Child allergic rhinitis desensitization treatment tracking system

Abstract

The application discloses a kind of based on children allergic rhinitis desensitization treatment tracking system, it is related to children allergic rhinitis desensitization treatment technical field, including: data acquisition module, for synchronously obtaining physiological signal data and discrete symptom evaluation data of target child in initial period of desensitization treatment;Signal processing module is used for the frequency domain analysis to physiological signal data, and the day rate heart rate variability index reflecting sympathetic and parasympathetic nervous balance degree of autonomous nervous system is calculated.This application establishes the dynamic calibration mechanism of objective physiological index to subjective symptom data, effectively eliminates the subjective perception deviation when physiological state is abnormal, so that symptom score can accurately reflect the real rhinitis inflammation load of child, effectively solve the traditional subjective and objective data static split, subjective data low reliability Problem, corrects data distortion problem, can reliably reflect real pathophysiological change, improve the accuracy of subsequent judgment result.

Description

Technical Field

[0001] This invention relates to the field of desensitization treatment technology for allergic rhinitis in children, specifically to a tracking system for desensitization treatment of allergic rhinitis in children. Background Technology

[0002] Allergic rhinitis is a common chronic inflammatory disease in children, and specific immunotherapy, also known as desensitization therapy, is the core approach to treating its underlying cause. This therapy induces immune tolerance in the body by administering allergen extracts regularly and over a long period, thereby achieving long-term symptom relief and even altering the natural course of the disease. However, this therapy typically requires 3 to 5 years of continuous treatment, and significant individual differences in efficacy exist. Currently, treatment follow-up and early efficacy assessment are mainly conducted through clinical practice.

[0003] Currently, the tracking and evaluation of treatment processes heavily rely on two types of information: one is the symptom diary recorded daily by the patient or guardian, which is essentially a discrete, subjective, and easily interfered-with report, especially for pediatric patients, where its accuracy and consistency are difficult to guarantee; the other is objective examinations conducted at discrete time points, such as nasal endoscopy and specific IgE testing, several months or even years after treatment.

[0004] With the widespread use of wearable devices, the tracking of desensitization therapy usually treats the patient's subjective symptom reports and objectively collected physiological signals as independent or simply parallel data streams. This processing mode has the defects of static separation and equal reliability. When the patient's autonomic nervous function is temporarily disordered due to fatigue or tension (HRV abnormality), their subjective perception and report of symptoms on that day may deviate significantly from their true inflammatory load. However, the existing system still treats it the same as the report when the state is stable, which leads to the tracking report on that day failing to reliably reflect the true pathophysiological changes and affecting the accuracy of subsequent judgment results. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a tracking system for desensitization treatment of allergic rhinitis in children.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows:

[0007] A tracking system for desensitization treatment of allergic rhinitis in children includes:

[0008] The data acquisition module is used to simultaneously acquire physiological signal data and discrete symptom assessment data of the target child during the initial period of desensitization treatment;

[0009] The signal processing module is used to perform frequency domain analysis on physiological signal data and calculate the daily heart rate variability index, which reflects the balance between the sympathetic and parasympathetic nervous systems of the autonomic nervous system.

[0010] The data fusion module is used to convert the daily heart rate variability index into a corresponding dynamic calibration weight coefficient for each day through a preset function. Based on the dynamic calibration weight coefficient, the discrete symptom assessment data of the same day are weighted and calculated to generate a dynamically weighted daily symptom score sequence.

[0011] The indicator construction module is used to perform statistical analysis on the dynamically weighted daily symptom score sequence within the initial period, calculate and generate the first health status indicator representing the degree of symptom fluctuation, arrange the daily heart rate variability indicators within the initial period in chronological order and perform linear regression analysis, and use the slope value of the regression line as the second health status indicator representing the trend of physiological state changes.

[0012] The prediction module is used to input the first and second health status indicators into the pre-trained efficacy potential prediction model and output the early efficacy potential prediction results for the target child.

[0013] A tracking method for desensitization treatment in children with allergic rhinitis includes the following steps:

[0014] Simultaneously acquire physiological signal data and discrete symptom assessment data of the target children during the initial cycle of desensitization therapy;

[0015] Frequency domain analysis of physiological signal data was performed to calculate the daily heart rate variability index, which reflects the balance between the sympathetic and parasympathetic nervous systems of the autonomic nervous system.

[0016] The daily heart rate variability index is converted into a corresponding dynamic calibration weight coefficient for each day through a preset function. Based on the dynamic calibration weight coefficient, the discrete symptom assessment data of the same day are weighted and calculated to generate a dynamically weighted daily symptom score sequence.

[0017] Statistical analysis was performed on the dynamically weighted daily symptom score sequence within the initial period to calculate and generate a first health status index representing the degree of symptom fluctuation. The daily heart rate variability index within the initial period was arranged in chronological order and linear regression analysis was performed. The slope of the regression line was used as a second health status index representing the trend of physiological state changes.

[0018] The first and second health status indicators are input into the pre-trained efficacy potential prediction model, which outputs the early efficacy potential prediction results for the target children.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0020] This invention establishes a dynamic calibration mechanism for subjective symptom data based on objective physiological indicators, effectively eliminating subjective perception bias when physiological states are abnormal. This allows symptom scores to accurately reflect the true inflammatory burden of rhinitis in children, effectively solving the problems of static separation between objective and subjective data and low reliability of subjective data in traditional methods. By generating a dynamic weighted daily symptom score sequence, it achieves the continuous and standardized transformation of discrete subjective data, corrects data distortion, and reliably reflects real pathophysiological changes, thereby improving the accuracy of subsequent judgments. Attached Figure Description

[0021] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts. Wherein:

[0022] Figure 1 This is a system diagram of the present invention;

[0023] Figure 2 This is a flowchart of the present invention. Detailed Implementation

[0024] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0025] Traditional methods have significant limitations in the clinical follow-up and early efficacy assessment of desensitization treatment for allergic rhinitis in children. Specifically, treatment follow-up relies heavily on subjective symptom diaries recorded by patients or guardians. This type of data is a discrete subjective report, which is affected by the child's expressive ability, the guardian's subjective perception, and external factors, making it difficult to guarantee accuracy and consistency. At the same time, objective efficacy assessment is only completed at discrete time points, such as nasal endoscopy and specific IgE testing, several months or even years after treatment, which cannot achieve continuous dynamic tracking of the treatment process.

