AI-based child nebulization treatment management system and method

Abstract

The invention discloses an AI-based child nebulization treatment management system and method, and relates to the field of treatment scheme management, and the system comprises a management and control module which is used for obtaining the medical general archive access and read-write authority of a target medical place, carrying out the editing and sending of a control instruction for each functional module, and sending the control instruction to the medical general archive access and read-write authority to the medical general archive access and read-write authority to the medical general archive access and read-write authority; summarizing and storing working results and analysis data of the functional modules; the data acquisition module is used for acquiring physique data, illness state data and treatment plans of the patient in the current period based on a medical general archive; the feature selection module is used for performing feature recognition on the data acquired by the data acquisition module by adopting a multi-modal feature coupling algorithm; by collecting and analyzing physical characteristics, disease progression, historical treatment response and feedback information of a patient in real time, a personalized treatment scheme is formulated according to specific conditions of each child, a general treatment mode is avoided, unnecessary side effects are reduced, and treatment effectiveness and safety are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of treatment plan management, and specifically to an AI-based pediatric atomization treatment management system and method. Background Art

[0002] Children are at high risk of respiratory diseases, which often require nebulizer inhalation treatment to relieve symptoms. Many regions face the problem of insufficient medical resources, especially in remote areas, where the number of doctors is limited and the medical care needs of children are difficult to meet; In recent years, AI technologies, such as machine learning and deep learning, have developed rapidly and can effectively process and analyze big data to provide support for medical decision-making. Many countries and regions are promoting policies related to digital medicine and personalized medicine, which provides a good environment for the development and application of AI systems. Modern medicine increasingly emphasizes active participation of patients. Parents have higher expectations for their children's treatment and need treatment plans that can respond and adjust in a timely manner. Children differ significantly in physiology, psychology, and immune systems, which often lead to different treatment effects for the same disease. Therefore, personalized treatment plans are needed. Traditional nebulizer treatment decisions often rely on the doctor's experience and subjective judgment, and may be affected by the doctor's knowledge level, personal preferences, and understanding of the drug, which can easily lead to inconsistent treatment results. The treatment standards between different doctors or medical institutions vary greatly. There is a lack of decision support based on quantitative data, which leads to non-optimal use of medical resources and increases the risk of unnecessary side effects. There is a lack of comprehensive evaluation and dynamic adjustment, making it difficult to meet the comprehensive needs of patients. Summary of the invention

[0003] 1. Technical issues to be resolved In view of the above-mentioned shortcomings of the prior art, the present invention provides an AI-based pediatric nebulizer treatment management system and method, which can effectively solve the problems in the prior art.

[0004] (II) Technical solution To achieve the above objectives, the present invention is implemented through the following technical solutions.

[0005] The present invention discloses an AI-based pediatric atomization treatment management system, comprising: The control module is used to obtain access and read and write permissions to the medical archives of the target medical site, and edit and send control instructions to each functional module; The data collection module is used to collect the patient's current period's physical data, condition data and treatment plan based on the general medical archive; A feature selection module is used to perform feature recognition on the data collected by the data collection module using a multi-modal feature coupling algorithm and select key features; The model building module is used to build an indicator adjustment model, retrieve the patient's physical fitness data, disease data, treatment plan data and treatment effect feedback data of several historical periods in the general medical archive as training samples, and conduct indicator adjustment model training; The indicator adjustment module is used to obtain the indicator adjustment model trained by the model building module, input the patient's physical data, disease data and treatment plan into the indicator adjustment model, output the abnormal adjustment coefficients of the medication frequency, dosage and temperature of the treatment plan, and perform a multi-objective optimization engine; The update prediction unit is used to adjust the medication frequency, dosage and temperature of the original treatment plan according to the abnormal adjustment coefficient output by the index adjustment module and the optimal solution of the current atomization efficacy index, and output a new treatment plan; The display module is used to receive the new treatment plan output by the update prediction unit, visualize the results, and support the export to external storage media.

[0006] Furthermore, the feature selection module adopts a multimodal feature coupling algorithm to dynamically segment the physiological parameter time series data into windows, extracts comprehensive time-frequency domain features through wavelet transform, and automatically eliminates feature combinations whose correlation with treatment effect is lower than the preset baseline through adversarial feature verification network. The preset baseline is updated in real time based on online learning, and the complexity of the patient's physiological parameter signal is dynamically judged by calculating the value of the physiological parameter time series data in the window. The window length is automatically expanded or contracted. When the value is greater than the program preset threshold, the window is reduced, and when the value is less than the program preset threshold, the window is expanded. The wavelet transform is parameterized and embedded in the neural network layer. The wavelet basis shape is adaptively optimized through back propagation to match the unique frequency domain pattern of children's respiratory signals. The unique frequency domain pattern includes high, medium, and low frequencies corresponding to wheezing or tidal volume. The continuous wavelet transform and the discrete wavelet transform are combined to integrate the daily nebulization medication response and the instantaneous response of a single nebulization inhalation as the key feature candidate standard.

