Family environment intervention method and system for chronic respiratory disease patient

Abstract

The invention discloses a chronic respiratory disease patient family environment intervention method and system. The method comprises the steps of sensing module construction, multi-node linkage traceability, sampling scheduling optimization and environment intervention. The invention relates to the technical field of environmental health monitoring and intelligent intervention, and the method achieves the real-time collection of household environment data through constructing a multi-mode micro sensor integrated with particulate matter, gas, temperature, humidity and allergen detection. An improved propagation consistency index and parameter gradient mutation analysis are adopted to identify the propagation path and health risk of the environmental factors inducing the symptoms; the sampling frequency is dynamically optimized in combination with a lightweight long-short-term memory network, so that the data acquisition efficiency is improved; and finally, the linkage regulation and control and health behavior prompt of the individual-oriented intelligent equipment are realized. The method has the advantages of individuation, high response and embeddability, and the intelligent level of chronic respiratory disease management is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental health monitoring and intelligent intervention, and specifically refers to a method and system for home environmental intervention for patients with chronic respiratory diseases. Background Art

[0002] A method and system for home environmental intervention for patients with chronic respiratory diseases generally refers to patients with chronic respiratory diseases such as asthma and chronic obstructive pulmonary disease (COPD). For factors in their daily living environment that may induce or exacerbate the condition (such as air pollution, abnormal temperature and humidity, allergens, etc.), through environmental monitoring, data analysis and intelligent control means, potential risks are identified in a timely manner and intervention measures such as air purification, ventilation adjustment or behavior guidance are implemented, so as to reduce the frequency of symptom attacks, improve the quality of life, and delay the progression of the disease. This system generally includes a sensor acquisition module, a data processing module and an intervention execution module, and has the characteristics of long-term operation, individual adaptation and high responsiveness.

[0003] However, in the existing methods for home environmental intervention for patients with chronic respiratory diseases, there are technical problems such as the lack of identification of the combined incentives of multiple microenvironmental factors, the lack of dynamic intervention adjustment strategies, the lag in system response, and the inability to adapt to the differences in patients' conditions in a personalized manner; in the process of risk tracing of environmental factor sampling nodes, there are technical problems such as the inability to effectively identify the pollution transmission path, the lack of symptom causality deduction ability, and the rough health risk scoring; in the process of optimizing the sampling scheduling of sensing nodes, there are technical problems such as waste of sensing resources, sampling not dynamically adjusted according to environmental state changes, and high algorithm computational complexity making it difficult to be deployed on edge devices. Summary of the Invention

[0004] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a method and system for home environmental intervention for patients with chronic respiratory diseases. Aiming at the technical problems in the existing methods for home environmental intervention for patients with chronic respiratory diseases, such as the lack of identification of the combined incentives of multiple microenvironmental factors, the lack of dynamic intervention adjustment strategies, the lag in system response, and the inability to adapt to the differences in patients' conditions in a personalized manner, this solution creatively adopts a comprehensive home environmental intervention method combining multi-modal micro-sensing module design, multi-node disease incentive risk linkage tracing and sampling node scheduling combined with disease incentive characteristics optimization, and realizes the induction factors (PM2.5, , co - modeling and co - regulation of the linkage (such as sudden changes in temperature and humidity, and dust mite allergens); in view of the technical problems existing in the risk tracing process of existing environmental factor sampling nodes, such as the inability to effectively identify the pollution transmission path, the lack of symptom causal deduction ability, and the rough health risk scoring, this solution creatively uses the spatio - temporal correlation parameter mutation analysis method combining the improved propagation consistency index and parameter gradient change to conduct multi - node linkage tracing, obtaining the characteristic data of the causes of respiratory disease symptoms, and realizing the full - chain analysis among "factor fluctuation - node collaboration - transmission path - symptom risk"; in view of the technical problems existing in the sampling scheduling optimization of existing sensing nodes, such as waste of sensing resources, sampling not being dynamically adjusted according to environmental state changes, and high algorithm computational complexity making it difficult to be deployed on edge devices, this solution creatively uses the lightweight improved long - short - term memory network prediction method combining parameter fluctuation characteristics to optimize the sampling scheduling of the sensing module, obtaining reference data for predictive sampling strategies, and realizing the state - aware dynamic allocation of the perception frequency of the home micro - environment.