[0026] With the widespread use of wearable devices, existing desensitization therapy tracking methods, while integrating subjective symptom reports and objective physiological signals, treat these two types of data as independent or simply parallel data streams. This results in static separation and equivalence of reliability, failing to consider the impact of physiological state on subjective symptom reports. When children experience temporary autonomic nervous system dysfunction due to fatigue, stress, or other factors (manifested as abnormal heart rate variability (HRV)), their subjective perception and reporting of rhinitis symptoms on that day will significantly deviate from the true inflammatory load. Existing systems still treat such distorted reports the same as reports when the physiological state is stable, leading to daily tracking data that cannot reliably reflect true pathophysiological changes. Consequently, the accuracy of subsequent efficacy judgments based on this type of data is greatly reduced.

[0027] For example, during the initial period of desensitization treatment for allergic rhinitis in children, poor sleep at night can lead to autonomic nervous system disorders and abnormal fluctuations in HRV (Heat Ratio) indicators. Guardians, due to the child's poor mental state, may exaggerate symptoms such as nasal congestion and sneezing. Existing tracking systems do not specifically calibrate these subjective symptom scores, directly merging them with symptom data from a stable physiological state. This results in treatment tracking reports that overemphasize subjective perception rather than reflecting actual inflammation improvement. When doctors subsequently assess treatment effectiveness based on these distorted reports, they are prone to misjudging the treatment outcome and may even make unnecessary adjustments to the treatment plan, affecting the regularity and effectiveness of desensitization treatment.

[0028] If the aforementioned problems are not addressed, the tracking and efficacy assessment of desensitization therapy for allergic rhinitis in children will face risks of data distortion, delayed assessment, and low accuracy in clinical applications. Specifically, the inherent defects of subjective symptom data and the unreasonable processing of subjective and objective data will lead to the inability to accurately capture dynamic pathophysiological changes during treatment. Early efficacy assessment will lack reliable data support, and the discrete time-point objective examination method will prevent doctors from promptly identifying children with poor efficacy in the early stages of treatment. This will result in some children receiving ineffective treatment for months or even years, increasing not only the physical burden on children and the economic costs on families, but also potentially delaying intervention and affecting the long-term control of the disease. Furthermore, distorted tracking data will also lead to a lack of unified and accurate quantitative standards for clinical evaluation of the efficacy of desensitization therapy.

[0029] In this regard, such as Figure 1-2 As shown, a tracking system for desensitization treatment of allergic rhinitis in children is proposed, including:

[0030] The data acquisition module is used to simultaneously acquire physiological signal data and discrete symptom assessment data of the target child during the initial period of desensitization treatment;

[0031] Specifically, in the follow-up of traditional desensitization treatment for allergic rhinitis in children, there are problems such as asynchronous acquisition of subjective and objective data, fragmented sources, and misaligned time dimensions: First, there is no unified time benchmark for the collection of subjective symptom assessment data (from records of guardians or doctors) and objective physiological signal data, which easily leads to cross-day and cross-time period matching; Second, there is a lack of a continuous mechanism for synchronous collection of subjective and objective data in the initial treatment cycle, relying only on objective examinations at discrete time points in the later stage, resulting in a lack of accurate two-dimensional data foundation in the early stage of treatment; Third, there is no standardized method for data storage and association, which lays the potential for error in subsequent data processing and fusion, and cannot guarantee the reliability of the analysis results.

[0032] Furthermore, the time frame for data collection is clearly defined as the initial period of desensitization treatment, i.e., the core clinical window period, which can reflect the characteristics of early treatment response. A standardized collection method is adopted to simultaneously acquire two types of core data: continuous physiological signal data collected by wearable devices and discrete symptom assessment data filled out daily by guardians or doctors. Through timestamp alignment technology, the two types of data are uniformly time-marked and stored in association to ensure that the physiological signal data and discrete symptom assessment data of the same natural day are matched one by one, forming a raw dataset with a unified time dimension.

[0033] The aforementioned technology effectively solves the problems of fragmented data sources and time misalignment in traditional tracking by establishing a continuous, synchronous, and precisely time-aligned foundation of subjective and objective raw data within the initial cycle of desensitization treatment. It achieves standardized and associated storage of the two types of data, eliminates analytical errors caused by time mismatch in subsequent data processing, and provides accurate and reliable raw data support for subsequent physiological indicator extraction and subjective and objective data fusion, laying the foundation for data analysis.

[0034] The signal processing module is used to perform frequency domain analysis on physiological signal data and calculate the daily heart rate variability index, which reflects the balance between the sympathetic and parasympathetic nervous systems of the autonomic nervous system.

[0035] Specifically, traditional treatment tracking relies solely on subjective symptom records, lacking objective core indicators that quantify the true pathophysiological state of children: First, allergic rhinitis directly interferes with the balance between the sympathetic and parasympathetic nervous systems of children's autonomic nervous system, and there are no continuous quantitative representation methods for this balance in clinical practice; second, the raw physiological signal data contains a large amount of noise, such as children's activities, equipment errors, and environmental interference, and without targeted preprocessing, effective physiological features cannot be extracted; third, without frequency domain analysis of physiological signals, it is impossible to extract core information reflecting autonomic nervous activity from time domain signals, resulting in objective physiological data failing to truly reflect changes in the pathological state.

[0036] Furthermore, the acquired physiological signal data undergoes targeted preprocessing. Signal segments within the daily nighttime sleep time window are extracted, as the data within this window is free from activity and emotional interference and exhibits the highest stability. Noise reduction and normalization operations are performed to eliminate noise interference and individual or device differences in signal amplitude. A Fast Fourier Transform is then performed on the preprocessed clean signal to convert the time-domain physiological signal into a frequency-domain signal, distinguishing the low-frequency segment corresponding to sympathetic nerve activity and the high-frequency segment corresponding to parasympathetic nerve activity. The power spectral density of the low-frequency segment and the high-frequency segment are calculated, and the ratio between the two is used as the core calculation basis to obtain the daily heart rate variability (HRV) index, which accurately reflects the balance between the sympathetic and parasympathetic nervous systems of the autonomic nervous system, thereby achieving a quantitative representation of daily physiological state.

