AI-based intelligent monitoring and reminding system and method for medicine inhaled by children for asthma

Abstract

The invention discloses an AI-based intelligent monitoring and reminding system and method for medicine inhaled for children asthma, and relates to the technical field of intelligent medical monitoring and AI application, the system comprises the following components: an intelligent monitoring terminal, an AI analysis platform, a multi-terminal reminding module and a medical cooperation terminal; through the intelligent monitoring terminal integrated on the asthma inhalation medicine container, various operation data and environment data in the medicine taking process of children can be accurately collected in real time, and the data are deeply processed and analyzed through the AI analysis platform; the system can accurately evaluate the medication operation standardization, the actual inhalation dose deviation and the medication compliance level of children, the system sends out a prompt signal at the preset medication time through the multi-terminal reminding module, and immediately sends out a warning signal when the operation is not standard, the dose is out of tolerance or the health risk exceeds the standard, so that the medication compliance of the children is effectively improved, and the medication compliance of the children is improved. And the medicine can be correctly inhaled on time according to the quantity, so that the treatment effect is improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent medical monitoring and AI application technology, specifically to an AI-based intelligent monitoring and reminder system and method for children's asthma inhaled medications. Background Technology

[0002] With the increasing severity of environmental pollution and the growing prevalence of allergies among children, the incidence of childhood asthma is showing a year-on-year upward trend. As a common chronic respiratory disease, asthma seriously affects children's physical and mental health and quality of life. In the treatment of childhood asthma, inhaled medications are one of the main means of controlling symptoms and preventing attacks. However, due to their young age and weak self-management ability, children often find it difficult to accurately master the techniques of using inhaled medications, leading to improper medication use and inaccurate dosage, which in turn affects the treatment effect. In addition, parents and doctors also find it difficult to understand children's medication use and health status in a timely and accurate manner, making it difficult to adjust the treatment plan without a scientific basis.

[0003] Traditional methods for monitoring inhaled medications in children with asthma rely primarily on parental supervision and children's self-reporting, which have several limitations. First, parental supervision is limited by time and energy, making it impossible to monitor continuously and comprehensively, which can easily lead to missed or improper medication administration. Second, children's self-reporting is highly subjective and easily influenced by memory bias and emotions, resulting in inaccurate medication data. Furthermore, traditional methods lack objective assessment of medication administration procedures, making it difficult to identify specific problems during medication administration, such as incorrect inhalation posture or insufficient inhalation time. Additionally, traditional methods cannot integrate environmental data and historical medication data in real time to predict future health risks for children, leaving doctors without foresight and scientific rigor when adjusting treatment plans.

[0004] Given the shortcomings of traditional methods for monitoring inhaled medications in children with asthma, it is therefore of great importance to develop an AI-based intelligent monitoring and reminder system and method for inhaled medications in children with asthma. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide an AI-based intelligent monitoring and reminder system and method for pediatric asthma inhaled medication. This system uses an intelligent monitoring terminal integrated into the asthma inhaled medication container to collect real-time and accurate operational and environmental data during the child's medication administration process. The system utilizes an AI analysis platform for in-depth processing and analysis, assessing the child's medication administration compliance, actual inhaled dose deviation, and medication adherence level. A multi-terminal reminder module issues alerts at preset medication times and immediately issues warnings when there is improper operation, dosage deviation, or excessive health risk. Furthermore, the system can combine historical medication data and environmental data to predict the probability of asthma attacks in the child over the next 7 days, providing doctors with a scientific basis to adjust treatment plans. This system effectively improves children's medication adherence, ensures correct medication inhalation, enhances treatment efficacy, and provides strong support for personalized and precise treatment of childhood asthma.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: On the one hand, an AI-based intelligent monitoring and reminder system for inhaled medications for childhood asthma, the system comprising the following components: an intelligent monitoring terminal, an AI analysis platform, a multi-terminal reminder module, and a medical collaboration terminal;

[0007] The intelligent monitoring terminal is detachably mounted on the asthma inhalation drug container and integrates a pressure sensor, an angle sensor, an airflow sensor, a facial image acquisition component, and an environmental sensor. The airflow sensor uses dual-channel airflow acquisition technology to simultaneously acquire the axial and radial airflow velocities at the inhalation port. The angle sensor has a built-in accelerometer sensor and supports dynamic benchmark calibration. It is used to collect operational and environmental data during children's medication administration.

