to feed back the analysis results to patients and medical staff in real time, and the feedback module has the functions of generating alarms, notifications and visual reports.
8. The device for monitoring, managing and analyzing adverse drug reactions according to claim 7, characterized in that: The data acquisition module has a data input interface and supports manual and automatic data recording.
9. The device for monitoring, managing and analyzing adverse drug reactions according to claim 7, characterized in that: The data analysis module includes: Data cleaning function: automatically identify and process outliers and missing values; Probability calculation function, used to calculate the probability of occurrence of adverse drug reactions.
10. The device for monitoring, managing and analyzing adverse drug reactions according to claim 7, characterized in that: The device also includes a user management module for managing user registration, login and permission settings, and supporting multiple user roles.