[0007] Furthermore, the indicator adjustment module dynamically calculates the optimal solution of the current nebulization efficacy index based on real-time biofeedback data, automatically activates gradient compensation when the blood oxygen drops beyond the preset value, and dynamically fine-tunes the centralized weights according to the real-time blood oxygen target and respiratory rate target during the multi-objective optimization engine. When a certain target continuously deviates from the expected range, its weight priority is automatically increased. Before compensation activation, based on the diffusion path of the nebulized drug, the expected blood oxygen recovery curves under different compensation schemes are calculated, and a balance scheme with the shortest recovery time and the minimum drug accumulation amount is selected.

[0008] Furthermore, the model construction module is interactively connected to a model verification module via a wireless network. The model verification module is used to obtain the patient's physical fitness data, disease data, treatment plan data and effect feedback data in the current cycle as an independent verification set, evaluate the indicator adjustment model applied in the previous cycle, calculate the accuracy, recall rate and F1 value indicators, and obtain the calculation results as a reference for adjusting the indicator adjustment model.

[0009] Furthermore, the update prediction unit is deployed with submodules at the lower level, and the submodules include a plan update module, a progress feedback module and a trend prediction module, and the plan update module, the progress feedback module and the trend prediction module are interactively connected through a wireless network, wherein: The plan update module is used to read the abnormal adjustment coefficients of the medication frequency, dosage and temperature of the treatment plan output by the indicator adjustment module, adjust the corresponding indicators, and output a new treatment plan; Progress feedback module, used to obtain the patient's physical and condition data changes under the new treatment plan in real time, and evaluate the treatment effect feedback coefficient based on the change status; The trend prediction module is used to analyze the patient's historical medical data based on the treatment effect feedback coefficient of the new treatment plan, and output the patient's medical data change trend in the future preset period through time series analysis. The S-type function is used to control the single dose and frequency adjustment amplitude of the new treatment plan, and all amplitude risk data are recorded. When the risk data of the adjustment amplitude of the target treatment plan is less than the predetermined standard, it is considered to have passed.

[0010] Furthermore, the management and control module is interactively connected to the data acquisition module through a wireless network, the data acquisition module is interactively connected to the feature selection module through a wireless network, the feature selection module is interactively connected to the model construction module through a wireless network, the model construction module is interactively connected to the indicator adjustment module through a wireless network, the indicator adjustment module is interactively connected to the update prediction unit through a wireless network, and the update prediction unit is interactively connected to the display module through a wireless network.

[0011] An AI-based pediatric atomization treatment management method comprises the following steps: Step 1: Obtain data read and write permissions for the medical archive of the target medical site; Step 2: Collect the patient's current cycle's physical data, condition data, and treatment plan, and perform cleaning, denoising, standardization, and format conversion processing; Step 3: Extract disease characteristics, physical characteristics and treatment characteristics from the processed data; Step 4: Construct an indicator adjustment model, obtain historical patient constitution, condition, treatment plan and feedback to train the indicator adjustment model, and adjust the parameters of the indicator adjustment model based on the training results; Step 5: Input the patient data of the current cycle into the indicator adjustment model, and output the adjustment coefficients of the medication frequency, dosage and atomization temperature of the treatment plan; Step 6: Revise the original treatment plan according to the adjustment coefficient, generate a new personalized treatment plan, and include the corresponding medication frequency, dosage and temperature settings, and update the patient's treatment plan; Step 7: Collect the patient's physical condition and condition change data under the new treatment plan in real time, conduct effect feedback evaluation, and use time series analysis to predict the future development trend of the disease based on the effect feedback evaluation, the patient's physical condition and the patient's condition change data; Step 8: Visualize the collection and output results and provide export function.

[0012] Furthermore, the expression of the indicator adjustment model in step 5 is: ; In the formula, Represents the output of the indicator adjustment model, which represents the set of adjustment coefficients for medication frequency, dosage, and atomization temperature in the treatment plan. represents the adjustment factor for medication frequency, represents the adjustment factor of the drug dosage, Represents the adjustment factor of the medication atomization temperature.