[0005] The technical solution adopted by the present invention is as follows: A method for intervening in the home environment of patients with chronic respiratory diseases provided by the present invention includes the following steps:

[0006] Step S1: Construction of the sensing module;

[0007] Step S2: Multi - node linkage tracing;

[0008] Step S3: Optimization of sampling scheduling;

[0009] Step S4: Home environment intervention.

[0010] Further, in step S1, the construction of the sensing module is used to construct a multi - modal micro - sensor module required for the home environment of patients with chronic respiratory diseases and realize real - time data acquisition. Specifically, a multi - modal micro - sensing module including particulate matter detection, gas detection, temperature and humidity detection, and dust mite allergen detection is constructed in sequence, and through the construction of the multi - modal micro - sensing module, home environment sensing data is collected to obtain an environmental original data set.

[0011] Further, in step S2, the multi - node linkage tracing is used for spatio - temporal correlation analysis of node data and identification of invisible health threats. Specifically, based on the environmental original data set, the spatio - temporal correlation parameter mutation analysis method combining the improved propagation consistency index and parameter gradient change is used to conduct multi - node linkage tracing, obtaining the characteristic data of the causes of respiratory disease symptoms, including the following steps:

[0012] Step S21: Pre - processing of node data. Specifically, based on the environmental original data set, by sequentially performing timestamp interpolation, Z - score normalization, and unit conversion on the asynchronous data of the sensing module, a multi - dimensional time - series matrix is constructed to obtain a multi - node basic data set;

[0013] The multi-node basic data set includes particulate matter detection data, gas detection data, temperature and humidity detection data, and dust mite allergen detection data;

[0014] Step S22: Propagation consistency index modeling, which is used to measure the synchronization of factor fluctuation propagation between different nodes. Specifically, based on the multi-node basic data set, by introducing a directional fluctuation gradient and an exponential time decay parameter, the correlation coefficient model is improved, and a propagation consistency index model is constructed;

[0015] Step S23: Parameter gradient change identification. Specifically, based on the multi-node basic data set, by constructing a locally adaptive parameter gradient mutation index based on a sliding window, the abnormal change modeling of environmental factors is carried out to obtain environmental parameter gradient mutation index data;

[0016] Step S24: Spatiotemporal interaction path modeling, which is used to dynamically construct the propagation path of a suspected pollution source from the starting point to the symptom node. Specifically, based on the propagation consistency index model and the environmental parameter gradient mutation index data, by sequentially constructing time-series propagation graph data and screening the environmental factor pollution propagation path, spatiotemporal interaction path modeling is carried out to obtain spatiotemporal interaction path reference data;

[0017] Step S25: Potential health risk identification. Specifically, based on the spatiotemporal interaction path reference data, by calculating the health risk score of each spatiotemporal interaction path based on an improved health risk weighting function, and by screening the spatiotemporal interaction path according to the health risk score, potential risk identification reference data is obtained;

[0018] Step S26: Symptom alignment linkage tracing. Specifically, by constructing a disease risk rule matching table, rule mapping modeling of symptom types and environmental factors is carried out, and through causal weight scoring and incentive feature screening, incentive feature data of respiratory disease symptoms is obtained.

[0019] Further, in step S3, the sampling scheduling optimization is used to adjust the sensor sampling frequency. Specifically, based on the incentive feature data of respiratory disease symptoms and the environmental original data set, a lightweight improved long short-term memory network prediction method combined with parameter fluctuation characteristics is used to optimize the sampling scheduling of the sensing module, and predictive sampling strategy reference data is obtained, including the following steps:

[0020] Step S31: Sampling state classification data preprocessing. Specifically, by based on the incentive feature data of respiratory disease symptoms and the environmental original data set, data annotation of sampling state categories is carried out to obtain sampling state classification data;

[0021] The sampling state classification data includes a stable state, a warning state, and a dangerous state;

[0022] Step S32: Parameter fluctuation partitioning, specifically by constructing a parameter fluctuation attention index, performing sampling attention partitioning, and obtaining parameter fluctuation partition data;

[0023] The parameter fluctuation attention index is specifically calculated by weighted calculation in combination with the parameter sharp jump score, the propagation distance normalization value, and the sensitivity score of the environmental factor;

[0024] The parameter sharp jump score is specifically calculated by combining the sliding window method with threshold setting;