[0008] The AI ​​analysis platform, deployed on a cloud server, establishes a real-time data transmission link with the intelligent monitoring terminal. It includes a data preprocessing module, an operational compliance judgment sub-model, a dose deviation calculation sub-model, a compliance assessment sub-model, and a health risk prediction sub-model. The data preprocessing module employs an improved outlier removal algorithm; the operational compliance judgment sub-model uses a multi-dimensional temporal alignment algorithm; the dose deviation calculation sub-model uses a drug residue correction algorithm; the compliance assessment sub-model uses a behavioral pattern recognition algorithm; and the health risk prediction sub-model uses a spatiotemporal fusion algorithm. It receives operational data and environmental data, sequentially performs data cleaning, noise reduction, and normalization preprocessing, inputs the data into a multi-task deep learning model, and outputs medication operation compliance results, actual inhaled dose deviation results, and medication compliance levels. It also generates a health risk assessment report by combining historical medication data and environmental data.

[0009] The multi-terminal reminder module includes a local audio-visual unit integrated into the smart monitoring terminal and a parent-side mobile application. It is used to issue reminder signals at preset medication times, issue warning signals when the operation is not standardized, the dosage is out of control, or the health risk exceeds the standard, and push medication plans, abnormal notifications, and health risk reports.

[0010] The medical collaboration terminal includes a doctor-side web platform and a mobile terminal, with a built-in medication plan optimization module. The medication plan optimization module uses three-dimensional feature matching + dynamic weighting technology to obtain medication data, evaluation results and health risk reports output by the AI ​​analysis platform, so that doctors can adjust the medication plan and push remote guidance information to the parent-side mobile application.

[0011] Furthermore, the airflow sensing component of the intelligent monitoring terminal adopts dual-channel airflow acquisition technology to simultaneously acquire the axial and radial airflow velocities at the inlet, and calculates the airflow stability using an airflow stability index algorithm. Where S 气流 v is the airflow stability index. i Let v be the airflow velocity at the i-th sampling point. avg The average airflow velocity during the sampling period is n, where n is the number of sampling points and k is k. corr The correction coefficient was determined through clinical trials: standardized medication tests were conducted on 30 children with asthma, 1200 sets of airflow data were collected, and the correction coefficient k was obtained through polynomial fitting. corr =1.2, so that the error in judging airflow stability is controlled within ±4%, when S 气流 A value ≥0.85 indicates stable airflow, 0.7-0.84 indicates basically stable airflow, and <0.7 indicates unstable airflow. The model is specifically optimized for the airflow fluctuation characteristics during children's inhalation and sets a dynamic threshold for children under 6 years old to address the problem of weak airflow control ability in young children. The recognition accuracy is improved by 32% compared to the traditional standard deviation algorithm.

[0012] Furthermore, the data preprocessing module of the AI ​​analysis platform employs an improved outlier removal algorithm, combining the Raida criterion with a dynamic threshold mechanism: P 异常 =|x i -μ|>3σ+α×μ, where P 异常 For the outlier determination result, x i Here, μ represents the original data value, σ represents the standard deviation, and α is a dynamic adjustment factor with a value of 0.15. The algorithm determines these parameters by analyzing 5000 sets of clinically collected pressure, angle, and airflow data: when the data distribution shows a significant right skew, α is automatically increased to 0.2; when the data approximates a normal distribution, α is decreased to 0.1. For time-series data, a continuity check is added: if three consecutive data points satisfy P... 异常Only when the conditions and trends are consistent are values ​​identified as outliers; otherwise, they are marked as suspicious values ​​and retained. Compared with the traditional 3σ criterion, the outlier misjudgment rate is reduced by 45%, and the accuracy rate for identifying transient interferences in children's medication process reaches 97.6%, ensuring that the effective data retention rate is ≥95%.

[0013] Furthermore, the operational compliance judgment sub-model of the AI ​​analysis platform employs a multi-dimensional temporal alignment algorithm to dynamically match actual operational data with standard templates: Where M 匹配 For the matching degree, d j Let w be the Euclidean distance between the data in the j-th dimension and the standard template. j For dimensional weights, τ j The dimension decay coefficient is w, and the weight is w. j The attenuation coefficient τ was determined by evaluation of 15 pediatric respiratory specialists using the Delphi method. j Analysis of variance based on 800 standard operating procedures showed that when M... 匹配 A score ≥0.85 is considered standard, and the matching degree of each dimension is output; a score of 70-84 indicates slight deviation, and the dimension with the largest deviation is marked; a score <70 indicates severe deviation, and the problem items are output in order of deviation degree. The model introduces a time elastic alignment mechanism, allowing an operation time deviation of ±1.5 seconds to adapt to the differences in operation rhythm among different children, and the matching accuracy rate reaches 96.8%.