[0013] Furthermore, the expression of the indicator adjustment model , and The calculation formula is: ; ; ; In the formula, The weight coefficient representing the ratio of the current actual value to the baseline value is used to measure the importance of the relationship between the current medication frequency, dosage and atomization temperature actual value and the baseline value. A larger value means that the actual medication situation has a significant impact on the adjustment coefficient. Represents the weight coefficient of the frequency adjustment factor based on patient characteristics, which is used to measure the impact of patient characteristics (such as gender, age, weight, height and physiological parameters) on the adjustment. When b is large, it means that the individual characteristics of the patient have a significant impact on the treatment adjustment. The weight coefficient representing the historical treatment effect factor is used to measure the impact of past treatment effects under similar conditions on the current treatment plan adjustment. If c is higher than the manually preset standard, it indicates that the historical effect is more important for future adjustment decisions, and supports the use of previously successful treatment methods. Represents the weight coefficient of the patient feedback factor, which is used to measure the importance of the patient's feedback after treatment to the adjustment of the treatment plan. When the value is high, it means that the patient's subjective feelings and feedback have a significant impact on the decision-making of drug adjustment, reminding medical staff to pay attention to the patient's experience. Represents the actual frequency of medication in the current cycle, Represents the baseline medication frequency value of the previous cycle, represents frequency adjustment factors based on patient characteristics, including sex, age, weight, height, and physiological parameters, Represents a frequency adjustment factor based on historical treatment effects, reflecting the response of similar conditions to treatments in the past. represents the frequency adjustment factor based on patient feedback, including the patient's subjective feelings and actual reactions to the effect of the last treatment. Represents the actual dosage of medication in the current cycle. Represents the standard dosage value of the previous cycle, represents the dose adjustment factor based on patient characteristics, represents the dose adjustment factor based on historical treatment effects, represents the dose adjustment factor based on patient feedback, Represents the actual atomization temperature of the current cycle. Represents the baseline atomization temperature value of the previous cycle, represents the adjustment factor of the nebulizer temperature based on patient characteristics, represents the atomization temperature adjustment factor based on historical treatment effects, Represents the nebulizer temperature adjustment factor based on patient feedback.

[0014] Furthermore, the expression of the indicator adjustment model , and The calculation logic is: Collect patients’ current medication frequency, dosage, atomization temperature and historical medication effect feedback information; Based on patient characteristics and historical data, appropriate adjustment factors were calculated; The adjustment coefficients of drug frequency, dosage and atomization temperature are calculated in sequence; Will , and Output as a new reference value for treatment plan.

[0015] (III) Beneficial effects Compared with the known prior art, the technical solution provided by the present invention has the following beneficial effects: By collecting and analyzing patients' physical characteristics, disease progression, historical treatment responses and feedback information in real time, we can develop personalized treatment plans based on the specific circumstances of each child, avoid one-size-fits-all treatments, reduce unnecessary side effects, and improve the effectiveness and safety of treatment.

[0016] By dynamically monitoring the patient's response during the treatment process and using the indicator adjustment model to update the treatment plan at any time, we can quickly respond to changes in patients, optimize the treatment plan in a timely manner, ensure that the treatment adapts to the rapidly changing condition, and thus improve the patient's quality of life. The indicator adjustment model is based on big data analysis and historical case data, extracts key information and generates recommended plans, improves the scientificity and accuracy of clinical decision-making, reduces doctors' subjective judgments, and makes treatment plans more reliable and consistent.

[0017] By combining multiple feedbacks on the patient's physical data, disease data, physiological data and treatment effects, a comprehensive treatment effect evaluation is formed, the ability to monitor and evaluate treatment effects is improved, patients can maintain effective communication throughout the treatment process, trust and cooperation between doctors and patients are enhanced, and through systematic data analysis and prediction, help to reasonably arrange subsequent treatment resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 This is a schematic diagram of the framework of the AI-based pediatric nebulizer therapy management system in the present invention.

[0020] The numbers in the figure represent, respectively, 1. Management and control module; 2. Data acquisition module; 3. Feature selection module; 4. Model building module; 5. Index adjustment module; 6. Update prediction unit; 61. Plan update module; 62. Progress feedback module; 63. Trend prediction module; 7. Display module; 8. Data preprocessing module; 9. Model verification module. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] The present invention will be further described below in conjunction with the embodiments.