[0037] The aforementioned technology fills the gap in traditional tracking by extracting objective indicators that can accurately quantify the true pathophysiological state of children from raw physiological signals. It also eliminates noise interference to the greatest extent possible through nighttime signal interception and preprocessing, ensuring the accuracy and stability of the HRV indicator. This indicator transforms the pathophysiological changes caused by allergic rhinitis in children into calculable and comparable quantitative values, providing a reliable objective basis for the calibration of subsequent subjective symptom data and realizing the transformation from raw physiological signals to effective physiological characteristics.

[0038] The data fusion module is used to convert the daily heart rate variability index into a corresponding dynamic calibration weight coefficient for each day through a preset function. Based on the dynamic calibration weight coefficient, the discrete symptom assessment data of the same day are weighted and calculated to generate a dynamically weighted daily symptom score sequence.

[0039] Specifically, traditional tracking suffers from static separation and equal credibility in the processing of subjective and objective data. It simply analyzes subjective symptom assessment data alongside objective physiological data without considering abnormal physiological states, such as abnormal HRV indicators or autonomic nervous system disorders. In such cases, the subjective symptom perception and reporting of the child or guardian will deviate significantly from the true inflammatory load, leading to distortion of subjective data. Furthermore, the processing of subjective symptom data is static and lacks a calibration mechanism that dynamically adjusts with physiological states, making it impossible to correct subjective biases.

[0040] Furthermore, by introducing a preset function, the daily HRV index is transformed into a dynamic calibration weight coefficient within the range of 0-1. The weight coefficient is dynamically adjusted according to the changes in the HRV index, accurately matching the child's physiological state on that day. The more abnormal the HRV, the greater the weight calibration amplitude. Using the daily dynamic calibration weight coefficient as the correction basis, the discrete symptom assessment data of the same day are weighted and calculated to eliminate the subjective perception bias caused by abnormal physiological state. The symptom scores obtained from the daily weighted calculation are integrated in chronological order to generate a dynamically weighted daily symptom score sequence, realizing the standardization and continuous correction of subjective symptom data.

[0041] The aforementioned technology effectively eliminates subjective perception bias when physiological states are abnormal by establishing a dynamic calibration mechanism for objective physiological indicators to subjective symptom data. This allows symptom scores to accurately reflect the true inflammatory burden of rhinitis in children, effectively solving the problems of static separation between objective and subjective data and low reliability of subjective data in traditional methods. Through the generated dynamic weighted daily symptom score sequence, the discrete subjective data is transformed into continuous and standardized data, correcting the data distortion problem and providing high-quality and highly reliable corrected data for the subsequent construction of core health status indicators.

[0042] The indicator construction module is used to perform statistical analysis on the dynamically weighted daily symptom score sequence within the initial period, calculate and generate the first health status indicator representing the degree of symptom fluctuation, arrange the daily heart rate variability indicators within the initial period in chronological order and perform linear regression analysis, and use the slope value of the regression line as the second health status indicator representing the trend of physiological state changes.

[0043] Specifically, due to the lack of a unified quantitative indicator system in traditional treatment tracking, it is impossible to extract the core state characteristics of the initial stage of treatment from the corrected subjective and objective data. It only performs simple summation and average statistics on the original symptom data, and cannot extract the core characteristics reflecting the stability of symptoms from the dynamically weighted symptom score sequence. Furthermore, it does not conduct time-dimensional trend analysis on the objective HRV indicator, and cannot quantify the long-term changing trend of the child's physiological state.

[0044] Furthermore, by performing statistical characteristic analysis on the dynamic weighted daily symptom score sequence within the initial period, the ratio of its standard deviation to the mean, i.e., the coefficient of variation, is calculated. This ratio is used as the first health status indicator to quantitatively characterize the degree of symptom fluctuation in the early stage of treatment. Then, all daily HRV indicators within the initial period are arranged in chronological order, and linear regression analysis is performed on them. The slope of the regression line is extracted as the second health status indicator to quantitatively characterize the trend of physiological state changes in the early stage of treatment. Thus, a dual-dimensional quantitative health status indicator system that combines symptom characteristics and physiological characteristics is constructed.

[0045] Furthermore, the daily heart rate variability index within the initial period was arranged in chronological order and subjected to linear regression analysis. The slope of the regression line was used as a second health status indicator to characterize the trend of physiological state changes, as detailed below:

[0046] First, the daily HRV indexes calculated during the initial cycle are organized into an ordered time series data according to the treatment time and HRV value. The higher the HRV value, the more dominant the sympathetic nervous system is, the more unbalanced the autonomic nervous system is, and the more severe the physiological disorders related to rhinitis are.

[0047] Example:

[0048] Treatment time (t) 1 2 3 4 5 6 7 8 9 10 Daily HRV value 2.9 2.8 2.7 2.6 2.5 2.4 2.3 2.2 2.1 2.0

[0049] Construct a linear regression model, the expression of which is: ;

[0050] Independent variable Treatment time, such as Day 1 = 1, Day 2 = 2, and so on;

[0051] Dependent variable : Daily HRV value for the corresponding time period;

[0052] The slope of the regression line, i.e., the second health status indicator;

[0053] The intercept of the regression line is only a fitting parameter and has no clinical significance.

[0054] The least squares method is used to calculate the sum of squared errors of all data points to the line. :

[0055] ;

[0056] : Total number of days in the initial cycle;

[0057] : Add up all the days;

[0058] : Sum of all HRV values;

[0059] : Add up the number of days × HRV each day;

[0060] Add up the squares of the number of days each day;

[0061] Example calculation:

[0062]

[0063] ;

[0064] ;

[0065] ;

[0066] ;

[0067] ;

[0068] ;

[0069] ;

[0070] The calculation yields:

[0071] This is the second health status indicator;

[0072] when When the value is less than 0, it indicates that the HRV value continues to decrease with the treatment time, the proportion of sympathetic nerves decreases, the balance of autonomic nerves improves, the desensitization treatment has taken effect at the physiological level, and the autonomic nerve disorder related to rhinitis is being relieved.