[0014] Furthermore, the dose deviation calculation sub-model of the AI ​​analysis platform adopts a drug residue correction algorithm, and the actual inhaled dose calculation formula is: D 实际 =(D 理论 ×η 释放 )-(R 残留 ×k 回收 ), where D 实际 D is the actual inhaled dose. 理论 η is the theoretical dose produced by pressing. 释放 For drug release efficiency, R 残留 k represents the amount of residual drug in the container. 回收 The residual drug recovery coefficient was determined through in vitro experiments on 20 mainstream asthma drugs: drug residues were measured under different pressure and humidity conditions, and the coefficient value was obtained through multiple regression analysis.

[0015] Furthermore, the compliance assessment sub-model of the AI ​​analysis platform employs a behavioral pattern recognition algorithm, combined with medication time distribution characteristics, to calculate the compliance index: Where C 依从 T is a compliance index. 按时 To ensure timely medication administration, T 总σ represents the total number of times the drug is applied. t S represents the standard deviation of medication time, ΔT represents the medication interval, and S represents the standard deviation of medication time. 连续 σ is the ratio of continuous medication days to total days, γ is the continuous medication reward coefficient, and the model parameters were optimized by analyzing medication data from 300 children over one year: t The weighting of the / ΔT term is set so that the contribution of time regularity accounts for 30%, S 连续 The item encourages the formation of medication habits, when C 依从 ≥0.9 indicates high adherence, 0.7-0.89 indicates medium adherence, and <0.7 indicates low adherence. A special seasonal adjustment factor was designed: γ was increased to 0.3 in spring and autumn to emphasize the importance of continuous medication. At the same time, the trend of adherence changes was recorded. If the decline exceeds 15% for two consecutive weeks, it was automatically marked as a deterioration in adherence and the cause was analyzed.

[0016] Furthermore, the health risk prediction sub-model of the AI ​​analysis platform uses a spatiotemporal fusion algorithm to calculate the probability of an asthma attack in the next 7 days: Where P 风险 X represents the probability of an asthma attack in the next 7 days. β0 is a constant term with a value of -2.35, and β1, β2, β3, and β4 are feature weights with values ​​of 1.82, 1.56, 2.13, and 0.97, respectively. 用药 To standardize medication adherence scoring, X 环境 X is the environmental risk index. 症状 X is the score for signs and symptoms over the past 3 days. 季节 The seasonal factor was set at 1.0 for spring and autumn, 0.6 for summer, and 0.8 for winter. The parameters were obtained by training logistic regression on the incidence data of 1200 children with asthma and dynamically adjusted using Bayesian optimization.

[0017] Furthermore, the angle sensing component of the intelligent monitoring terminal adopts dual position reference + real-time compensation technology to realize real-time correction of the container tilt angle measurement value. The specific steps are as follows: two miniature tilt sensors are symmetrically installed on the outer side of the middle of the container. The main sensor collects real-time angle data at a sampling frequency of 30Hz, and the secondary sensor simultaneously collects pressure distribution data of the container sidewall. The system presets a pressure distribution threshold corresponding to the standard medication angle range. If the angle detected by the main sensor deviates from the standard range during medication, the secondary sensor immediately compares the deviation between the current pressure distribution and the threshold and calculates the compensation coefficient. The angle value of the main sensor is multiplied by the compensation coefficient to obtain the correction amount, which is superimposed on the original angle value of the main sensor to generate the final corrected angle. The whole process is completed within 0.5 seconds, and the angle measurement error after correction is ≤±0.3°, which is suitable for the shaking scenario when children hold the container.

[0018] Furthermore, the medication optimization module of the medical collaboration terminal uses three-dimensional feature matching + dynamic weighting technology to recommend medication frequency adjustments. The specific steps are as follows: The system constructs a pediatric asthma medication feature database, containing four core features: age, weight, symptom improvement rate in the past 30 days, and previous adverse drug reaction records. When administering medication, the system extracts and standardizes the four feature data of the current child. The standardized data is matched with the medication frequency and efficacy data of more than 1,000 similar cases in the feature database, and the matching degree is calculated. Cases with a matching degree ≥ 85% are included in the candidate set. The medication frequency of the candidate set cases is calculated as a weighted average based on symptom improvement rate, adverse reaction incidence rate, and medication adherence, which serves as the initial recommended frequency. The final recommended frequency is generated by combining the current health risk level of the child, with a 10% increase for high-risk cases and an 8% decrease for low-risk cases. The entire calculation process is completed within 2 seconds, and the recommended frequency fit reaches more than 92%.