[0023] Embodiment 1: This embodiment is an AI-based pediatric atomization treatment management system, such as Figure 1 As shown, including: Management and control module 1 is used to obtain access and read and write permissions to the medical archive of the target medical site, edit and send control instructions to each functional module, carry out heterogeneous system compatible interfaces, support cross-platform HL7 / FHIR medical data protocol conversion, and aggregate and store the work results and analysis data of each functional module; through the coordinated operation of each functional module, the system can efficiently process instructions and analyze data, and provide accurate basic data for subsequent modules; The data acquisition module 2 is used to collect the patient's current period's physical data, condition data and treatment plan based on the general medical archives; the data acquisition module 2 is interactively connected to the data preprocessing module 8 through a wireless network, and the data preprocessing module 8 is used to clean, denoise, standardize and convert the collected data, receive feedback on missing and abnormal data for supplementation and modification; and is used to ensure data quality, thereby improving the accuracy of subsequent analysis; The feature selection module 3 is used to perform feature recognition on the data collected by the data acquisition module 2 using a multimodal feature coupling algorithm, dynamically segment the physiological parameter time series data using a multimodal feature coupling algorithm, extract the comprehensive features in the time and frequency domains through wavelet transform, and automatically eliminate the feature combination whose correlation with the treatment effect is lower than the preset baseline through an adversarial feature verification network, update the preset baseline in real time based on online learning, and select key features; by identifying key pathological, physical and treatment features, it can provide important information for personalized treatment, and the selected key features enable the system to better understand the patient's condition; feature selection Module 3 dynamically determines the complexity of the patient's physiological parameter signal by calculating the value of the physiological parameter time series data in the window, automatically expands or shrinks the window length, shrinks the window when the value is greater than the program preset threshold, and expands the window when the value is less than the program preset threshold. The wavelet transform is parameterized and embedded in the neural network layer. The wavelet basis shape is adaptively optimized through back propagation to match the unique frequency domain pattern of the child's respiratory signal. The unique frequency domain pattern includes high, medium and low frequencies corresponding to wheezing or tidal volume. The continuous wavelet transform and discrete wavelet transform are combined to integrate the daily nebulization medication response and the instantaneous response of a single nebulization inhalation as the key feature candidate standard; The key features selected by feature selection module 3 include: Pathological characteristics: type of condition, duration of illness, intensity of symptoms, and treatment history; Physical characteristics: gender, age, weight, height and physiological parameters; Treatment characteristics: medication type, dose, dosing frequency, and nebulization temperature; The model building module 4 is used to build an indicator adjustment model, retrieve the patient's physical fitness data, disease data, treatment plan data and treatment effect feedback data of several historical periods in the general medical archive as training samples, build a dual-branch network architecture, the main branch processes structured medical data, and the auxiliary branch processes unstructured diagnosis and treatment record texts, and train the indicator adjustment model; the model building module 4 is interactively connected to the model verification module 9 through a wireless network, and the model verification module 9 is used to obtain the patient's physical fitness data, disease data, treatment plan data and effect feedback data of the current period as an independent verification set, evaluate the indicator adjustment model applied in the previous period, calculate the accuracy, recall rate and F1 value indicators, and obtain the calculation results as a reference for adjusting the indicator adjustment model; the indicator adjustment model is constructed by training historical data, and the current period data is applied to model evaluation to ensure the accuracy and effectiveness of the model; The index adjustment module 5 is used to obtain the index adjustment model trained by the model construction module 4, input the patient's physical data, disease data and treatment plan into the index adjustment model, output the abnormal adjustment coefficients of the medication frequency, dosage and temperature of the treatment plan, and perform a multi-objective optimization engine to dynamically calculate the optimal solution of the current atomization efficacy index according to the real-time biofeedback data, and automatically activate the gradient compensation when the blood oxygen drops beyond the preset value; thereby facilitating the timely adjustment of the medication frequency, dosage and atomization temperature in actual treatment; During the multi-objective optimization engine, the indicator adjustment module 5 dynamically fine-tunes the weights according to the real-time blood oxygen target and the respiratory rate target. When a certain target continuously deviates from the expected range, its weight priority is automatically increased. Before the compensation is activated, the expected blood oxygen recovery curve under different compensation schemes is calculated based on the diffusion path of the atomized drug, and the balance scheme with the shortest recovery time and the minimum drug accumulation is selected; During the process of automatic activation of gradient compensation, an individualized safety threshold is established based on clinical standards and historical data. The normal range of standard blood oxygen saturation is 92%-98%. The safety threshold is manually adjusted secondary according to the patient's condition. Once the blood oxygen saturation is detected to drop below the critical value of the manually set safety threshold, the automatic activation mechanism is triggered. The optimal threshold adjustment plan is calculated through real-time monitoring data, and the adjustment strategy of drug dosage and frequency is set, that is, the balance plan between the shortest recovery time and the minimum drug accumulation; The update prediction unit 6 is used to adjust the medication frequency, dosage and temperature of the original treatment plan according to the abnormal adjustment coefficient output by the indicator adjustment module 5 and the optimal solution of the current atomization efficacy index, output a new treatment plan, and connect the digital twin simulation system to preview the treatment plan risk; the update prediction unit 6 is deployed with submodules at the lower level, and the submodules include a plan update module 61, a progress feedback module 62 and a trend prediction module 63. The plan update module 61, the progress feedback module 62 and the trend prediction module 63 are interactively connected through a wireless network, wherein: The plan updating module 61 is used to read the abnormal adjustment coefficients of the medication frequency, dosage and temperature of the treatment plan output by the index adjustment module 5, adjust the corresponding indexes, and output a new treatment plan; realize feedback adjustment of medication indexes to maintain the dynamic adaptability of the treatment plan; The progress feedback module 62 is used to obtain the changes in the patient's physical data and condition data under the new treatment plan in real time, and evaluate the treatment effect feedback coefficient based on the change status; continuously monitor the patient's status and provide real-time feedback, thereby improving the effectiveness of the treatment plan; The trend prediction module 63 is used to analyze the patient's historical condition data based on the treatment effect feedback coefficient of the new treatment plan, output the patient's condition data change trend in the future preset period through time series analysis, use the S-type function to control the single dose and frequency adjustment amplitude of the new treatment plan, record all amplitude risk data, and when the risk data of the adjustment amplitude of the target treatment plan is less than the predetermined standard, it is considered to be passed; through historical analysis and time series prediction, potential changes in the condition can be identified and prevented in advance, providing reference for subsequent treatment; The display module 7 is used to receive the new treatment plan output by the update prediction unit 6, visualize the results, and support the export to external storage media; through the intuitive visualization of the treatment plan and its effects, doctors and patients can quickly understand the progress of treatment and provide further data support for medical decision-making.