[0073] when When the value is equal to 0, it indicates that the HRV value has no trend change, the autonomic nervous system balance has not improved or has worsened, the treatment in the initial cycle has not shown physiological effects, and continuous observation is required.

[0074] when When the value is greater than 0, it indicates that the HRV value continues to rise with the treatment time, the proportion of sympathetic nerves increases, the autonomic nervous system disorder worsens, the treatment may be ineffective, or the child's condition may even deteriorate, and the treatment plan needs to be adjusted in time.

[0075] In the aforementioned technology, by constructing a two-dimensional indicator system that can accurately quantify the core status of children in the initial treatment cycle, the problem of traditional tracking lacking standardized core quantitative characteristics is effectively solved. The first health status indicator realizes the quantification of the stability of the child's symptoms. The larger the value, the more unstable the symptoms and the higher the uncertainty of the efficacy. The second health status indicator realizes the quantitative capture of the long-term trend of the child's physiological status. A positive slope indicates the improvement of the physiological status, a negative slope indicates the deterioration of the physiological status, and a slope close to 0 indicates no significant change in the physiological status.

[0076] The prediction module is used to input the first and second health status indicators into the pre-trained efficacy potential prediction model and output the early efficacy potential prediction results for the target child.

[0077] Specifically, due to the lag, subjectivity, and inefficiency of traditional desensitization therapy efficacy assessment, efficacy can only be assessed months or even years later through objective examinations at discrete time points such as nasal endoscopy and specific IgE testing. It is impossible to make early predictions at the beginning of the treatment cycle, resulting in children with poor efficacy receiving long-term ineffective treatment and delaying the intervention opportunity. Furthermore, early efficacy assessment relies on the experience analysis of clinicians, which is highly subjective. Different doctors have different judgment results, lacking unified judgment standards and standardized efficacy prediction models. It is impossible to transform core quantitative indicators into accurate efficacy potential results, resulting in a lack of scientific and reliable decision-making basis for clinical intervention.

[0078] Furthermore, since the calculated raw indicators have dimensional differences and are inconsistent with the input requirements preset by the pre-trained model, it is necessary to first process the first health status indicators according to the preprocessing rules built into the efficacy potential prediction model. Second health status indicators Standardized preprocessing is performed to ensure that the indicators can be correctly identified and calculated by the efficacy potential prediction model. The specific steps are as follows:

[0079] Accessing the normalized parameters fixed in the efficacy potential prediction model: This involves reading the normalized parameters stored in the efficacy potential prediction model, i.e., those determined during the pre-training phase. Minimum value ( ), maximum value ( ),as well as Minimum value of the indicator ( ), maximum value ( This parameter represents the known preset conditions of the efficacy potential prediction model, which can be directly called.

[0080] Indicator normalization processing: This involves normalizing the indicators of the target child. , The original indicators are substituted into the normalization formula built into the efficacy potential prediction model and mapped to the [0,1] interval to obtain the normalized indicators. The normalization formula is as follows:

[0081] ;

[0082] ;

[0083] Will and Inference calculations are performed on the input therapeutic potential prediction model:

[0084] First, the fixed parameters of the efficacy potential prediction model are invoked: The efficacy potential prediction model automatically reads the core parameters fixed during the pre-training phase, including the intercept term. , Feature weights and Feature weights ;

[0085] Will , , , and Substitute into the formula to calculate:

[0086] ;

[0087] The result of the calculation The probability of effective desensitization therapy for the target child.

[0088] Example calculation:

[0089] Assumption , , =0.12、 and ;

[0090] ;

[0091] ;

[0092] The success rate of desensitization treatment for the target children is approximately 60.3%.

[0093] After completing the inference calculation, the efficacy potential prediction model automatically converts the probability values ​​into clinically usable binary prediction conclusions based on built-in preset thresholds. It also outputs auxiliary information to facilitate doctors' decision-making in conjunction with clinical practice, as detailed below:

[0094] The predictive threshold built into the efficacy potential prediction model is preset to 0.5. This threshold is the optimal threshold determined during the pre-training phase and is fixed. It can be flexibly adjusted between 0.4 and 0.6 according to actual clinical needs. After adjustment, there is no need to retrain the model; only the threshold parameter needs to be modified. The purpose of setting the threshold is to balance the sensitivity and specificity of the prediction, avoiding missed effective cases or misclassified ineffective cases. The determination is as follows:

[0095] If the effective probability is ≥0.5, it indicates a high potential for early treatment of the target children.

[0096] If the effective probability is <0.5, it indicates that the early therapeutic potential for the target child is low;

[0097] The output consists of three parts to ensure that doctors have a comprehensive understanding of the prediction and avoid misdiagnosis due to a single conclusion:

[0098] Key predictive conclusions: Clearly indicate whether the early efficacy potential is high or low, and use this as a core reference for clinical decision-making.

[0099] Supplementary reference information: The effective probability calculated by the efficacy potential prediction model is marked, which helps doctors to make further judgments based on their own clinical experience;

[0100] Clinical guidance tips:

[0101] If the therapeutic potential is high: it is recommended to continue the current desensitization treatment plan, maintain the regular follow-up frequency, and continuously monitor changes in indicators;

[0102] If the therapeutic potential is low: it is recommended to combine the child's clinical symptoms and objective examinations to promptly assess and adjust the treatment plan to avoid long-term ineffective treatment.

[0103] In the aforementioned technology, the first and second health status indicators are used as core input features and fed into a pre-trained efficacy potential prediction model. Through standardized computational processing of the model, the early efficacy potential prediction results of desensitization treatment for the target child are directly output. This achieves early, accurate, and quantitative prediction of the child's efficacy potential within the initial cycle of desensitization treatment, effectively solving the core problems of lagging and subjective traditional efficacy judgment. It advances the time window for efficacy judgment from months or years to the early stage of treatment, replacing doctors' experience-based judgment, establishing a standardized efficacy prediction system, ensuring the consistency and scientific nature of the judgment results, and enabling clinicians to promptly identify children with low efficacy potential. This provides accurate decision-making basis for early adjustment of treatment plans, cessation of ineffective treatments, and replacement of intervention methods, effectively reducing the physical burden on children and the economic costs on families, and significantly improving the clinical intervention efficiency and accuracy of desensitization treatment for childhood allergic rhinitis.