[0025] Step S33: Lightweight prediction network improvement, specifically by using a lightweight long short-term memory neural network and combining a custom parameter fluctuation partition gating function to perform lightweight prediction network improvement, training a prediction network model based on the sampling state classification data, obtaining a lightweight prediction network model, and using the lightweight prediction network model to perform parameter distribution prediction based on the respiratory disease symptom inducement feature data and the environmental original data set to obtain sampling node prediction state data;

[0026] Step S34: Sampling strategy generation, specifically by constructing a scoring function to adjust the sampling frequency based on the sampling node prediction state data to obtain node sampling frequency reference data;

[0027] Step S35: Dynamic adjustment of the sampling scheduling strategy, specifically by dynamically adjusting the multimodal micro-sensing module based on the node sampling frequency reference data to obtain sampling scheduling strategy reference data.

[0028] Further, in step S4, the home environment intervention is used to perform targeted device control and user behavior prompts, specifically by performing comprehensive home environment intervention based on the respiratory disease symptom inducement feature data and the predictive sampling strategy reference data to obtain personalized respiratory protection adjustment reference data.

[0029] A home environment intervention system for chronic respiratory disease patients provided by the present invention includes a sensing and acquisition module, a detection and traceability module, a sampling and regulation module, and an environment intervention module;

[0030] The sensing and acquisition module is used for constructing a sensing module. By constructing the sensing module, an environmental original data set is obtained, and the environmental original data set is sent to the detection and traceability module, the sampling and regulation module, and the environment intervention module;

[0031] The detection and traceability module is used for multi-node linkage traceability. By multi-node linkage traceability, predictive sampling strategy reference data is obtained, and the predictive sampling strategy reference data is sent to the sampling and regulation module;

[0032] The sampling control module is used for sampling scheduling optimization. Through sampling scheduling optimization, reference data of sampling scheduling strategies is obtained and sent to the environmental intervention module.

[0033] The environmental intervention module is used for home environmental intervention. Through home environmental intervention, reference data of sampling scheduling strategies is obtained.

[0034] The beneficial effects achieved by the present invention using the above solution are as follows:

[0035] (1) Aiming at the technical problems existing in the existing home environmental intervention methods for patients with chronic respiratory diseases, such as the lack of identification of comprehensive linkage incentives of microenvironment multi-factors, the lack of dynamic intervention adjustment strategies, the lag of system response, and the inability to adapt to the differences in patients' conditions individually, this solution creatively adopts a comprehensive home environmental intervention method combining multi-modal micro-sensing module design, multi-node disease incentive risk linkage traceability, and sampling node scheduling combined with disease incentive characteristics optimization, realizing the linkage modeling and linkage control of inducing factors (PM2.5, , sudden changes in temperature and humidity, dust mite allergens) in the home environment of chronic respiratory patients;

[0036] (2) Aiming at the technical problems existing in the risk traceability process of existing environmental factor sampling nodes, such as the inability to effectively identify pollution transmission paths, the lack of symptom causal deduction ability, and the rough health risk scoring, this solution creatively adopts a spatio-temporal correlation parameter mutation analysis method combining improved propagation consistency index and parameter gradient change for multi-node linkage traceability to obtain symptom incentive characteristic data of respiratory diseases, realizing the full-chain analysis among "factor fluctuation - node collaboration - transmission path - symptom risk";

[0037] (3) Aiming at the technical problems existing in the sampling scheduling optimization process of existing sensing nodes, such as waste of sensing resources, sampling not dynamically adjusted according to environmental status changes, and high algorithm computational complexity making it difficult to be deployed on edge devices, this solution creatively adopts a lightweight improved long short-term memory network prediction method combining parameter fluctuation characteristics for sampling scheduling optimization of sensing modules to obtain predictive sampling strategy reference data, realizing state-aware dynamic allocation of the perception frequency of the home microenvironment. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a schematic flowchart of a home environmental intervention method for patients with chronic respiratory diseases provided by the present invention;

[0039] Figure 2 It is a schematic diagram of a home environmental intervention system for patients with chronic respiratory diseases provided by the present invention;

[0040] Figure 3It is a schematic flowchart of multi-node linkage traceability for step S2;

[0041] Figure 4 It is a schematic flowchart of sampling scheduling optimization for step S3.

[0042] The accompanying drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, but do not constitute a limitation to the present invention. Detailed implementation manners

[0043] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0044] In the description of the present invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation to the present invention.

