Deep learning-based early warning method and system for infantile asthmatic attack

Abstract

The invention discloses a children asthmatic attack early warning method and system based on deep learning, and relates to the technical field of intelligent health monitoring, and the method comprises the steps: constructing an electronic medical record system, collecting disease data according to the electronic medical record system, training the disease data, and marking a known asthmatic attack event; analyzing the extracted asthmatic attack event through a time sequence, and capturing the physiological parameter change trend of the infantile asthma; and establishing an evaluation index system of confrontation training, integrating a result after confrontation training into an electronic medical record system according to an evaluation result, and customizing a child asthma early warning intervention plan. According to the method, the trend report is automatically generated, the treatment scheme is helped to be adjusted, the intelligent level of asthma management is improved, the coping ability of parents and medical staff is enhanced, asthmatic attack is effectively prevented, and the life quality of child patients is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent health monitoring technology, and in particular to a method and system for early warning of pediatric asthma attacks based on deep learning. Background Art

[0002] With the rapid development of information technology, the collection, processing and analysis of health data have become the key to improving the efficiency of personal health management. At present, the widespread application of technologies such as smart wearable devices, the Internet of Things, big data analysis and artificial intelligence has made it possible to monitor and analyze personal health status in real time. People are increasingly relying on these technologies to obtain continuous and accurate health data, thereby achieving refined management of their own health. Especially in the field of children's health, parents and caregivers hope to use advanced information and communication technologies to have a more comprehensive understanding of their children's health status, timely identify potential problems and take preventive measures. This demand has driven the continuous innovation and development of health data processing technology.

[0003] Although existing health data management technologies can collect a large amount of physiological parameters and environmental data, they have obvious deficiencies in data integration and intelligent analysis. Traditional data collection platforms can usually only provide a single type of health data, such as heart rate or respiratory rate, and lack the ability to comprehensively process multi-source data. In addition, existing systems also have limitations in data cleaning, feature extraction, and model training, making it difficult to effectively identify complex time series patterns and individual differences. Rule-based early warning systems are often too rigid and cannot adapt to the specific circumstances of different users. Summary of the Invention

[0004] In view of the problems existing in the existing pediatric asthma attack warning methods and systems based on deep learning, the present invention is proposed.

[0005] Therefore, the present invention addresses the problems of inaccurate predictions and untimely intervention caused by reliance on subjective reports and experience in the existing technology. By integrating multi-source health status information and utilizing deep learning and adversarial training optimization models, the present invention achieves accurate prediction and personalized early warning intervention for pediatric asthma attacks.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In the first aspect, the present invention provides a pediatric asthma attack warning method based on deep learning, which includes: designing a data collection platform for pediatric asthma attacks, obtaining health status information, integrating historical medical records in the health status information, building an electronic medical record system, collecting disease data according to the electronic medical record system, training the disease data, and marking known asthma attack events; using the trained disease data as a learning model, and using the learning model to extract the marked known asthma attack events, and analyzing the extracted asthma attack events through time series to capture the changing trends of physiological parameters of pediatric asthma; based on the changing trends of physiological parameters, customizing the network layer architecture of pediatric asthma, introducing the learning model and the network layer architecture to form adversarial training, establishing an evaluation index system for adversarial training, and integrating the results of adversarial training into the electronic medical record system according to the evaluation results to customize a pediatric asthma warning intervention plan.

[0007] As a preferred embodiment of the deep learning-based pediatric asthma attack warning method described in the present invention, the acquisition of health status information includes physiological parameter monitoring, environmental factor monitoring and symptom log recording of children, the physiological parameter monitoring of children includes heart rate parameters, respiratory rate parameters, blood oxygen saturation parameters, body temperature parameters and skin electrical response parameters, the environmental factor monitoring includes air quality parameter monitoring and temperature and humidity parameter monitoring, the symptom log recording includes preprocessing of symptom log records based on physiological parameter monitoring and environmental factor monitoring of children, the preprocessing includes establishing a mobile application development model for symptom log recording, setting an automatic reminder function based on the mobile application development model, marking the abnormal data points to be reminded, and interactively confirming the marked abnormal data points.

[0008] As a preferred embodiment of the deep learning-based early warning method for pediatric asthma attacks of the present invention, the air quality parameter includes an air quality parameter calculated by an air quality monitor, and the specific calculation formula for the air quality parameter is: ; in, represents the pollutant concentration, Indicates the minimum pollutant concentration, Indicates the maximum pollutant concentration, Indicates the minimum pollutant concentration corresponding to index, Indicates the maximum pollutant concentration corresponding to index, Indicates air quality parameters; The temperature and humidity parameters include a comprehensive formula for calculating temperature and relative humidity using a temperature and humidity sensor. The calculation formula for the comprehensive formula for temperature and relative humidity is: ; in, Indicates temperature and humidity parameters, represents the water vapor pressure, 、 and represents a constant, Indicates temperature; Based on air quality parameters, the data collection platform is used to provide additional recommendations in combination with temperature and humidity parameters; When The threshold value is ;when When The threshold value is ,when When The threshold value is ; when When , the environmental factor monitoring is in the first level state; when When , the environmental factor monitoring is in the secondary state; when When , the environmental factor monitoring is in the third level state.