[0019] On the other hand, an AI-based intelligent monitoring and reminder method for inhaled medications in children with asthma is characterized by the following specific steps:

[0020] S1. Data Acquisition Steps: Through a smart monitoring terminal that is detachably mounted on the asthma inhalation drug container, the operation data and environmental data during the child's medication process are collected simultaneously. The operation data includes the pressure applied, container tilt angle, inhalation airflow parameters, and facial posture data. The environmental data includes the temperature, relative humidity, and PM2.5 concentration of the medication environment.

[0021] S2. Data processing and analysis steps: The collected operational data and environmental data are transmitted to the AI ​​analysis platform via wireless communication. Data cleaning, noise reduction, and normalization preprocessing are performed in sequence. Then, the data is input into the preset multi-task deep learning model, and the results of medication operation standardization, actual inhaled dose deviation, and medication compliance level are output. A health risk assessment report is generated by combining historical medication data and environmental data.

[0022] S3. Reminder and Notification Steps: Tiered reminders are implemented through a multi-terminal reminder module. The local audio-visual unit of the multi-terminal reminder module issues a prompt signal at the preset medication time and an alarm signal when the operation is not standardized, the dosage is out of control, or the health risk exceeds the standard. The parent's mobile application receives medication plan push, abnormal situation notification and health risk report.

[0023] S4. Medical Collaboration Steps: The medical collaboration terminal obtains medication data, assessment results, and health risk reports output by the AI ​​analysis platform, which are then used by doctors to adjust medication plans and push remote guidance information through the parents' mobile application.

[0024] Compared with existing technologies, this AI-based intelligent monitoring and reminder system and method for childhood asthma inhaled medication has the following beneficial effects:

[0025] I. This system, through an intelligent monitoring terminal integrated into the asthma inhalation drug container, can collect various operational and environmental data during children's medication administration in real time and with high accuracy. After in-depth processing and analysis by the AI ​​analysis platform, this data can accurately assess the standardization of children's medication operation, deviation of actual inhaled dosage, and medication adherence level. The system issues prompt signals at preset medication times through a multi-terminal reminder module, and immediately issues warning signals when the operation is not standardized, the dosage is out of control, or the health risk exceeds the standard. This effectively improves children's medication adherence, ensures that the drug can be inhaled correctly on time and in the correct dosage, thereby improving the treatment effect.

[0026] Second, this system, through its AI analysis platform, can not only analyze current medication data in real time, but also combine historical medication data with environmental data. Utilizing advanced spatiotemporal fusion algorithms, it predicts the probability of asthma attacks in children over the next 7 days, generating detailed health risk assessment reports. These reports and data analysis results are available for remote viewing by doctors via medical collaboration terminals, enabling them to promptly understand changes in the child's condition and medication status. Doctors can adjust medication plans based on this data-driven information and push remote guidance information to parents through a mobile application, achieving personalized and precise treatment. This data-driven health risk assessment and remote medical collaboration model greatly optimizes treatment plans for childhood asthma, improving treatment effectiveness and safety.

[0027] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0029] Figure 1 This is a schematic diagram of the architecture of an AI-based intelligent monitoring and reminder system for inhaled medications in children with asthma.

[0030] Figure 2 This is a flowchart illustrating an AI-based intelligent monitoring and reminder method for inhaled medications in children with asthma. Detailed Implementation

[0031] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0032] Example 1

[0033] A smart monitoring terminal was installed on a pressurized metered-dose inhaler for a 5-year-old child with asthma. This terminal features a detachable design and can be securely mounted on the outside of the medication container. During the child's regular medication administration times of 8:00 AM and 8:00 PM, the terminal activates its various sensing components to collect data: the pressure sensing component captures the force with which the child presses the medication container in real time; the angle sensing component incorporates an accelerometer and supports dynamic benchmark calibration, simultaneously recording changes in the container's tilt angle; and the airflow sensing component utilizes dual-channel airflow acquisition technology to simultaneously acquire axial and radial airflow velocity data from the inhalation port and calculates airflow stability using an airflow stability index algorithm. Where S 气流 v is the airflow stability index. i Let v be the airflow velocity at the i-th sampling point. avg The average airflow velocity during the sampling period is n, where n is the number of sampling points and k is k. corr To correct the coefficients; the facial image acquisition component captures the child's facial posture during medication to determine if there are any problems such as improper mouth opening; the environmental sensing component collects environmental data such as temperature, relative humidity, and PM2.5 concentration in the medication environment. All operational data and environmental data are stored in the terminal's local cache in real time during the acquisition process to ensure that no data is lost.