[0045] Example 1, refer to Figure 1 , a method for family environment intervention for patients with chronic respiratory diseases provided by the present invention, the method includes the following steps:

[0046] Step S1: Construction of the sensing module;

[0047] Step S2: Multi-node linkage traceability;

[0048] Step S3: Sampling scheduling optimization;

[0049] Step S4: Family environment intervention.

[0050] By performing the above operations, in the existing method for family environment intervention for patients with chronic respiratory diseases, there are technical problems such as the lack of recognition of the combined incentives of multiple microenvironment factors, the lack of dynamic intervention adjustment strategies, the lag of system response, and the inability to adapt to the differences in patient conditions individually. This solution creatively adopts a comprehensive family environment intervention method combining multi-modal micro-sensing module design, multi-node disease incentive risk linkage traceability, and sampling node scheduling combined with disease incentive characteristics optimization, and realizes the induction factors (PM2.5, , co - modeling and co - regulation of the linkage (such as sudden changes in temperature and humidity, dust mite allergens); for example, in the homes of elderly asthma patients during the plum rain season, the combined changes in dust mite concentration and relative humidity may trigger airway allergic reactions within a short period of time, and traditional systems cannot provide early warnings. However, the present invention can combine multi - modal sensing data with spatio - temporal propagation models to identify risk factors in advance and link execution devices such as air conditioners and dehumidifiers, improving the timeliness of intervention and individual adaptability.

[0051] Example two, refer to Figure 1 and Figure 2 , in step S1, the construction of the sensing module is used to construct a multi - modal micro - sensor module required for the home environment of chronic respiratory disease patients and realize real - time data collection. Specifically, a multi - modal micro - sensing module including particulate matter detection, gas detection, temperature and humidity detection, and dust mite allergen detection is constructed in sequence. By constructing the multi - modal micro - sensing module, home environment sensing data is collected to obtain an environmental original data set;

[0052] Preferably, the multi - modal micro - sensing module includes a particulate matter detection module, a gas detection module, a temperature and humidity detection module, a dust mite allergen detection module, a communication module, and a power supply module;

[0053] Table 1 is a structural schematic table of the multi - modal micro - sensing module. As shown in the table, for the particulate matter detection module, a laser scattering type PM2.5 / PM10 sensor is selected, with a detection range of 0 - 1000 μg / m³, an error not greater than ±10%, and a resolution of 1 μg / m³, which is used for real - time monitoring of common particulate matter pollution sources such as kitchen fumes and dust accumulation;

[0054] The gas detection module includes a VOC detection sub - module and a detection sub - module; for the VOC detection sub - module, an SGP30 metal oxide semiconductor gas sensor is used, with a detection range covering common decoration pollutants and volatile components of cleaning agents, and having the characteristics of fast response and low power consumption; for the detection sub - module, a SenseAir S8 sensor based on the non - dispersive infrared (NDIR) principle is used, with a detection range of 400 - 5000 ppm and an accuracy of ±(50 ppm + 3%), which is used to reflect the ventilation state and the degree of carbon dioxide accumulation;

[0055] The temperature and humidity detection module uses an integrated digital temperature and humidity sensor SHT31, with a temperature measurement accuracy of ±0.3℃ and a humidity accuracy of ±2% RH, which is used to assist in evaluating the risk of allergen growth and regulating respiratory comfort, and assist in establishing a correlation model between the temperature and humidity environment and symptom induction;

[0056] The dust mite allergen detection module uses immunochromatographic test strips combined with camera colorimetric analysis to detect the main allergen, Der p1, a derivative of dust mite. Users can regularly replace the test strips and perform colorimetric concentration analysis to achieve a quantitative assessment of dust mite levels.

[0057] The communication module adopts Bluetooth BLE 5.2 low-power communication protocol, supports multi-node networking and scheduled data broadcasting;

[0058] The power supply module can be powered by a rechargeable button battery solution and a USB power supply solution. The rechargeable button battery solution supports low-frequency sampling operation for more than one month and is suitable for low-light environments such as the back of furniture and under beds. The USB power supply solution is suitable for long-term fixed deployment locations.

[0059] In terms of structural integration, all sensors are connected via a unified I²C bus, with the main control chip (STM32L or ESP32 series) performing polling sampling. The sensors can be configured to enter sleep or active states through software, achieving dynamic power consumption management.