[0009] As a preferred embodiment of the deep learning-based pediatric asthma attack early warning method of the present invention, the construction of an electronic medical record system includes monitoring status based on different environmental factors, developing an EMR system for pediatric asthma management, developing an API interface to connect historical medical records with the EMR system, and formulating an episode data migration plan. The episode data migration plan includes dividing data migration into N phases and gradually migrating the data. The migrated episode data is then evaluated for consistency with the unmigrated episode data. The determination of whether the disease occurrence data is consistent with the unmigrated data includes analyzing data synchronization between the old and new systems, and the data synchronization includes an incremental synchronization mechanism, a timed synchronization mechanism, and a bidirectional synchronization mechanism; The incremental synchronization mechanism includes the generation of new medical records and changes to patient information. When a new medical record is generated, a medical record is created or updated in the old system, sent to the new system through the API interface, marked as synchronized, and the timestamp of the medical record operation is recorded in the old system. When patient information changes, it is modified in the old system, triggering an incremental synchronization task to push the latest patient information to the new system. At the same time, the new system is checked to see if there is a record with the same patient ID. If so, it is updated; otherwise, a new record is created. The timing synchronization mechanism includes making a judgment based on a preset time interval and a detected network connection recovery state; When the time interval between the old and new systems arrives on time, a batch synchronization task is started to extract the new or modified data from the old system since the last synchronization and import it into the new system. At the same time, the data consistency between the old and new systems is compared to ensure that all data has been synchronized. When the network connection is restored after being interrupted, a full synchronization task is triggered to prevent the data generated during the network failure from being lost. At the same time, the synchronization log is recorded to obtain the asthma attack event.

[0010] As a preferred embodiment of the deep learning-based early warning method for pediatric asthma attacks of the present invention, the learning model includes normalizing the synchronized data to generate a feature vector suitable for input to the learning model, extracting time series features, wherein the time series features include mean, variance, and peak value; training the onset data includes labeling the asthma attack events, using the labels to perform supervised learning training on the model, and employing cross-validation and hyperparameter optimization techniques to complete the training of the learning model; The calculation formula of the mean is: ; in, represents the mean, represents the number of samples, Indicates the data points; The specific calculation formula of the variance is: ; in, represents the variance, represents the square of the deviation; The specific calculation formula of the peak value is: ; in, Indicates the peak value, Represents the maximum value function.

[0011] As a preferred embodiment of the deep learning-based early warning method for pediatric asthma attacks described in the present invention, capturing the changing trends of physiological parameters of pediatric asthma includes providing dynamic charts on both the old and new systems to display the changing trends of the physiological parameters of the child, observing the changes in the child's heart rate, respiratory rate, and blood oxygen saturation parameters through the charts, generating historical comparison charts, and analyzing individual differences among the children based on the historical comparison charts; Generating a historical comparison chart includes displaying the time points of asthma attacks according to a dynamic chart, selecting different time periods for comparison, and then automatically generating a corresponding comparison chart with the old and new systems to provide statistical analysis results; Customize the analysis model based on the statistical analysis results provided. When the old and new systems detect abnormal fluctuations in the physiological parameters of pediatric asthma, the old and new systems will automatically call the analysis model, generate an analysis report, and provide health advice. The new and old systems track the physiological parameters of each child with asthma over a long period of time, recording their changing trends over time and adjusting treatment plans by analyzing long-term data; When the system detects a change in the long-term trend, it will generate a trend change report and remind the staff to recommend whether to continue the current treatment plan.

[0012] As a preferred embodiment of the deep learning-based early warning method for pediatric asthma attacks of the present invention, the customized pediatric asthma early warning intervention plan includes generating an intervention plan based on the captured physiological parameter change trend by the new and old systems; When it is detected that the air quality does not meet the standard, take air purification measures, increase ventilation frequency or install a fresh air system; When it is detected that the air quality has returned to the standard, the network layer architecture for pediatric asthma is customized to complete real-time monitoring of pediatric asthma attack warnings.