[0104] The data acquisition module includes:

[0105] Wearable physiological signal acquisition unit is used to continuously acquire raw photoplethysmography (PPG) signals from the target child.

[0106] The user interaction unit is used to receive daily input of discrete symptom assessment data.

[0107] The time synchronization unit is used to align and store the raw photoplethysmography pulse wave signal and discrete symptom assessment data based on a unified timestamp to obtain physiological signal data.

[0108] Specifically, the raw photoplethysmography (PPG) signal is obtained by emitting light of a specific wavelength through a sensor, which penetrates skin tissue to detect periodic changes in blood volume, forming a continuous pulse wave signal. Continuous acquisition ensures the integrity and continuity of the signal. The raw PPG signal is used to calculate daily heart rate variability, reflecting the balance between the sympathetic and parasympathetic nervous systems in the child's autonomic nervous system. This directly correlates with the pathophysiological changes in rhinitis, replacing traditional methods that rely solely on subjective descriptions of physiological states. This provides objective, continuous, and quantitative physiological signal acquisition, offering authentic, unbiased raw data for subsequent physiological indicator extraction.

[0109] Discrete symptom assessment data refers to the graded and quantitative scoring of core symptoms of allergic rhinitis in children, such as sneezing, nasal congestion, runny nose, nasal itching, and eye itching. It is typically scored on a 1-5 scale, with 0 indicating no symptoms and higher scores indicating more severe symptoms. This is standardized, discrete, subjective assessment data, entered at a fixed frequency daily, usually once a day, corresponding to the overall symptom situation on the day of treatment, forming a daily discrete symptom score. This score is filled out by the guardian or doctor based on the child's actual symptom presentation that day. It is the most commonly used form of symptom tracking in clinical practice. After standardization, it can be used for statistical and quantitative analysis, transforming vague, verbal symptom descriptions into standardized, calculable, and comparable discrete score data, providing the basic data for subjective dimensions in subsequent symptom score weighting and health status indicator construction.

[0110] The raw photoplethysmography (PPG) signal and discrete symptom assessment data are aligned and stored based on a unified timestamp, ultimately yielding physiological signal data that the system can directly use.

[0111] Uniform timestamp: Using the system standard time as the sole reference, each segment of raw photoplethysmography pulse wave signal and each set of daily discrete symptom assessment data are marked with a completely consistent standard time stamp.

[0112] Data alignment: Based on the same natural day as the matching rule, all raw photoplethysmography signals continuously collected on the same day are bound one by one with the discrete symptom assessment data entered on the same day to prevent time misalignment, cross-day matching, and data fragmentation.

[0113] Unified storage: The raw photoplethysmography pulse wave signal and discrete symptom assessment data, which have been time-aligned, are stored in a structured manner in chronological order to form standardized physiological signal data with time tags and subjective-objective binding, which can be directly called by subsequent signal processing modules.

[0114] It effectively solves the shortcomings of traditional tracking, such as the asynchronous and non-corresponding time of physiological data and symptom data, ensuring the accuracy of subsequent dynamic calibration weights and weighted calculation of symptom scores on the same day, avoiding analysis errors caused by time misalignment, realizing structured and traceable data storage, and improving the data reliability and analysis accuracy of the entire system.

[0115] The signal processing module is specifically configured as follows:

[0116] Based on a unified timestamp, signal segments within a preset nighttime sleep time window are extracted from physiological signal data each day, and the signal segments are preprocessed by denoising and normalization.

[0117] Perform a fast Fourier transform on the preprocessed signal segment and calculate the power spectral density in the low-frequency and high-frequency bands.

[0118] The daily heart rate variability index is calculated based on the ratio of power spectral density in the low-frequency band to that in the high-frequency band.

[0119] Specifically, signal interception: Based on a unified timestamp, the original photoplethysmography (PPG) signal segment within the nighttime sleep window from 22:00 to 06:00 the next day is intercepted, and only the signal during this period is retained, eliminating daytime interference.

[0120] Signal denoising: A combination of wavelet denoising and moving average filtering is used for processing.

[0121] Wavelet denoising: Perform a 3-level decomposition of the signal using db4 wavelets, retain low-frequency coefficients, and reconstruct the denoised signal;

[0122] Moving average filtering: Using N=8 sampling points as a window, calculate the local mean.

[0123] ;

[0124] : This is the signal after wavelet denoising;

[0125] : This is the filtered signal;

[0126] Normalization preprocessing, formula:

[0127] ;

[0128] : for the normalized first The amplitude of each sampling point is mapped to the interval [0,1].

[0129] Fast Fourier Transform (FFT): Perform an FFT on the normalized signal segment (length M), formula:

[0130] ;

[0131] : is a complex sequence in the frequency domain. The imaginary unit;

[0132] Frequency band allocation (child-friendly):

[0133] Low frequency band (LF): 0.04-0.15Hz (sympathetic nerve characteristics);

[0134] High frequency band (HF): 0.15-0.4Hz (parasympathetic nerve characteristics);

[0135] Power spectral density (PSD) calculation:

[0136] Using the periodogram method, the formula is:

[0137] ;

[0138] : This refers to the sampling rate of the original photoplethysmography (PPG) signal, such as 100Hz;

[0139] The amplitude is in the frequency domain.

[0140] For all frequencies in the LF and HF bands respectively Summing the values, we get the total PSD for the frequency band:

[0141] ;

[0142] Calculating the daily HRV index: The daily HRV index is obtained based on the LF / HF ratio, using the following formula:

[0143] ;

[0144] Elevated levels indicate that the sympathetic nervous system is dominant over the parasympathetic nervous system, resulting in an imbalance of the autonomic nervous system. This suggests that the child's rhinitis inflammation has led to sympathetic nerve excitation and a poor physiological state.