[0060] Table 1 Schematic diagram of the structure of the multimodal micro-sensor module

[0061]

[0062] The multimodal micro-sensor module described in this embodiment can be flexibly deployed according to the type of home room, supports continuous monitoring of key health factors such as particulate matter, volatile gases, temperature and humidity, and biological allergens, and supports sampling scheduling optimization, health risk reasoning, and home environment intervention control through subsequent algorithms, providing patients with chronic respiratory diseases with efficient and intelligent environmental support means.

[0063] Example 3, see Figure 1 、 Figure 2 and Figure 3 This embodiment is based on the above embodiment. In step S2, the multi-node linkage tracing is used for spatiotemporal correlation analysis of node data and identification of invisible health threats. Specifically, based on the original environmental data set, a spatiotemporal correlation parameter mutation analysis method combining an improved propagation consistency index and parameter gradient changes is adopted to perform multi-node linkage tracing to obtain characteristic data of respiratory disease symptom inducers, including the following steps:

[0064] Step S21: Node data preprocessing, specifically, constructing a multi-dimensional time series matrix by performing timestamp interpolation, Z-score normalization, and unit conversion on the asynchronous data of the sensor module based on the original environmental data set, thereby obtaining a multi-node basic data set;

[0065] The multi-node basic data set includes particulate matter detection data, gas detection data, temperature and humidity detection data, and dust mite allergen detection data. The calculation formula is:

[0066] ;

[0067] In the formula, X t is the multi-node basic data set, is the particulate matter detection data, is the gas detection data, is the temperature detection data, is the humidity detection data, is the dust mite allergen detection data;

[0068] Step S22: Propagation consistency index modeling, which is used to measure the synchronization of factor fluctuation propagation between different nodes. Specifically, based on the multi-node basic data set, by introducing a directional fluctuation gradient and an exponential time decay parameter, the correlation coefficient model is improved, and a propagation consistency index model is constructed;

[0069] The calculation formula of the propagation consistency index model is:

[0070] ;

[0071] In the formula, C i,j (t) is the propagation consistency index model, i is the sampling node index, j is the sampling adjacent node index, T is the total number of sampling times, is the parameter gradient time index, t is the time index, is the parameter gradient of the i-th sampling node at time t, is the parameter gradient of the j-th sampling adjacent node after a delay of to capture the time delay characteristics during environmental pollution propagation, is the magnitude of the parameter gradient of the i-th sampling node at time t, is a very small correction positive number, as a whole is the exponential time decay parameter, which is used to represent that as the time delay parameter increases, the impact of environmental pollution propagation gradually weakens, is the time decay coefficient, which is default set to 0.3;

[0072] Step S23: Parameter gradient change identification. Specifically, based on the multi-node basic data set, by constructing a local adaptive parameter gradient mutation index based on a sliding window, the abnormal change modeling of environmental factors is carried out to obtain environmental parameter gradient mutation index data;

[0073] The calculation formula for the abnormal change modeling is as follows:

[0074] ;

[0075] In the formula, is the environmental parameter gradient mutation index data, is the parameter gradient of the i-th sampling node at time t, is the historical parameter gradient mean value of the i-th sampling node, is the standard deviation value of the parameter gradient of the i-th sampling node;

[0076] Step S24: Spatiotemporal interaction path modeling, which is used to dynamically construct the propagation path of the suspected pollution source from the starting point to the symptom node. Specifically, based on the propagation consistency index model and the environmental parameter gradient mutation index data, through sequentially constructing the time-series propagation graph data and screening the environmental factor pollution propagation path, spatiotemporal interaction path modeling is performed to obtain spatiotemporal interaction path reference data;

[0077] The construction of the time-series propagation graph data is specifically to use the sampling points as graph nodes, the pollution propagation path as edges, and based on the environmental parameter gradient mutation index data and the propagation consistency index model, define the edge weights and use the edge weights as the propagation intensity weights;

[0078] The calculation formula for the propagation intensity weight is as follows:

[0079] ;

[0080] In the formula, is the propagation intensity weight, is the environmental parameter gradient mutation index data, C i,j (t) is the propagation consistency index model, E is the edge structure of the time-series propagation graph data, where, is used to indicate that the sampling node i and the sampling adjacent node j are adjacent nodes and are connected in the home space layout;

[0081] The screening of the environmental factor pollution propagation path specifically uses a graph traversal algorithm to perform threshold screening on the propagation intensity weights corresponding to each environmental factor. Through the threshold screening, the propagation path of the suspected pollution source from the starting point to the symptom node is screened as the spatiotemporal interaction path reference data;