[0013] In a second aspect, the present invention provides a pediatric asthma attack early warning system based on deep learning, which includes: a construction module that designs a data collection platform for pediatric asthma attacks, obtains health status information, integrates historical medical records in the health status information, constructs an electronic medical record system, collects disease data based on the electronic medical record system, trains the disease data, and marks known asthma attack events; The extraction module uses the trained onset data as a learning model to extract asthma attack events with known markers. It then analyzes the extracted asthma attack events through time series analysis to capture the changing trends of physiological parameters of pediatric asthma. The intervention module customizes the network layer architecture of pediatric asthma based on the changing trends of physiological parameters, introduces the learning model and the network layer architecture to form adversarial training, establishes an evaluation index system for adversarial training, and integrates the results of adversarial training into the electronic medical record system based on the evaluation results to customize the pediatric asthma early warning intervention plan.

[0014] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the processor executes the computer program, it implements any step of the above-mentioned deep learning-based pediatric asthma attack warning method.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the above-mentioned deep learning-based pediatric asthma attack warning method is implemented.

[0016] The beneficial effects of the present invention are: the present invention integrates children's physiological parameters, environmental factors and symptom logs to build a comprehensive pediatric asthma attack warning system, uses deep learning models to analyze multi-source data, and realizes the precise extraction and prediction of asthma attack events, significantly improving the accuracy and timeliness of the warning. The system can monitor environmental factors such as air quality, temperature and humidity in real time, and provide personalized health advice and intervention measures according to the specific situation of the child. Through the incremental synchronization and timed synchronization mechanism of historical medical records, the consistency and integrity of the data are ensured, supporting doctors to make more scientific treatment decisions. In addition, the system can track the changing trends of the physiological parameters of children in the long term, automatically generate trend reports, and help adjust treatment plans. The present invention not only improves the level of intelligence of asthma management, but also enhances the coping ability of parents and medical staff, effectively prevents asthma attacks, and improves the quality of life of children. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them: Figure 1 This is a specific flow chart of a deep learning-based pediatric asthma attack warning method provided by one embodiment of the present invention.

[0018] Figure 2 A schematic diagram of a deep learning-based pediatric asthma attack warning method and system provided in one embodiment of the present invention. DETAILED DESCRIPTION

[0019] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field 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. However, the present invention may also be implemented in other ways different from those described herein. 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] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0022] The present invention is described in detail with reference to schematic diagrams. For ease of illustration, cross-sectional views of device structures may be partially enlarged and not to scale when describing embodiments of the present invention. Furthermore, the schematic diagrams are merely illustrative and should not limit the scope of the present invention. Furthermore, in actual production, the three-dimensional dimensions of length, width, and depth should be included.

[0023] In the description of the present invention, it should be noted that the terms "upper, lower, inner, and outer" and other references to orientations or positional relationships are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first, second, or third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0024] In this disclosure, unless otherwise specified or limited, the terms "mounted," "connected," and "connected" should be interpreted broadly. For example, they may refer to fixed, removable, or integral connections. They may also refer to mechanical, electrical, or direct connections, indirect connections through an intermediary, or internal communication between two components. Those skilled in the art will understand the specific meanings of these terms in this disclosure.

[0025] Example 1: Reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a method for warning of pediatric asthma attacks based on deep learning, comprising: S1: Design a data collection platform for pediatric asthma attacks, obtain health status information, integrate historical medical records in the health status information, build an electronic medical record system, collect disease data based on the electronic medical record system, train the disease data, and mark known asthma attack events.

[0026] Among them, obtaining health status information includes physiological parameter monitoring of children, environmental factor monitoring and symptom log recording. Physiological parameter monitoring of children includes heart rate parameters, respiratory rate parameters, blood oxygen saturation parameters, body temperature parameters and skin electrical response parameters. Environmental factor monitoring includes air quality parameter monitoring and temperature and humidity parameter monitoring. Symptom log recording includes preprocessing of symptom log records based on physiological parameter monitoring and environmental factor monitoring of children. Preprocessing includes establishing a mobile application development model for symptom log recording, setting an automatic reminder function based on the mobile application development model, marking the abnormal data points to be reminded, and interactively confirming the marked abnormal data points.

[0027] S1.1: Air quality parameters include those calculated by air quality monitors. The specific calculation formula for air quality parameters is: ; in, represents the pollutant concentration, Indicates the minimum pollutant concentration, Indicates the maximum pollutant concentration, Indicates the minimum pollutant concentration corresponding to index, Indicates the maximum pollutant concentration corresponding to index, Indicates air quality parameters; The temperature and humidity parameters include a comprehensive formula for calculating temperature and relative humidity using a temperature and humidity sensor. The calculation formula for the comprehensive formula for temperature and relative humidity is: ; in, Indicates temperature and humidity parameters, represents the water vapor pressure, 、 and represents a constant, Indicates temperature; Based on air quality parameters, the data collection platform is used to provide additional recommendations in combination with temperature and humidity parameters; When The threshold value is ;when When The threshold value is ,when When The threshold value is ; when When , the environmental factor monitoring is in the first level state; when When , the environmental factor monitoring is in the secondary state; when When , the environmental factor monitoring is in the third level state.