[0034] The intelligent monitoring terminal transmits the collected raw data in real time to an AI analysis platform deployed on a cloud server via a wireless communication link. First, the platform's data preprocessing module employs an improved outlier removal algorithm, combining the Laida criterion and a dynamic threshold mechanism. The formula is: P 异常 =|x i -μ|>3σ+α×μ, where P 异常 For the outlier determination result, x i The original data values ​​are represented by μ, σ, and α. The original data is cleaned to remove abnormal data caused by accidental touches by the child or momentary interference from the sensor. Then, noise reduction and normalization are performed to adjust the data to a uniform standard range.

[0035] The processed standardized data is input into a pre-defined multi-task deep learning model: the operation standardization judgment sub-model uses a multi-dimensional temporal alignment algorithm to dynamically match the child's actual operation data with the standard medication operation template to determine whether the medication operation is standardized. The formula is: Where M匹配 For the matching degree, d j Let w be the Euclidean distance between the data in the j-th dimension and the standard template. j For dimensional weights, τ j The dimensional attenuation coefficient is used. The dose deviation calculation sub-model employs a drug residue correction algorithm, combining parameters such as the theoretical dose generated by pressing, drug release efficiency, residual drug amount in the container, and residual drug recovery coefficient to calculate the deviation between the actual inhaled dose and the theoretical dose. The compliance assessment sub-model uses a behavioral pattern recognition algorithm, combining medication time distribution characteristics to calculate the compliance index and determine the medication compliance level. The health risk prediction sub-model uses a spatiotemporal fusion algorithm, with the formula: Where P 风险 X represents the probability of an asthma attack in the next 7 days, where β0 is a constant, and β1, β2, β3, and β4 are feature weights. 用药 To standardize medication adherence scoring, X 环境 X is the environmental risk index. 症状 X is the score for signs and symptoms over the past 3 days. 季节 Using seasonal factors, combined with the child's historical medication data, current environmental data, physical signs and symptoms scores over the past 3 days, and seasonal factors, a health risk assessment report on the risk of asthma attacks in the next 7 days is generated.

[0036] The multi-terminal reminder module activates a tiered reminder mechanism based on the output of the AI ​​analysis platform. Fifteen minutes before the preset medication times of 8:00 AM and 8:00 PM, the local audio-visual unit of the intelligent monitoring terminal emits a soft beeping sound and a flashing green light to remind the child and parents to prepare for medication. If the AI ​​analysis platform determines that the operation is not standardized, the dosage is out of control, or the health risk assessment report shows that the risk exceeds the standard when the child is taking medication, the local audio-visual unit immediately switches to a rapid beeping warning sound and a solid red light. At the same time, the parent's mobile application receives push information in real time, including details of the medication plan for the day, notifications of abnormal situations such as non-standard operation / dosage deviation, and a complete health risk assessment report, so that parents can keep track of the child's medication status in a timely manner.

[0037] The medical collaboration terminal's doctor-side web platform and mobile terminal obtain real-time medication data, various assessment results, and health risk assessment reports from the AI ​​analysis platform. Doctors can view this information through the terminal and, combined with the child's past medical history, if they find that the child has recently experienced multiple instances of improper operation leading to insufficient dosage, low compliance, and an upward trend in health risk, they can adjust the medication plan through the medication plan optimization module. This module uses three-dimensional feature matching + dynamic weighting technology to first extract and standardize four types of feature data: the child's current medication characteristics, physical signs, environmental characteristics, etc., and match them with the medication plans and efficacy data of more than 1,000 similar cases in the feature database. After calculating the matching degree, a weighted average is calculated for the candidate medication plans based on symptom improvement rate, adverse reaction incidence rate, and medication compliance. Combined with the child's current health risk level, a final recommended medication plan is generated. Subsequently, the doctor pushes remote guidance information through the parent's mobile application, including details of the adjusted medication plan, links to correct medication operation videos, and daily care suggestions, ensuring that parents can accurately implement the new medication plan and maintain communication with the doctor.