[0145] Decreased: This indicates that the parasympathetic nervous system is relatively dominant and the autonomic nervous system is becoming more balanced, suggesting the effect of desensitization therapy on improving physiological state;

[0146] The HRV index is calculated daily to form a daily sequence, which can intuitively reflect the changing trend of the child's autonomic nervous system state with desensitization treatment.

[0147] The default function uses the Sigmoid function form, and its specific expression is as follows:

[0148] ;

[0149] in, The output is the dynamic calibration weight coefficient, and ;

[0150] The input is the daily heart rate variability index;

[0151] These are the preset sensitivity gain parameters;

[0152] These are the preset reference baseline parameters;

[0153] It is a natural exponential function.

[0154] Specifically, Used for weighted calculation of subsequent symptom scores; the weight reflects the degree of importance that the symptom scores should be given.

[0155] Typical range for children: 0.5-3.0;

[0156] Child fit value: 1.0-5.0, usually 2.0, used to control the steepness of the Sigmoid curve. The larger, The more sensitive it is to changes in HRV, The smaller the value, the smoother the change.

[0157] Pediatric fit value: 1.5 (mean HRV based on large clinical data), the physiological baseline of HRV balance, HRV = hour, , that is, the midpoint of the Sigmoid curve;

[0158] Example calculation:

[0159] Combined with typical HRV values ​​for children ( , ), analyze the variation of W with HRV:

[0160] :

[0161] ;

[0162] The child's autonomic nervous system was in baseline equilibrium, with a symptom score weight of 0.5 (neutral).

[0163] (like (Inflammation leads to sympathetic excitation)

[0164] ;

[0165] The higher the temperature, the worse the physiological state. The closer a score is to 1, the higher its weight, and the more attention should be paid to the symptoms on that day.

[0166] (like (The autonomic nervous system tends to be in balance)

[0167] ;

[0168] The lower the level, the better the physiological state. The closer the score is to 0, the lower the weight of the symptom score, and the lower the reference value of the symptoms on that day.

[0169] Preset functions To adaptively and dynamically adjust parameters, specifically:

[0170] Obtain the daily heart rate variability index sequence generated by the target child in the initial cycle, and calculate the mean and standard deviation of the daily heart rate variability index sequence;

[0171] Sensitivity gain parameters The update formula is:

[0172] ;

[0173] in, This represents the standard deviation of the daily heart rate variability index series.

[0174] This is the preset reference sensitivity constant;

[0175] It is a very small positive number used to prevent division by zero errors.

[0176] Specifically, by adjusting the sensitivity gain parameter of the Sigmoid function... The design incorporates adaptive, dynamic parameter adjustment to ensure that the sensitivity of the weight calibration matches the HRV fluctuation characteristics of the individual child. This addresses the distortion in weight calibration caused by significant differences in HRV fluctuation amplitude among different children, such as children with severe allergies exhibiting large HRV fluctuations while those with mild symptoms show smaller fluctuations. Ultimately, the goal is to achieve dynamic weight calibration. More tailored to the individual physiological characteristics of children, as detailed below:

[0177] Obtain the initial cycle HRV index sequence: Extract all daily HRV indices generated by the signal processing module during the initial cycle of desensitization treatment for the target child, forming a sequence: (n is the initial cycle number of days);

[0178] Calculate the mean and standard deviation of the HRV series:

[0179] mean ( ): Reflects the average HRV level in the initial cycle of the child, although it does not directly participate in Calculated, but used for individual baseline comparison, formula:

[0180] ;

[0181] Standard deviation ( ): Reflects the fluctuation range (dispersion) of the child's initial cycle HRV, formula:

[0182] ;

[0183] The larger the heart rate, the more drastic the fluctuations in the child's HRV, such as repeated allergy symptoms leading to large fluctuations in the autonomic nervous system state; The smaller the value, the more stable the HRV (smaller fluctuations in physiological state).

[0184] implement Adaptive updates:

[0185] .

[0186] The discrete symptom assessment data of the same day are weighted based on dynamically calibrated weighting coefficients, and the specific formula is as follows:

[0187] ;

[0188] in, No. Daily dynamic weighted total symptom score;

[0189] For the first Daily dynamic calibration weighting coefficient;

[0190] For the first day The original assessment scores for each symptom dimension.

[0191] Specifically, : 0-(5×m) (m is the number of symptom dimensions), quantifying the actual severity of symptoms on the day, with weights calibrated in conjunction with physiological state;

[0192] : It is calculated from the HRV of the day using the Sigmoid function and reflects the degree of calibration of the physiological state of the day on the symptom score;

[0193] Scores: 0-5 (0 = no symptoms, 5 = severe symptoms), core dimensions of childhood allergic rhinitis: (sneeze), (Nasal congestion) (runny nose) (Itchy nose) (Itchy eyes);

[0194] Example calculation:

[0195] Taking a typical scenario of allergic rhinitis in children (m=5 symptom dimensions) as an example, the calculation process is shown intuitively:

[0196] Determine the basic data:

[0197] No. Daily dynamic calibration weight: , that day > Baseline Poor physiological condition, high weight;

[0198] No. Original scores for each symptom on a given day:

[0199] (Sneezing, moderate) (Severe nasal congestion) (Runny nose, moderate) (Itchy nose, mild) (Itchy eyes, mild).

[0200] Perform weighted calculation:

[0201] ;

[0202] ;

[0203] Comparing the results with fixed weights: If the weights are fixed at 0.5, then:

[0204] ;

[0205] The significant difference from the weighted result of 11.44 indicates that when the physiological state is poor, the total symptom score is reasonably amplified, which is more in line with the actual severity of the condition.

[0206] Statistical analysis was performed on the dynamically weighted daily symptom score sequence within the initial period, specifically including:

[0207] The ratio of the standard deviation to the mean of the dynamically weighted daily symptom score sequence within the initial period is calculated, and this ratio is used as the primary indicator of health status.

[0208] Specifically, the steps are as follows:

[0209] Initial period: refers to the early observation phase of desensitization therapy;

[0210] Scoring sequence: Extract the daily dynamic weighted symptom score within the initial period. , forming a sequence:

[0211] ;

[0212] Calculate the average of the sequences to reflect the overall severity of symptoms:

[0213] ;

[0214] The higher the value, the more severe the symptoms are overall in the initial period; the lower the value, the milder the symptoms are overall.