[0024] As an implementation method in this embodiment, Figure 1 As shown, the management and control module 1 is interactively connected to the data acquisition module 2 through a wireless network, the data acquisition module 2 is interactively connected to the feature selection module 3 through a wireless network, the feature selection module 3 is interactively connected to the model construction module 4 through a wireless network, the model construction module 4 is interactively connected to the indicator adjustment module 5 through a wireless network, the indicator adjustment module 5 is interactively connected to the update prediction unit 6 through a wireless network, and the update prediction unit 6 is interactively connected to the display module 7 through a wireless network.

[0025] In the specific implementation of this embodiment, the global function module is controlled by the management and control module 1 to obtain the access rights to the patient archive of the medical system, the electronic medical records of the target patients are collected by the data collection module 2 to obtain the patient information and treatment plan, the collected data is filtered and filled by the data preprocessing module 8, the key data in the collected data is extracted by the feature selection module 3, the indicator adjustment model is constructed by the model construction module 4, the indicator adjustment module 5 inputs the key data and outputs the adjustment parameters of the treatment plan, the original treatment plan is adjusted based on the adjustment parameters by the update prediction unit 6, the patient's treatment feedback data is collected in real time by the progress feedback module 62, the treatment plan of the future cycle is predicted based on the patient's treatment feedback data and the adjusted treatment plan by the trend prediction module 63, and all output results are displayed by the display module 7.

[0026] Embodiment 2: In other aspects, this embodiment also provides an AI-based pediatric atomization treatment management method, comprising the following steps: Step 1: Obtain data read and write permissions for the medical archive of the target medical site; Step 2: Collect the patient's current cycle's physical data, condition data, and treatment plan, and perform cleaning, denoising, standardization, and format conversion processing; Step 3: Extract disease characteristics, physical characteristics and treatment characteristics from the processed data; Step 4: Construct an indicator adjustment model, obtain historical patient constitution, condition, treatment plan and feedback to train the indicator adjustment model, and adjust the parameters of the indicator adjustment model based on the training results; Step 5: Input the patient data of the current cycle into the indicator adjustment model, and output the adjustment coefficients of the medication frequency, dosage and atomization temperature of the treatment plan; Step 6: Revise the original treatment plan according to the adjustment coefficient, generate a new personalized treatment plan, and include the corresponding medication frequency, dosage and temperature settings, and update the patient's treatment plan; Step 7: Collect the patient's physical condition and condition change data under the new treatment plan in real time, conduct effect feedback evaluation, and use time series analysis to predict the future development trend of the disease based on the effect feedback evaluation, the patient's physical condition and the patient's condition change data; Step 8: Visualize the collection and output results and provide export function.