[0028] Furthermore, the air quality monitor calculates the air quality index (AQI) by measuring the concentration of pollutants. When the concentration of PM2.5 is 75 micrograms per cubic meter, the system will calculate according to the preset concentration range and the corresponding index. If the concentration range of PM2.5 is 0-35 micrograms per cubic meter, the corresponding AQI index is 0-50, and 35-75 micrograms per cubic meter corresponds to the AQI index 51-100, then a PM2.5 concentration of 75 micrograms per cubic meter will be calculated as an AQI value of 100, indicating that the air quality is in a state of light pollution. Table 1 Correspondence between air quality index (AQI) and PM2.5 concentration. Table 1 Correspondence between air quality index AQI and PM2.5 concentration PM2.5 concentration range (μg / m3) Corresponding AQI index range PM2.5 concentration example (µg / m³) Calculated AQI value Air quality status 0 - 35 0 - 50 35 50 good 35 - 75 51 - 100 75 100 Light pollution 75 - 115 101 - 150 115 150 Moderate pollution 115 - 150 151 - 200 150 200 Severe pollution 150 - 250 201 - 300 250 300 Severe pollution >250 >300 300 >300 Danger Table 1 shows the correspondence between PM2.5 concentration and the air quality index AQI. It specifically lists the PM2.5 values ​​in different concentration ranges and their corresponding AQI index and air quality status. When the PM2.5 concentration is 75 micrograms / cubic meter, the calculated AQI value is 100, indicating that the air quality is in a state of mild pollution.

[0029] When the AQI value is less than 50, the air quality is good and normal outdoor activities are recommended. When the AQI value is between 51 and 100, the air quality is slightly polluted and it is recommended to reduce long-term outdoor activities, especially for children with asthma. When the AQI value is between 101 and 150, the air quality is moderately polluted and it is recommended to reduce outdoor activities and use an air purifier. When the AQI value is greater than 150, the air quality is severely polluted and it is recommended to avoid outdoor activities, use an air purifier and consult a doctor. The correspondence between the air quality index AQI and health recommendations is shown in Table 2 below: Table 2 Corresponding data of air quality index AQI and health recommendations AQI index range Air quality status Health Advice 0 - 50 good Normal outdoor activities are recommended 51 - 100 Light pollution It is recommended to reduce long-term outdoor activities, especially for children with asthma 101 - 150 Moderate pollution It is recommended to reduce outdoor activities and use air purifiers 151 - 200 Severe pollution It is recommended to avoid outdoor activities, use air purifiers and consult a doctor 201 - 300 Severe pollution It is strongly recommended to avoid outdoor activities, use an air purifier and consult a doctor as soon as possible >300 Danger You must avoid outdoor activities, use an air purifier and consult a doctor immediately Table 2 shows the air quality status within different AQI index ranges, from good to hazardous, and provides corresponding health advice, such as normal outdoor activities, reducing outdoor time, using air purifiers, and consulting a doctor. Through clear classification and specific recommendations, it helps parents and medical staff take appropriate protective measures based on air quality to ensure the health and safety of children.

[0030] The temperature and humidity sensor measures temperature and relative humidity to assess the comfort of the indoor environment. When the temperature is 25 degrees Celsius and the relative humidity is 40%, the system combines these data for a comprehensive assessment. If the temperature is between 20 and 25 degrees Celsius and the relative humidity is between 40% and 60%, the system considers the environment comfortable and suitable for daily activities. The temperature and humidity parameters and the indoor environment comfort assessment table are shown in Table 3 below: Table 3 Temperature and humidity parameters and indoor environment comfort evaluation table Temperature range (Celsius) Relative humidity range Environmental Assessment Health Advice 20 - 25 40 - 60 Comfort Suitable for daily activities <20 <40 Cooler and dry It is recommended to use a humidifier and increase the indoor temperature appropriately >25 >60 Hot and humid It is recommended to use air conditioning or dehumidifier to keep it cool and dry 20 - 25 <40 dry It is recommended to use a humidifier 20 - 25 >60 damp It is recommended to use a dehumidifier <20 40 - 60 cooler It is recommended to increase the indoor temperature appropriately >25 40 - 60 Hotter It is recommended to use air conditioning to cool down Table 3 provides a more intuitive understanding of the indoor environmental comfort assessment and corresponding health recommendations under different temperature and relative humidity combinations, helping parents and medical staff to take appropriate measures based on actual conditions to ensure the health and comfort of children.