[0038] Example 2

[0039] A smart monitoring terminal was installed on the dry powder inhaler carried to school by an 8-year-old child with asthma. The terminal features a lightweight and detachable design, adapted to the container structure of the dry powder inhaler. During the child's daily medication time at noon at school, the terminal automatically activates various sensing components: the pressure sensor detects the force and duration of the child pressing the dry powder inhaler's dosing button; the angle sensor uses dual-position reference + real-time compensation technology, with two miniature tilt sensors symmetrically installed on the outer side of the container to accurately measure the container's tilt angle during medication; the airflow sensor uses dual-channel airflow acquisition technology to simultaneously acquire the axial and radial airflow velocity at the inhalation port to determine if the inhalation action meets the standard; the facial image acquisition component captures facial posture data such as whether the child's face is aligned with the inhalation port and whether there is any improper breathing coordination during medication; the environmental sensor collects environmental data such as temperature, relative humidity, and PM2.5 concentration in the school classroom or infirmary, paying particular attention to environmental parameters related to potential allergens such as dust and pollen in group activity areas. All collected data is temporarily stored on the terminal in real time and automatically uploaded to the cloud after the terminal connects to the school's WiFi or the child's mobile hotspot.

[0040] After connecting to the network, the intelligent monitoring terminal transmits the stored operational data and environmental data to the AI ​​analysis platform. The platform's data preprocessing module first uses an improved outlier removal algorithm, combined with the Raida criterion and a dynamic threshold mechanism, to remove abnormal data caused by personnel movement and equipment interference in the campus environment. Then, it performs noise reduction and normalization to unify the data format and numerical range. The preprocessed data analysis process is as follows: The operational compliance judgment sub-model uses a multi-dimensional temporal alignment algorithm to dynamically match the child's medication operation timeline data on campus with a standard template to determine whether there are operational errors caused by distraction in the campus environment; the dosage deviation calculation sub-model uses a drug residue correction algorithm. The method calculates the actual inhaled dose deviation based on the theoretical dose, drug release efficiency, residual drug amount in the container, and residual drug recovery coefficient of the dry powder inhaler, and investigates whether insufficient or excessive medication is caused by operational problems. The compliance assessment sub-model uses a behavioral pattern recognition algorithm to calculate the compliance index by combining the time distribution characteristics of children's medication use at school and at home, and assesses medication compliance during school. The health risk prediction sub-model uses a spatiotemporal fusion algorithm to generate an asthma attack risk assessment report for the next 7 days by combining children's medication use data at school, school environment data, scores of whether they have had signs and symptoms such as cough and wheezing in the past 3 days, and seasonal factors, focusing on the risks that may be caused by group activities at school.

[0041] Considering the unique nature of the school setting, the multi-terminal reminder module employs a local low-level reminder + parent-teacher dual-terminal notification mode. Ten minutes before the daily 12 PM medication time, the local audio-visual unit of the intelligent monitoring terminal emits a low-volume humming alert and a slowly flashing blue light signal to avoid disturbing group activities on campus while ensuring no medication reminders are missed. If the AI ​​analysis platform determines that the child's medication operation is not standardized, the dosage is excessive, or the health risk assessment report shows that the risk exceeds the standard, the terminal's local audio-visual unit switches to a medium-volume beeping warning sound and a flashing orange light signal to remind the child to correct the situation promptly. Simultaneously, the parent's mobile application immediately receives notifications of the abnormal situation and the health risk report. The teacher responsible for the child's health management at the school also receives a simplified version of the abnormal notification, facilitating the teacher's assistance in correcting the child's medication operation on campus. In addition, the parent's mobile application pushes a daily medication summary report for the child at school every evening, including medication time, standardization of operation, dosage, etc., allowing parents to fully understand the child's medication status at school.

[0042] The medical collaboration terminal receives real-time data on children's medication use at school, assessment results, and health risk assessment reports from the AI ​​analysis platform. When doctors view the data through the doctor's web platform, if they find that the child has repeatedly made dosage errors due to unfamiliarity with medication use at school, and if the PM2.5 concentration in the school environment has been high recently, increasing the child's health risk, they can adjust the medication plan through the medication plan optimization module. This module uses three-dimensional feature matching and dynamic weighting technology to extract and standardize four types of data: the child's medication use characteristics at school, school environment characteristics, and physical signs. It then matches the medication plan data with the medication plans of more than 1,000 similar cases of children with asthma at school in the feature database. After calculating the matching degree, it calculates the weighted average of candidate plans based on symptom improvement rate, adverse reaction rate, and medication adherence in the school setting. Combined with the child's current health risk level, it generates a medication plan suitable for the school setting. The doctor pushes the adjusted medication plan and school medication precautions through the parent's mobile application, and also sends a simplified version of the medication guide to the school teacher, forming a three-way collaboration between doctor, parent, and teacher to ensure that the child can take medication safely and in accordance with regulations at school.