[0027] Compared with existing technologies, by comprehensively considering the patient's physical data, disease characteristics, historical treatment effects and other personalized factors, a tailored treatment plan can be implemented, which can improve the treatment effect, reduce unnecessary side effects and waste of medical resources, use massive historical data for model training, continuously optimize and adjust strategies, and thus improve the accuracy and reliability of medical decision-making; It has the ability to monitor the patient's condition in real time and can dynamically adjust the treatment plan to better adapt to the patient's changes, thereby quickly responding to the patient's sudden illness or changes and ensuring the timeliness and effectiveness of treatment.

[0028] Embodiment 3: In this embodiment, an expression of an indicator adjustment model is provided, specifically: ; In the formula, Represents the output of the indicator adjustment model, which represents the set of adjustment coefficients for medication frequency, dosage, and atomization temperature in the treatment plan. represents the adjustment factor for medication frequency, represents the adjustment factor of the drug dosage, Represents the adjustment factor of the medication atomization temperature; , and The calculation formula is: ; ; ; In the formula, The weight coefficient representing the ratio of the current actual value to the baseline value is used to measure the importance of the relationship between the current medication frequency, dosage and atomization temperature actual value and the baseline value. A larger value means that the actual medication situation has a significant impact on the adjustment coefficient. Represents the weight coefficient of the frequency adjustment factor based on patient characteristics, which is used to measure the impact of patient characteristics (such as gender, age, weight, height and physiological parameters) on the adjustment. When b is large, it means that the individual characteristics of the patient have a significant impact on the treatment adjustment. The weight coefficient representing the historical treatment effect factor is used to measure the impact of past treatment effects under similar conditions on the current treatment plan adjustment. If c is higher than the manually preset standard, it indicates that the historical effect is more important for future adjustment decisions, and supports the use of previously successful treatment methods. Represents the weight coefficient of the patient feedback factor, which is used to measure the importance of the patient's feedback after treatment to the adjustment of the treatment plan. When the value is high, it means that the patient's subjective feelings and feedback have a significant impact on the decision-making of drug adjustment, reminding medical staff to pay attention to the patient's experience. Represents the actual frequency of medication in the current cycle, Represents the baseline medication frequency value of the previous cycle, represents frequency adjustment factors based on patient characteristics, including sex, age, weight, height, and physiological parameters, Represents a frequency adjustment factor based on historical treatment effects, reflecting the response of similar conditions to treatments in the past. represents the frequency adjustment factor based on patient feedback, including the patient's subjective feelings and actual reactions to the effect of the last treatment. Represents the actual dosage of medication in the current cycle. Represents the standard dosage value of the previous cycle, represents the dose adjustment factor based on patient characteristics, represents the dose adjustment factor based on historical treatment effects, represents the dose adjustment factor based on patient feedback, Represents the actual atomization temperature of the current cycle. Represents the baseline atomization temperature value of the previous cycle, represents the adjustment factor of the nebulizer temperature based on patient characteristics, represents the atomization temperature adjustment factor based on historical treatment effects, represents the nebulizer temperature adjustment factor based on patient feedback; In the expression of the indicator adjustment model , and The calculation logic is: Collect patients’ current medication frequency, dosage, atomization temperature and historical medication effect feedback information; Based on patient characteristics and historical data, appropriate adjustment factors were calculated; The adjustment coefficients of drug frequency, dosage and atomization temperature are calculated in sequence; Will , and Output as a new reference value for treatment plan.

[0029] The above calculation formula combines multiple data sources and uses dynamic feedback and personalized information to provide an advanced, flexible and precise adjustment plan for the management of nebulizer treatment for children. This design can not only improve the treatment effect, but also help improve the patient experience and promote the implementation of personalized medicine.

[0030] In summary, the present invention uses AI technology to analyze a large amount of patient historical data, and find the best treatment plan through algorithm learning, providing personalized management based on data, being able to collect and process data in real time during the treatment process, and adjusting the treatment plan in time according to the patient's response, making the treatment more flexible and adaptable; It can collect and process data in real time during the treatment process, adjust the treatment plan in time according to the patient's response, making the treatment more flexible and adaptable, and can provide medical staff with suggestions based on data analysis to reduce their decision-making burden. At the same time, it can improve the consistency and standardization of treatment plans, and can significantly improve the safety and effectiveness of nebulizer treatment for children, meet the individual needs of different patients, optimize treatment results, and improve the overall quality of care.