[0031] The system divides environmental factor monitoring into three levels based on different combinations of the air quality index (AQI) and temperature and humidity parameters. When the AQI value is less than 50 and the relative humidity is between 40% and 60%, the environmental factor monitoring is in level one, indicating that the environment is good and no special intervention is required; when the AQI value is between 51 and 100 or the relative humidity is below 40% or above 60%, the environmental factor monitoring is in level two, and it is recommended to take appropriate measures, such as using an air purifier or humidifier; when the AQI value is greater than 100, or the temperature is below 20 degrees Celsius and the relative humidity is below 40%, or the temperature is above 25 degrees Celsius and the relative humidity is above 60%, the environmental factor monitoring is in level three, and it is recommended to take immediate action, such as avoiding outdoor activities, using air purifiers and air conditioners, to protect children's health. The environmental factor monitoring levels and recommendations are shown in Table 4 below: Table 4 Environmental factor monitoring levels and recommendations AQI index range Temperature range / degrees Celsius Relative humidity range% Environmental monitoring level Environmental Assessment Health Advice <50 - 40 - 60 Level 1 status Good environment No special intervention required 51 - 100 - 40 - 60 Secondary status Light pollution It is recommended to take appropriate measures, such as using an air purifier 51 - 100 - <40 or >60 Secondary status Discomfort It is recommended to take appropriate measures, such as using a humidifier or dehumidifier >100 - - Level 3 status Moderate to severe pollution Immediate actions are recommended, such as avoiding outdoor activities, using air purifiers and air conditioners - <20 <40 Level 3 status Cooler and dry Immediate action is recommended, such as using a humidifier and raising the indoor temperature appropriately - >25 >60 Level 3 status Hot and humid It is recommended to take immediate action, such as using air conditioning or a dehumidifier to keep it cool and dry Table 4 provides a more intuitive understanding of the environmental factor monitoring levels and corresponding health recommendations under different AQI values ​​and temperature and humidity combinations, helping parents and medical staff take appropriate measures based on actual conditions to ensure the health and safety of children.

[0032] S1.2: Building an electronic medical record system includes monitoring status based on different environmental factors, developing an EMR system for pediatric asthma management, developing an API interface to connect historical medical records with the EMR system, and developing an episode data migration plan. The episode data migration plan includes dividing the data migration into N phases and gradually migrating it. The migrated episode data is then evaluated for consistency with the unmigrated episode data. Determining whether the disease data is consistent with the unmigrated data includes analyzing the data synchronization between the old and new systems. Data synchronization includes incremental synchronization, scheduled synchronization, and bidirectional synchronization mechanisms. The incremental synchronization mechanism includes the generation of new medical records and changes to patient information. When a new medical record is generated, a medical record is created or updated in the old system, sent to the new system through the API interface, marked as synchronized, and the timestamp of the medical record operation is recorded in the old system. When patient information changes, it is modified in the old system, triggering an incremental synchronization task to push the latest patient information to the new system. At the same time, the new system is checked to see if there is a record with the same patient ID. If so, it is updated; otherwise, a new record is created. The timing synchronization mechanism includes making judgments based on the preset time interval and the detected network connection recovery status; When the time interval between the old and new systems arrives on time, a batch synchronization task is started to extract the new or modified data from the old system since the last synchronization and import it into the new system. At the same time, the data consistency between the old and new systems is compared to ensure that all data has been synchronized. When the network connection is restored after being interrupted, a full synchronization task is triggered to prevent the data generated during the network failure from being lost. At the same time, the synchronization log is recorded to obtain the asthma attack event.

[0033] S2: Use the trained disease data as a learning model and use the learning model to extract asthma attack events with known markers. Through time series analysis of the extracted asthma attack events, the changing trends of physiological parameters of pediatric asthma are captured.

[0034] The learning model involves normalizing the synchronized data to generate a feature vector suitable for the learning model input, extracting time series features such as mean, variance, and peak value, and training the disease data by labeling asthma attack events. The labels are then used to train the model through supervised learning. Cross-validation and hyperparameter optimization techniques are then used to complete the training of the learning model. The formula for calculating the mean is: ; in, represents the mean, represents the number of samples, Indicates the data points; The specific calculation formula for variance is: ; in, represents the variance, represents the square of the deviation; The specific calculation formula for the peak value is: ; in, Indicates the peak value, Represents the maximum value function.

[0035] S2.1: Capturing the changing trends of physiological parameters of pediatric asthma includes providing dynamic charts on both the old and new systems to display the changing trends of physiological parameters of the child. This chart can be used to observe changes in the child's heart rate, respiratory rate, and blood oxygen saturation parameters, generate historical comparison charts, and analyze individual differences among children based on these historical comparison charts. Generating a historical comparison chart involves displaying the time points of asthma attacks using dynamic charts and selecting different time periods for comparison. At this point, the new and old systems will automatically generate corresponding comparison charts and provide statistical analysis results. Customize the analysis model based on the statistical analysis results provided. When the old and new systems detect abnormal fluctuations in the physiological parameters of pediatric asthma, the old and new systems will automatically call the analysis model, generate an analysis report, and provide health advice. The new and old systems track the physiological parameters of each child with asthma over a long period of time, recording their changing trends over time and adjusting treatment plans by analyzing long-term data; When the system detects a change in the long-term trend, it will generate a trend change report and remind the staff to recommend whether to continue the current treatment plan.