[0215] Calculate the standard deviation of the sequence to reflect the degree of absolute fluctuation in symptoms:

[0216] ;

[0217] The higher the score, the greater the absolute fluctuation of the daily symptom score; the lower the score, the more stable the symptoms.

[0218] Calculate the ratio (coefficient of variation) as the primary indicator of health status:

[0219] .

[0220] The therapeutic potential prediction model was obtained through the following method:

[0221] Acquire physiological signal data and discrete symptom assessment data of historical patient cohorts, and execute data processing procedures to generate corresponding first and second health status indicators for each historical patient.

[0222] Using the generated first and second health status indicators as input features and the final treatment results corresponding to historical patients as training labels, the machine learning model is trained to obtain a treatment potential prediction model.

[0223] The machine learning model is a logistic regression model, and the specific discriminant function is:

[0224] ;

[0225] in, These are model parameters;

[0226] For the intercept term;

[0227] and These are the first and second health status indicators, respectively;

[0228] and These are the weighting coefficients corresponding to the first and second health status indicators, respectively;

[0229] The training process obtains the optimal result through optimization algorithms. , and .

[0230] Specifically, the training method for the efficacy potential prediction model is as follows:

[0231] Training data collection: A multi-center, prospective clinical follow-up study was conducted on children with allergic rhinitis undergoing desensitization treatment from 10 or more tertiary hospitals and 5 or more primary hospitals to ensure data coverage of children from different regions, age groups, and severity of the disease, thus avoiding sample bias. A total of ≥1000 historical cases were included, of which 50%-60% were effective cases (i.e., those with a final effective treatment) and 40%-50% were ineffective cases (i.e., those with a final ineffective treatment) to avoid model bias caused by sample imbalance. Initial desensitization treatment data was collected for each historical case, consistent with the initial treatment period of this system, usually set at 1-3 months, to ensure consistency of input features and long-term final efficacy data. The complete desensitization treatment period was ≥3 years to ensure the authenticity of the labels.

[0232] Training data preprocessing: For each historical patient, strictly follow the process of this system to calculate the first health status index and the second health status index of the corresponding patient to form an input feature set, and normalize the first health status index and the second health status index to eliminate the difference in the units of measurement between the first health status index and the second health status index.

[0233] Initial parameter settings:

[0234] Intercept term The initial value is set to 0.0 to adjust the overall model offset and counteract the impact of normalization on the input features.

[0235] Feature weights : The first health status index, which corresponds to the input feature, is initially set to 0.5, representing the initial influence weight of the first health status index on the efficacy prediction result;

[0236] Feature weights : The second health status index corresponds to the input feature. The initial value is set to 0.5, which represents the initial influence weight of the first health status index on the efficacy prediction result.

[0237] Additional notes: The initial parameters are only the starting values ​​for training. They will be iteratively updated to the optimal parameters through training optimization algorithms. The core principle of setting the initial values ​​is to balance the initial weights and ensure that training converges quickly.

[0238] Discriminant function determined:

[0239] .

[0240] Model training and optimization:

[0241] Stochastic gradient descent (SGD) is used as the model optimization algorithm. The core training objective is to minimize the error between the model's predictions and the true training labels. The model parameters are updated through multiple iterations. To obtain the optimal parameter combination, the specific training and optimization steps are as follows:

[0242] Training objective: Minimize the error between the model's predicted probability and the true labels of the training set samples. The cross-entropy loss function is used, and its formula is:

[0243] ;

[0244] The average loss value of the model. The smaller the value, the smaller the deviation between the model's prediction and the true label, and the better the model's training effect.

[0245] Total number of samples in the training set;

[0246] : No. Example: the true labels of the training set samples;

[0247] : No. The effective probability of model prediction for example training set samples is calculated using a discriminant function;

[0248] : Natural logarithm function;

[0249] Error calculation frequency: Calculate the average loss value of the training set once per iteration. It is used to determine the training effect of the model and the timing of the iteration termination.

[0250] Parameter update: Stochastic gradient descent is used. The direction of parameter update is determined by calculating the partial derivative of the loss function with respect to each model parameter. The parameters are then updated iteratively. Specific steps:

[0251] Partial derivative calculation: for the loss function Find them separately The partial derivatives are used to obtain the gradients of each parameter. The formula for the partial derivatives is:

[0252] loss function right Partial derivatives: ;

[0253] loss function right Partial derivatives: ;

[0254] loss function right Partial derivatives: ;

[0255] In the formula, , The first Example: Normalized values ​​of training set samples.

[0256] Parameter update formula: Iteratively update each parameter based on the partial derivatives. Update formula:

[0257] ;

[0258] : Parameter value at the current iteration number;

[0259] : Updated parameter values;

[0260] The learning rate, with a preset initial value of 0.01, controls the step size for parameter updates, preventing parameter oscillations caused by an excessively large step size and slow training caused by an excessively small step size.

[0261] : The partial derivative of the loss function with respect to this parameter at the current iteration number;

[0262] Parameter update frequency: Update all model parameters once per iteration to ensure synchronized parameter optimization.

[0263] Iteration termination condition:

[0264] Loss convergence: The average loss value of the training set after 100 consecutive iterations. If the change is ≤0.0001, it indicates that the model loss value tends to stabilize and the parameters are close to the optimal value.

[0265] Reaching the maximum number of iterations: Training stops when the number of iterations reaches the preset maximum of 1000, even if the loss value has not fully converged, to avoid overtraining.

[0266] A tracking method for desensitization treatment in children with allergic rhinitis includes the following steps:

[0267] Simultaneously acquire physiological signal data and discrete symptom assessment data of the target children during the initial cycle of desensitization therapy;

[0268] Frequency domain analysis of physiological signal data was performed to calculate the daily heart rate variability index, which reflects the balance between the sympathetic and parasympathetic nervous systems of the autonomic nervous system.

[0269] The daily heart rate variability index is converted into a corresponding dynamic calibration weight coefficient for each day through a preset function. Based on the dynamic calibration weight coefficient, the discrete symptom assessment data of the same day are weighted and calculated to generate a dynamically weighted daily symptom score sequence.