[0031] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An AI-based pediatric atomization treatment management system, characterized in that: include: The control module (1) is used to obtain access and read and write permissions to the medical archive of the target medical site, and edit and send control instructions to each functional module; The data collection module (2) is used to collect the patient's current period physical data, disease data and treatment plan based on the medical archive; A feature selection module (3) is used to perform feature recognition on the data collected by the data collection module (2) using a multi-modal feature coupling algorithm to obtain key features; The model building module (4) is used to build an indicator adjustment model, retrieve the patient's physical condition data, disease condition data, treatment plan data and treatment effect feedback data of several historical periods in the medical archive as training samples, and perform indicator adjustment model training; The indicator adjustment module (5) is used to obtain the indicator adjustment model trained by the model building module (4), input the patient's physical data, disease data and treatment plan into the indicator adjustment model, output the abnormal adjustment coefficients of the medication frequency, dosage and temperature of the treatment plan, and perform a multi-objective optimization engine; An updating prediction unit (6) is used to adjust the medication frequency, dosage and temperature of the original treatment plan according to the abnormal adjustment coefficient output by the index adjustment module (5) and the optimal solution of the current atomization efficacy index, and output a new treatment plan; The display module (7) is used to receive the new treatment plan output by the update prediction unit (6), perform visual display of the results, and support export to an external storage medium.

2. The AI-based pediatric atomization treatment management system according to claim 1, characterized in that: The feature selection module (3) adopts a multimodal feature coupling algorithm to dynamically window segment the patient's current cycle of physical data, disease data and treatment plan time series data, extracts time-frequency domain comprehensive features through wavelet transform, and automatically eliminates feature combinations with a correlation with treatment effect lower than a preset baseline through an adversarial feature verification network, updates the preset baseline in real time based on online learning, dynamically determines the complexity of the patient's physiological parameter signal by calculating the value of the physiological parameter time series data in the window, automatically expands or shrinks the window length, shrinks the window when the value is greater than the program preset threshold, and expands the window when the value is less than the program preset threshold, parameterizes the wavelet transform, embeds it into a neural network layer, adaptively optimizes the wavelet basis shape through back propagation, matches the unique frequency domain pattern of the child's respiratory signal, the unique frequency domain pattern includes high, medium and low frequencies corresponding to wheezing or tidal volume, combines continuous wavelet transform with discrete wavelet transform, integrates the daily nebulization medication response and the instantaneous response of a single nebulization inhalation as the key feature candidate standard.

3. The AI-based pediatric atomization treatment management system according to claim 1, characterized in that: The index adjustment module (5) dynamically calculates the optimal solution of the current atomization efficacy index based on the real-time biofeedback data, automatically activates gradient compensation when the blood oxygen level drops beyond a preset value, and dynamically fine-tunes the weight of the target according to the real-time blood oxygen level target and the respiratory rate target during the multi-objective optimization engine process. When a certain target continuously deviates from the expected range, its weight priority is automatically increased. Before the compensation is activated, the expected blood oxygen recovery curve under different compensation schemes is calculated based on the diffusion path of the atomized drug, and a balance scheme with the shortest recovery time and the minimum drug accumulation amount is selected.

4. The AI-based pediatric atomization treatment management system according to claim 1, characterized in that: The model building module (4) is interactively connected to a model verification module (9) via a wireless network. The model verification module (9) is used to obtain the patient's physical fitness data, disease condition data, treatment plan data and effect feedback data in the current cycle as an independent verification set, evaluate the indicator adjustment model applied in the previous cycle, calculate the accuracy, recall rate and F1 value indicators, and obtain the calculation results as a reference for adjusting the indicator adjustment model.

5. The AI-based pediatric atomization treatment management system according to claim 1, characterized in that: The update prediction unit (6) has submodules deployed at a lower level, the submodules including a plan update module (61), a progress feedback module (62) and a trend prediction module (63), the plan update module (61), the progress feedback module (62) and the trend prediction module (63) are interactively connected via a wireless network, wherein: A plan updating module (61) is used to read the abnormal adjustment coefficients of the medication frequency, dosage and temperature of the treatment plan output by the indicator adjustment module (5), adjust the corresponding indicators, and output a new treatment plan; A progress feedback module (62) is used to obtain in real time changes in the patient's physical fitness data and condition data under the new treatment plan, and to evaluate the treatment effect feedback coefficient based on the change status; The trend prediction module (63) is used to analyze the patient's historical medical data based on the treatment effect feedback coefficient of the new treatment plan, output the patient's medical data change trend in the future preset period through time series analysis, use the S-type function to control the single dose and frequency adjustment amplitude of the new treatment plan, record all amplitude risk data, and when the risk data of the adjustment amplitude of the target treatment plan is less than the predetermined standard, it is considered to be passed.