[0036] S3: Based on the changing trends of physiological parameters, a customized network layer architecture for pediatric asthma is developed. The learning model and the network layer architecture are introduced to form adversarial training. An evaluation index system for adversarial training is established. Based on the evaluation results, the results of adversarial training are integrated into the electronic medical record system to customize a pediatric asthma early warning intervention plan.

[0037] Furthermore, customized pediatric asthma early warning intervention plans include generating intervention plans based on captured trends in physiological parameters using both the old and new systems; When it is detected that the air quality does not meet the standard, take air purification measures, increase ventilation frequency or install a fresh air system; When it is detected that the air quality has returned to the standard, the network layer architecture for pediatric asthma is customized to complete real-time monitoring of pediatric asthma attack warnings.

[0038] In a preferred embodiment, a pediatric asthma attack warning system based on deep learning is provided, which includes a construction module, which designs a data collection platform for pediatric asthma attacks, obtains health status information, integrates historical medical records in the health status information, builds an electronic medical record system, collects disease data based on the electronic medical record system, trains the disease data, and marks known asthma attack events; an extraction module, which uses the trained disease data as a learning model, and uses the learning model to extract marked asthma attack events, and captures the changing trends of physiological parameters of pediatric asthma through time series analysis of the extracted asthma attack events; an intervention module, which customizes the network layer architecture of pediatric asthma based on the changing trends of physiological parameters, introduces the learning model and the network layer architecture to form adversarial training, establishes an evaluation index system for adversarial training, integrates the results of adversarial training into the electronic medical record system according to the evaluation results, and customizes a pediatric asthma warning intervention plan.

[0039] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0040] In summary, the present invention has built a comprehensive pediatric asthma attack warning system by integrating children's physiological parameters, environmental factors and symptom logs, and uses deep learning models to analyze multi-source data, thereby realizing the precise extraction and prediction of asthma attack events, significantly improving the accuracy and timeliness of the warning. The system can monitor environmental factors such as air quality, temperature and humidity in real time, and provide personalized health advice and intervention measures based on the specific situation of the child. Through the incremental synchronization and timed synchronization mechanism of historical medical records, the consistency and integrity of the data are ensured, supporting doctors to make more scientific treatment decisions. In addition, the system can track the changing trends of the physiological parameters of children over a long period of time, automatically generate trend reports, and help adjust treatment plans. The present invention not only improves the level of intelligence in asthma management, but also enhances the coping capabilities of parents and medical staff, effectively preventing asthma attacks and improving the quality of life of children.

[0041] Example 2: Reference Figure 1 and Figure 2 , which is the second embodiment of the present invention, provides a pediatric asthma attack warning method based on deep learning. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0042] First, in a six-month experiment, 200 children with asthma were recruited as research subjects. Each child was equipped with a smart wearable device for real-time monitoring of physiological parameters such as heart rate, respiratory rate, blood oxygen saturation, body temperature, and galvanic skin response. At the same time, air quality monitors and temperature and humidity sensors were installed in the home to continuously record environmental data such as PM2.5 concentration, PM10 concentration, CO2 concentration, temperature, and relative humidity. In addition, parents used a mobile app to record their children's symptom logs, including the onset and severity of symptoms such as coughing, wheezing, and chest tightness. During the experiment, the system collected physiological parameters and environmental data every 5 minutes and uploaded them to a cloud server for processing. According to a preset algorithm, the system analyzed this data in real time and calculated the air quality index (AQI) and temperature and humidity parameters. When the PM2.5 concentration was 80 micrograms per cubic meter, the system calculated an AQI value of 120, indicating that the air quality was in a state of light pollution. When the temperature was 22 degrees Celsius and the relative humidity was 45%, the system considered the environment comfortable and suitable for daily activities. The experimental data of the present invention are shown in Table 5 below: Table 5 Experimental data table of the present invention Data Type Specific data Remark Experimental cycle 6 months Number of research subjects 200 children with asthma Monitoring equipment Smart Wearable Devices - Air Quality Monitor - Temperature and Humidity Sensor For real-time monitoring of physiological parameters and environmental data Physiological parameter monitoring frequency Collect data every 5 minutes Including heart rate, respiratory rate, blood oxygen saturation, body temperature, skin galvanic response Environmental data monitoring frequency Collect data every 5 minutes Including PM2.5 concentration, PM10 concentration, CO2 concentration, temperature, relative humidity Symptom Logging Mode Parents record via mobile app Record the onset and severity of symptoms such as coughing, wheezing, and chest tightness Data upload frequency Upload to cloud server every 5 minutes Air Quality Index (AQI) When the PM2.5 concentration is 80 micrograms per cubic meter, the AQI value is 120 Indicates that the air quality is slightly polluted Temperature and humidity parameters The temperature is 22 degrees Celsius and the relative humidity is 45%. The system considers the environment comfortable and suitable for daily activities Table 5 can more intuitively understand the specific data collected during the experiment and its corresponding system evaluation results, helping readers better understand the design and implementation of the experiment. The comparison between the present invention and the prior art is shown in Table 6 below: Table 6 Comparison between the present invention and the prior art Comparison items Existing technology Technical solution of the present invention Advantages of the present invention Data Source It mainly relies on doctors' experience and patients' subjective reports, and lacks multi-source data fusion. Integrate multi-source data of physiological parameters, environmental factors, and symptom diaries. Provide a more comprehensive and objective assessment of health status. Monitoring frequency Data collection is infrequent, typically daily or weekly. The system collects and uploads data in real time every 5 minutes. Real-time monitoring ensures data timeliness and accuracy. Early warning mechanism The warning is based on the doctor's experience and the prediction is inaccurate. Using deep learning models for accurate asthma attack prediction and early warning. Intelligent early warning improves forecast accuracy and timeliness. Personalized intervention Interventions are generally generic and lack personalized recommendations. Generate customized intervention plans based on individual data. Provide personalized health advice and interventions. Table 6 can clearly demonstrate the significant advantages of the present invention in terms of data source, monitoring frequency, early warning mechanism and personalized intervention.