[0043] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An AI-based intelligent monitoring and reminder system for inhaled medications in children with asthma, characterized in that, The system comprises the following components: intelligent monitoring terminal, AI analysis platform, multi-terminal alert module, and medical collaboration terminal: The intelligent monitoring terminal is detachably mounted on the asthma inhalation drug container and integrates a pressure sensor, an angle sensor, an airflow sensor, a facial image acquisition component, and an environmental sensor. The airflow sensor uses dual-channel airflow acquisition technology to simultaneously acquire the axial and radial airflow velocities at the inhalation port. The angle sensor has a built-in accelerometer sensor and supports dynamic benchmark calibration. It is used to collect operational and environmental data during children's medication administration. The AI ​​analysis platform is deployed on a cloud server and establishes a real-time data transmission link with the intelligent monitoring terminal. It includes a data preprocessing module, an operation standardization judgment sub-model, a dose deviation calculation sub-model, a compliance assessment sub-model, and a health risk prediction sub-model. The data preprocessing module adopts an improved outlier removal algorithm. The operational compliance judgment sub-model employs a multi-dimensional temporal alignment algorithm; the dose deviation calculation sub-model employs a drug residue correction algorithm; the compliance assessment sub-model employs a behavioral pattern recognition algorithm; and the health risk prediction sub-model employs a spatiotemporal fusion algorithm. It receives operational data and environmental data, sequentially performs data cleaning, denoising, and normalization preprocessing, inputs the data into a multi-task deep learning model, and outputs medication operation compliance results, actual inhaled dose deviation results, and medication compliance levels. It also generates a health risk assessment report by combining historical medication data and environmental data. The multi-terminal reminder module includes a local audio-visual unit integrated into the smart monitoring terminal and a parent-side mobile application. It is used to issue reminder signals at preset medication times, issue warning signals when the operation is not standardized, the dosage is out of control, or the health risk exceeds the standard, and push medication plans, abnormal notifications, and health risk reports. The medical collaboration terminal includes a doctor-side web platform and a mobile terminal, with a built-in medication plan optimization module. The medication plan optimization module uses three-dimensional feature matching + dynamic weighting technology to obtain medication data, evaluation results and health risk reports output by the AI ​​analysis platform, so that doctors can adjust the medication plan and push remote guidance information to the parent-side mobile application.

2. The AI-based intelligent monitoring and reminder system for inhaled medications in children with asthma according to claim 1, characterized in that, The airflow sensing component of the intelligent monitoring terminal adopts dual-channel airflow acquisition technology to simultaneously acquire the axial and radial airflow velocities at the inlet, and calculates airflow stability using an airflow stability index algorithm. Where S 气流 v is the airflow stability index. i Let v be the airflow velocity at the i-th sampling point. avg The average airflow velocity during the sampling period is n, where n is the number of sampling points and k is k. corr This is the correction factor.

3. The AI-based intelligent monitoring and reminder system for inhaled medications in children with asthma according to claim 1, characterized in that, The data preprocessing module of the AI ​​analysis platform employs an improved outlier removal algorithm, combining the Laida criterion and a dynamic threshold mechanism: P 异常 =|x i -μ|>3σ+α×μ, where P 异常 For the outlier determination result, x i α represents the original data values, μ represents the data mean, σ represents the standard deviation, and α represents the dynamic adjustment factor.

4. The AI-based intelligent monitoring and reminder system for inhaled medications in children with asthma according to claim 1, characterized in that, The operational compliance judgment sub-model of the AI ​​analysis platform adopts a multi-dimensional temporal alignment algorithm to dynamically match actual operational data with standard templates: Where M 匹配 For the matching degree, d j Let w be the Euclidean distance between the data in the j-th dimension and the standard template. j For dimensional weights, τ j This is the dimensional decay coefficient.

5. The AI-based intelligent monitoring and reminder system for inhaled medications in children with asthma according to claim 1, characterized in that, The dose deviation calculation sub-model of the AI ​​analysis platform uses a drug residue correction algorithm. The actual inhaled dose calculation formula is: D 实际 =(D 理论 ×η 释放 )-(R 残留 ×k 回收 ), where D 实际 D is the actual inhaled dose. 理论 η is the theoretical dose produced by pressing. 释放 For drug release efficiency, R 残留 k represents the amount of residual drug in the container. 回收 The residual drug recovery coefficient is given.