6. The AI-based pediatric atomization treatment management system according to claim 1, characterized in that: The control module (1) is interactively connected to the data acquisition module (2) via a wireless network, the data acquisition module (2) is interactively connected to the feature selection module (3) via a wireless network, the feature selection module (3) is interactively connected to the model construction module (4) via a wireless network, the model construction module (4) is interactively connected to the indicator adjustment module (5) via a wireless network, the indicator adjustment module (5) is interactively connected to the update prediction unit (6) via a wireless network, and the update prediction unit (6) is interactively connected to the display module (7) via a wireless network.

7. An AI-based pediatric nebulization therapy management method, the method is an implementation method of the AI-based pediatric nebulization therapy management system according to any one of claims 1 to 6, characterized in that: The following steps are involved: Step 1: Obtain data read and write permissions for the medical archive of the target medical site; Step 2: Collect the patient's current cycle's physical data, condition data, and treatment plan, and perform cleaning, denoising, standardization, and format conversion processing; Step 3: Extract disease characteristics, physical characteristics and treatment characteristics from the processed data; Step 4: Construct an indicator adjustment model, obtain historical patient constitution, condition, treatment plan and feedback to train the indicator adjustment model, and adjust the parameters of the indicator adjustment model based on the training results; Step 5: Input the patient data of the current cycle into the indicator adjustment model, and output the adjustment coefficients of the medication frequency, dosage and atomization temperature of the treatment plan; Step 6: Revise the original treatment plan according to the adjustment coefficient, generate a new personalized treatment plan, and include the corresponding medication frequency, dosage and temperature settings, and update the patient's treatment plan; Step 7: Collect the patient's physical condition and condition change data under the new treatment plan in real time, conduct effect feedback evaluation, and use time series analysis to predict the future development trend of the disease based on the effect feedback evaluation, the patient's physical condition and the patient's condition change data; Step 8: Visualize the collection and output results and provide export function.

8. The AI-based pediatric aerosol therapy management method according to claim 7, characterized in that: The expression of the indicator adjustment model in step 5 is: ； In the formula, Represents the output of the indicator adjustment model, which represents the set of adjustment coefficients for medication frequency, dosage, and atomization temperature in the treatment plan. represents the adjustment factor for medication frequency, represents the adjustment factor of the drug dosage, Represents the adjustment factor of the medication atomization temperature.

9. The AI-based pediatric aerosol therapy management method according to claim 8, characterized in that: The indicator adjustment model expression is , and The calculation formula is: ； ； ； In the formula, The weight coefficient representing the ratio of the current actual value to the baseline value, represents the weight coefficient of the frequency adjustment factor based on patient characteristics, The weight coefficient representing the historical treatment effect factor is used to measure the impact of past treatment effects under similar conditions on the current treatment plan adjustment. If c is higher than the manually preset standard, it indicates that the historical effect is more important for future adjustment decisions, and supports the use of previously successful treatment methods. represents the weight coefficient of the patient feedback factor, Represents the actual frequency of medication in the current cycle, Represents the baseline medication frequency value of the previous cycle, represents the frequency adjustment factor based on patient characteristics, represents a frequency adjustment factor based on historical treatment effects, represents the frequency adjustment factor based on patient feedback, Represents the actual dosage of medication in the current cycle. Represents the standard dosage value of the previous cycle, represents the dose adjustment factor based on patient characteristics, represents the dose adjustment factor based on historical treatment effects, represents the dose adjustment factor based on patient feedback, Represents the actual atomization temperature of the current cycle. Represents the baseline atomization temperature value of the previous cycle, represents the adjustment factor of the nebulizer temperature based on patient characteristics, represents the atomization temperature adjustment factor based on historical treatment effects, Represents the nebulizer temperature adjustment factor based on patient feedback.

10. The AI-based pediatric aerosol therapy management method according to claim 8, characterized in that: The indicator adjustment model expression is , and The calculation logic is: Collect patients’ current medication frequency, dosage, atomization temperature and historical medication effect feedback information; Based on patient characteristics and historical data, appropriate adjustment factors were calculated; The adjustment coefficients of drug frequency, dosage and atomization temperature are calculated in sequence; Will , and Output as a new reference value for treatment plan.

Images