[0043] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. 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 pediatric asthma attack warning method based on deep learning, characterized by: include: Design a data collection platform for pediatric asthma attacks, obtain health status information, integrate historical medical records in health status information, build an electronic medical record system, collect disease data based on the electronic medical record system, train the disease data, and mark known asthma attack events; The trained disease data is used as a learning model, and the learning model is used to extract asthma attack events with known labels. The extracted asthma attack events are analyzed through time series to capture the changing trend of physiological parameters of pediatric asthma. Based on the changing trend of physiological parameters, a network layer architecture for pediatric asthma is customized, and the learning model and network layer architecture are introduced to form adversarial training. An evaluation index system for adversarial training is established. Based on the evaluation results, the results of adversarial training are integrated into the electronic medical record system to customize an early warning intervention plan for pediatric asthma.

2. The method for early warning of childhood asthma attacks based on deep learning according to claim 1, characterized in that: The acquisition of health status information includes physiological parameter monitoring, environmental factor monitoring and symptom log recording of children. The physiological parameter monitoring of children includes heart rate parameters, respiratory rate parameters, blood oxygen saturation parameters, body temperature parameters and skin electrical response parameters. The environmental factor monitoring includes air quality parameter monitoring and temperature and humidity parameter monitoring. The symptom log recording includes preprocessing of symptom log records according to physiological parameter monitoring and environmental factor monitoring of children. The preprocessing includes establishing a mobile application development model for symptom log recording, setting an automated reminder function based on the mobile application development model, marking abnormal data points to be reminded, and interactively confirming the marked abnormal data points.

3. The method for early warning of childhood asthma attacks based on deep learning according to claim 2, characterized in that: The air quality parameter includes an air quality parameter calculated by an air quality monitor, and the specific calculation formula for calculating the air quality parameter is: ； in, represents the pollutant concentration, represents the minimum pollutant concentration, Indicates the maximum pollutant concentration, Indicates the minimum pollutant concentration corresponding to index, Indicates the maximum pollutant concentration corresponding to index, Indicates air quality parameters; The temperature and humidity parameters include a comprehensive formula for calculating temperature and relative humidity through a temperature and humidity sensor. The calculation formula of the comprehensive formula for temperature and relative humidity is: ； in, Indicates temperature and humidity parameters, represents the water vapor pressure, , and represents a constant, Indicates temperature; Based on air quality parameters, the data collection platform combines temperature and humidity parameters to provide additional recommendations; When The threshold value is ;when When The threshold value is ,when When The threshold value is ; when When , the environmental factor monitoring is in the first level state; when When , the environmental factor monitoring is in the secondary state; when When , the environmental factor monitoring is in the third level state.