6. The AI-based intelligent monitoring and reminder system for inhaled medications in children with asthma according to claim 1, characterized in that, The compliance assessment sub-model of the AI ​​analysis platform uses a behavioral pattern recognition algorithm, combined with medication time distribution characteristics, to calculate the compliance index: Where C 依从 T is a compliance index. 按时 To ensure timely medication administration, T 总 σ represents the total number of times the drug is applied. t S represents the standard deviation of medication time, ΔT represents the medication interval, and S represents the standard deviation of medication time. 连续 γ is the ratio of the number of consecutive days of medication to the total number of days, and γ is the continuous medication incentive coefficient.

7. The AI-based intelligent monitoring and reminder system for inhaled medications in children with asthma according to claim 1, characterized in that, The health risk prediction sub-model of the AI ​​analysis platform uses a spatiotemporal fusion algorithm to calculate the probability of an asthma attack in the next 7 days: Where P 风险 X represents the probability of an asthma attack in the next 7 days, where β0 is a constant, and β1, β2, β3, and β4 are feature weights. 用药 To standardize medication adherence scoring, X 环境 X is the environmental risk index. 症状 X is the score for signs and symptoms over the past 3 days. 季节 It is a seasonal factor.

8. The AI-based intelligent monitoring and reminder system for inhaled medications in children with asthma according to claim 1, characterized in that, The angle sensing component of the intelligent monitoring terminal uses dual position reference + real-time compensation technology to realize real-time correction of the container tilt angle measurement value. The specific steps are as follows: two miniature tilt sensors are symmetrically installed on the outer side of the middle of the container. The main sensor collects real-time angle data at a sampling frequency of 30Hz, and the secondary sensor simultaneously collects pressure distribution data of the container sidewall. The system presets a pressure distribution threshold corresponding to the standard medication angle range. If the angle detected by the main sensor deviates from the standard range during medication, the secondary sensor immediately compares the deviation between the current pressure distribution and the threshold and calculates the compensation coefficient. The angle value of the main sensor is multiplied by the compensation coefficient to obtain the correction amount, which is then added to the original angle value of the main sensor to generate the final corrected angle.

9. The AI-based intelligent monitoring and reminder system for inhaled medications in children with asthma according to claim 1, characterized in that, The medication plan optimization module of the medical collaboration terminal uses three-dimensional feature matching + dynamic weighting technology to recommend medication frequency adjustment. The specific steps are as follows: the system constructs a medication feature database for childhood asthma, extracts four types of feature data of the current child and standardizes them when administering medication; the standardized data is matched with the medication frequency and efficacy data of more than 1,000 similar cases in the feature database, and the matching degree is calculated; the medication frequency of the candidate set of cases is calculated as a weighted average based on symptom improvement rate, adverse reaction incidence rate and medication compliance, which is used as the initial recommended frequency. The final recommended frequency is generated based on the current level of children's health risks.

10. An AI-based intelligent monitoring and reminder method for inhaled medications in children with asthma, applicable to the AI-based intelligent monitoring and reminder system for inhaled medications in children with asthma as described in any one of claims 1-9, characterized in that, The specific steps of this method are as follows: S1. Data Acquisition Steps: Through a smart monitoring terminal that is detachably mounted on the asthma inhalation drug container, the operation data and environmental data during the child's medication process are collected simultaneously. The operation data includes the pressure applied, container tilt angle, inhalation airflow parameters, and facial posture data. The environmental data includes the temperature, relative humidity, and PM2.5 concentration of the medication environment. S2. Data processing and analysis steps: The collected operational data and environmental data are transmitted to the AI ​​analysis platform via wireless communication. Data cleaning, noise reduction, and normalization preprocessing are performed in sequence. Then, the data is input into the preset multi-task deep learning model, and the results of medication operation standardization, actual inhaled dose deviation, and medication compliance level are output. A health risk assessment report is generated by combining historical medication data and environmental data. S3. Reminder and Notification Steps: Tiered reminders are implemented through a multi-terminal reminder module. The local audio-visual unit of the multi-terminal reminder module issues a prompt signal at the preset medication time and an alarm signal when the operation is not standardized, the dosage is out of control, or the health risk exceeds the standard. The parent's mobile application receives medication plan push, abnormal situation notification and health risk report. S4. Medical Collaboration Steps: The medical collaboration terminal obtains medication data, assessment results, and health risk reports output by the AI ​​analysis platform, which are then used by doctors to adjust medication plans and push remote guidance information through the parents' mobile application.

Images