[0082] Preferably, the threshold is set to 1.8, and the graph traversal algorithm uses the networkx.dfs_edges() algorithm to extract the pollution propagation paths with a path length less than or equal to 4;

[0083] The calculation formula for the threshold screening is as follows:

[0084] ;

[0085] In the formula, is the propagation intensity weight, is the threshold, and path is the total set of all edges;

[0086] Step S25: Potential health risk identification, specifically, based on the spatio-temporal interaction path reference data, through an improved health risk weighting function, calculate the health risk score for each spatio-temporal interaction path, and screen the spatio-temporal interaction paths according to the health risk score to obtain potential risk identification reference data;

[0087] The calculation formula of the improved health risk weighting function is:

[0088] ;

[0089] In the formula, is the health risk score, where is the spatio-temporal interaction path reference data, is the propagation intensity weight, with a default value of 0.5, and PathStrength is the sum of all propagation intensity edge weights on the path, is the sensitivity weight, with a default value of 0.3, and NodeRiskIndex is the sensitivity parameter of the sampling point at the end of the path, used to represent the risk contribution degree to symptom triggering, is the sensitization intensity weight, with a default value of 0.2, and FactorSeverity is the sensitization intensity score of the dominant environmental factor in the path;

[0090] Step S26: Symptom alignment and linkage traceability, specifically, by constructing a disease risk rule matching table, perform rule mapping modeling of symptom types and environmental factors, and through combining the health risk score, perform causal weight scoring and incentive feature screening to obtain respiratory disease symptom incentive feature data;

[0091] Preferably, Table 2 is the association rule mapping instance table for the rule mapping modeling. As shown in the table, the symptom types specifically include dry cough, nocturnal asthma, laryngeal irritation, and nasal congestion and sneezing, and the corresponding weight scores are 1.0, 0.9, 0.85, and 0.8 respectively. The weight scores are used to assist in the calculation of the sensitization intensity score; the number of matching factors is used to represent the number of associated environmental factors that can trigger a certain symptom type; the number of matching factors and the weight scores are valued and calculated according to the association rule mapping instance table;

[0092] The sensitization intensity score is calculated according to the number of matching factors and the weight scores in the association rule mapping instance table, and the calculation formula is:

[0093] ;

[0094] In the formula, is the sensitization intensity score of the dominant environmental factor in the path calculated based on the spatio-temporal interaction path reference data, is the number of matching factors corresponding to the symptom type s, K is the total number of environmental factors in the path, k is the environmental factor index, and I k is the symptom-environment matching indication parameter, which is used to represent the matching situation between the k-th environmental factor and the previous symptom type s. When the k-th environmental factor matches the symptom type s, the value is 1, and when it does not match, the value is 0. W s is the weight score of the symptom type s.

[0095] Table 2 Association rule mapping instance table for rule mapping modeling

[0096]

[0097] By performing the above operations, aiming at the technical problems of being unable to effectively identify the pollution propagation path, lacking the ability of symptom causal deduction, and having rough health risk scoring during the risk tracing process of existing environmental factor sampling nodes, this solution creatively adopts a spatio-temporal correlation parameter mutation analysis method combining the improved propagation consistency index and parameter gradient change to conduct multi-node linkage tracing, obtaining the characteristic data of the inducing factors of respiratory disease symptoms, and realizing the full-chain analysis among "factor fluctuation - node collaboration - propagation path - symptom risk"; specifically, taking the example that kitchen fume in the home affects the PM2.5 concentration in the bedroom and induces nocturnal dyspnea in COPD, this solution can identify the aerosol migration path during the peak period of fume through the directional propagation index, then combine the sliding window mutation model to identify the "key transition point", and finally deduce the inducing factor combination based on the symptom-factor rule base, providing a targeted decision-making basis for personalized intervention.