4. The method for early warning of childhood asthma attacks based on deep learning as claimed in claim 3, characterized in that: The construction of the electronic medical record system includes monitoring the status according to different environmental factors, formulating an EMR system for pediatric asthma management, developing an API interface to connect historical medical records with the EMR system, and formulating an incidence data migration plan, wherein the incidence data migration plan includes dividing the data migration into N stages and gradually migrating the data, and judging whether the migrated incidence data is consistent with the unmigrated incidence data; The determination of whether the disease data is consistent with the unmigrated disease data includes analyzing the data synchronization between the new and old systems, and the data synchronization includes an incremental synchronization mechanism, a timed synchronization mechanism, and a bidirectional synchronization mechanism; The incremental synchronization mechanism includes the generation of new medical records and changes to patient information; when a new medical record is generated, a medical record is created or updated in the old system, and the medical record is sent to the new system through the API interface and marked as synchronized, and the timestamp of the medical record operation is recorded in the old system; When patient information changes, it is modified in the old system, triggering an incremental synchronization task to push the latest patient information to the new system. At the same time, check whether there is a record with the same patient ID in the new system. If so, update it, otherwise create a new record; The timing synchronization mechanism includes making a judgment based on a preset time interval and a detected network connection recovery state; When the time interval set by the new and old systems arrives on time, the batch synchronization task is started to extract the new or modified data since the last synchronization from the old system and import it into the new system. At the same time, the data consistency between the new and old systems is compared to ensure that all data has been synchronized; When the network connection is restored after being interrupted, a full synchronization task is triggered to ensure that the data generated during the network failure is not lost. At the same time, the synchronization log is recorded to obtain asthma attack events.

5. The method for early warning of childhood asthma attacks based on deep learning according to claim 4, characterized in that: The learning model includes normalizing the synchronized data, generating a feature vector suitable for inputting the learning model, extracting time series features, wherein the time series features include mean, variance and peak value, and training the onset data includes labeling the asthma attack events, using the labels to perform supervised learning training on the model, and using cross-validation and hyperparameter optimization techniques to complete the training of the learning model; The calculation formula of the mean is: ； in, represents the mean, represents the number of samples, Indicates data points; The specific calculation formula of the variance is: ； in, represents the variance, represents the square of the deviation; The specific calculation formula of the peak value is: ； in, represents the peak value, Represents the maximum value function.

6. The method for early warning of childhood asthma attacks based on deep learning according to claim 5, characterized in that: The capturing of the changing trend of physiological parameters of pediatric asthma includes providing dynamic charts on the old and new systems to display the changing trend of physiological parameters of the children, observing the changes of the heart rate, respiratory rate, and blood oxygen saturation parameters of the children through the charts, generating historical comparison charts, and analyzing the individual differences of the children based on the historical comparison charts; The generating of the historical comparison chart includes displaying the time point of the asthma attack event according to the dynamic chart, selecting different time periods for comparison, at which time, the new and old systems will automatically generate corresponding comparison charts and provide statistical analysis results; Customize the analysis model based on the statistical analysis results provided; when the old and new systems detect abnormal fluctuations in the physiological parameters of pediatric asthma, the old and new systems will automatically call the analysis model, generate an analysis report, and provide health advice; The new and old systems track the physiological parameters of each child with asthma over a long period of time, record their changing trends over time, and adjust treatment plans by analyzing long-term data; When the system detects a change in the long-term trend, it will generate a trend change report and remind the staff to recommend whether to continue the current treatment plan.

7. The method for early warning of childhood asthma attacks based on deep learning according to claim 6, characterized in that: The customized pediatric asthma early warning intervention plan includes generating an intervention plan by the old and new systems based on the captured trend of changes in physiological parameters; When it is detected that the air quality does not meet the standard, take air purification measures, increase ventilation times or install a fresh air system; When it is detected that the air quality has returned to the standard, the network layer architecture for pediatric asthma is customized to complete real-time monitoring of pediatric asthma attack warnings.

8. A pediatric asthma attack warning system based on deep learning, based on the pediatric asthma attack warning method based on deep learning according to any one of claims 1 to 7, characterized in that: include: The construction module designs a data collection platform for pediatric asthma attacks, obtains health status information, integrates historical medical records in the health status information, builds an electronic medical record system, collects disease data based on the electronic medical record system, trains the disease data, and marks known asthma attack events; The extraction module uses the trained disease data as a learning model and uses the learning model to extract asthma attack events with known labels. The extracted asthma attack events are analyzed through time series to capture the changing trend of physiological parameters of pediatric asthma. The intervention module customizes the network layer architecture of pediatric asthma based on the changing trends of physiological parameters, introduces the learning model and the network layer architecture to form adversarial training, establishes an evaluation index system for adversarial training, and integrates the results of adversarial training into the electronic medical record system based on the evaluation results to customize the pediatric asthma early warning intervention plan.

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 pediatric asthma attack warning method based on deep learning are implemented as described in any one of claims 1 to 7.

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 pediatric asthma attack warning method based on deep learning are implemented as described in any one of claims 1 to 7.

Images