[0270] Statistical analysis was performed on the dynamically weighted daily symptom score sequence within the initial period to calculate and generate a first health status index representing the degree of symptom fluctuation. The daily heart rate variability index within the initial period was arranged in chronological order and linear regression analysis was performed. The slope of the regression line was used as a second health status index representing the trend of physiological state changes.

[0271] The first and second health status indicators are input into the pre-trained efficacy potential prediction model, which outputs the early efficacy potential prediction results for the target children.

[0272] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A tracking system for desensitization treatment of allergic rhinitis in children, characterized in that, include: The data acquisition module is used to simultaneously acquire physiological signal data and discrete symptom assessment data of the target child during the initial period of desensitization treatment; The signal processing module is used to perform frequency domain analysis on physiological signal data and calculate the daily heart rate variability index, which reflects the balance between the sympathetic and parasympathetic nervous systems of the autonomic nervous system. The data fusion module is used to convert the daily heart rate variability index into a corresponding dynamic calibration weight coefficient for each day through a preset function. Based on the dynamic calibration weight coefficient, the discrete symptom assessment data of the same day are weighted and calculated to generate a dynamically weighted daily symptom score sequence. The indicator construction module is used to perform statistical analysis on the dynamically weighted daily symptom score sequence within the initial period, calculate and generate the first health status indicator representing the degree of symptom fluctuation, arrange the daily heart rate variability indicators within the initial period in chronological order and perform linear regression analysis, and use the slope value of the regression line as the second health status indicator representing the trend of physiological state changes. The prediction module is used to input the first and second health status indicators into the pre-trained efficacy potential prediction model and output the early efficacy potential prediction results for the target child.

2. The tracking system for desensitization treatment of allergic rhinitis in children according to claim 1, characterized in that: The data acquisition module includes: Wearable physiological signal acquisition unit is used to continuously acquire raw photoplethysmography (PPG) signals from the target child. The user interaction unit is used to receive daily input of discrete symptom assessment data. The time synchronization unit is used to align and store the raw photoplethysmography pulse wave signal and discrete symptom assessment data based on a unified timestamp to obtain physiological signal data.

3. The tracking system for desensitization treatment of allergic rhinitis in children according to claim 2, characterized in that: The signal processing module is specifically configured as follows: Based on a unified timestamp, signal segments within a preset nighttime sleep time window are extracted from physiological signal data each day, and the signal segments are preprocessed by denoising and normalization. Perform a fast Fourier transform on the preprocessed signal segment and calculate the power spectral density in the low-frequency and high-frequency bands. The daily heart rate variability index is calculated based on the ratio of power spectral density in the low-frequency band to that in the high-frequency band.

4. The tracking system for desensitization treatment of allergic rhinitis in children according to claim 1, characterized in that: The preset function takes the form of a Sigmoid function, and its specific expression is as follows: ； in, The output is the dynamic calibration weight coefficient, and ; The input is the daily heart rate variability index; These are the preset sensitivity gain parameters; These are the preset reference baseline parameters; It is a natural exponential function.

5. The tracking system for desensitization treatment of allergic rhinitis in children according to claim 4, characterized in that: The preset function To adaptively and dynamically adjust parameters, specifically: Obtain the daily heart rate variability index sequence generated by the target child in the initial cycle, and calculate the mean and standard deviation of the daily heart rate variability index sequence; Sensitivity gain parameters The update formula is: ； in, This represents the standard deviation of the daily heart rate variability index series. This is the preset reference sensitivity constant; It is a very small positive number used to prevent division by zero errors.

6. The tracking system for desensitization treatment of allergic rhinitis in children according to claim 4, characterized in that: The discrete symptom assessment data of the same day are weighted based on dynamically calibrated weighting coefficients, and the specific formula is as follows: ； in, No. Daily dynamic weighted total symptom score; For the first Daily dynamic calibration weighting coefficient; For the first day The original assessment scores for each symptom dimension.

7. The tracking system for desensitization treatment of allergic rhinitis in children according to claim 6, characterized in that: Statistical analysis was performed on the dynamically weighted daily symptom score sequence within the initial period, specifically including: The ratio of the standard deviation to the mean of the dynamically weighted daily symptom score sequence within the initial period is calculated, and this ratio is used as the primary indicator of health status.

8. The tracking system for desensitization treatment of allergic rhinitis in children according to claim 1, characterized in that: The therapeutic potential prediction model was obtained through the following method: Acquire physiological signal data and discrete symptom assessment data of historical patients, and execute data processing procedures to generate corresponding first and second health status indicators for each historical patient. Using the generated first and second health status indicators as input features and the final treatment results corresponding to historical patients as training labels, the machine learning model is trained to obtain a treatment potential prediction model.

9. The tracking system for desensitization treatment of allergic rhinitis in children according to claim 8, characterized in that: The machine learning model is a logistic regression model, and the specific discriminant function is: ； in, These are model parameters; For the intercept term; and These are the first and second health status indicators, respectively; and These are the weighting coefficients corresponding to the first and second health status indicators, respectively; The training process obtains the optimal result through optimization algorithms. , and .

10. A method for tracking desensitization treatment in children with allergic rhinitis, characterized in that, Includes the following steps: Simultaneously acquire physiological signal data and discrete symptom assessment data of the target children during the initial cycle of desensitization therapy; Frequency domain analysis of physiological signal data was performed to calculate the daily heart rate variability index, which reflects the balance between the sympathetic and parasympathetic nervous systems of the autonomic nervous system. The daily heart rate variability index is converted into a corresponding dynamic calibration weight coefficient for each day through a preset function. Based on the dynamic calibration weight coefficient, the discrete symptom assessment data of the same day are weighted and calculated to generate a dynamically weighted daily symptom score sequence. Statistical analysis was performed on the dynamically weighted daily symptom score sequence within the initial period to calculate and generate a first health status index representing the degree of symptom fluctuation. The daily heart rate variability index within the initial period was arranged in chronological order and linear regression analysis was performed. The slope of the regression line was used as a second health status index representing the trend of physiological state changes. The first and second health status indicators are input into the pre-trained efficacy potential prediction model, which outputs the early efficacy potential prediction results for the target children.

Images