[0098] Example 4, referring to Figure 1 , Figure 2 and Figure 4 , based on the above example, in step S3, the sampling scheduling optimization is used to adjust the sensor sampling frequency. Specifically, according to the characteristic data of the inducing factors of respiratory disease symptoms and the original environmental data set, a lightweight improved long short-term memory network prediction method combining parameter fluctuation characteristics is adopted to optimize the sampling scheduling of the sensing module, obtaining predictive sampling strategy reference data, including the following steps:

[0099] Step S31: Sampling status classification data preprocessing, specifically, by labeling the sampling status category data according to the characteristic data of the inducing factors of respiratory disease symptoms and the original environmental data set, the sampling status classification data is obtained;

[0100] The sampled state classification data includes a stable state, a warning state, and a dangerous state;

[0101] Step S32: Parameter fluctuation partitioning, specifically, by constructing a parameter fluctuation attention index, performing sampling attention partitioning, and obtaining parameter fluctuation partition data;

[0102] The parameter fluctuation attention index is specifically calculated by weighted combination of the parameter sharp jump score, the propagation distance normalization value, and the sensitivity score of the environmental factor. The calculation formula is:

[0103] ;

[0104] In the formula, is the parameter fluctuation attention index, is the parameter sharp jump weight, with a default value of 0.4, JumpScore t is the parameter sharp jump score, which is specifically calculated by combining the sliding window method with threshold setting, is the propagation distance weight, with a default value of 0.4, Proximity t is the propagation distance normalization value, which is used to represent the propagation distance from the sampling node i to the end of the propagation path corresponding to the closest respiratory disease symptom inducement feature data, is the environmental factor sensitivity weight, with a default value of 0.2, FactorWeight is the sensitivity score of the environmental factor, which is specifically calculated according to the weight scores in Table 2;

[0105] Step S33: Improvement of the lightweight prediction network, specifically, adopting a lightweight long short-term memory neural network and combining a custom parameter fluctuation partition gating function to improve the lightweight prediction network. Based on the sampled state classification data, the prediction network model is trained to obtain a lightweight prediction network model, and by using the lightweight prediction network model, based on the respiratory disease symptom inducement feature data and the environmental original data set, parameter distribution prediction is performed to obtain the sampled node prediction state data;

[0106] The calculation formula for the improvement of the lightweight prediction network is:

[0107] ;

[0108] In the formula, is the output hidden state of the lightweight prediction network model, is the custom parameter fluctuation partition gating function, where is the parameter fluctuation attention index, LSTM(·) is the single-layer long short-term memory neural network model function, x t is the sampled state classification data, h t-1 is the hidden state of the previous moment, The whole is the activation partition, which is used to identify dynamic mutations. The whole is the retention partition, which is used to maintain prediction stability.

[0109] The calculation formula of the custom parameter fluctuation partition gating function is:

[0110] ;

[0111] In the formula, is the upper gating threshold, is the lower gating threshold. The custom parameter fluctuation partition gating function is specifically used to activate the prediction ability in the high-fluctuation state and only retain the historical state in the low-fluctuation state, reducing the risk of overfitting.

[0112] Preferably, the value of the upper gating threshold is set to 0.7, and the value of the lower gating threshold is set to 0.3.

[0113] Preferably, Table 3 is an example table of the parameters of the lightweight prediction network model. As shown in the table, the parameter categories of the lightweight prediction network model include input dimension, number of hidden units, time step, gating activation function, network activation function, and loss function.

[0114] Table 3 Example Table of the Parameters of the Lightweight Prediction Network Model

[0115]

[0116] Step S34: Sampling strategy generation, specifically, according to the predicted state data of the sampling nodes, the sampling frequency is adjusted by constructing a scoring function to obtain the reference data of the node sampling frequency.

[0117] The calculation formula of the scoring function is:

[0118] ;

[0119] In the formula, f i (t) is the reference data of the node sampling frequency, f max is the high-frequency sampling frequency, f med is the medium-frequency sampling frequency, f min is the low-frequency sampling frequency, is the predicted state data of the sampling nodes, is the set of sampling nodes whose sampling states are classified as warning state and dangerous state, is the scoring threshold;

[0120] Preferably, the specific sampling frequency of the high-frequency sampling frequency is set to be sampled once every 1 minute, the specific sampling frequency of the medium-frequency sampling frequency is set to be sampled once every 5 minutes, and the specific sampling frequency of the low-frequency sampling frequency is set to be sampled once every 15 minutes; the default value of the scoring threshold is set to 0.15;

[0121] Step S35: Dynamically adjust the sampling scheduling strategy, specifically, dynamically adjust the multimodal micro-sensing module according to the node sampling frequency reference data to obtain sampling scheduling strategy reference data.

[0122] By performing the above operations, in the process of optimizing the sampling scheduling of existing sensing nodes, there are technical problems such as waste of sensing resources, sampling not dynamically adjusted according to environmental state changes, and high algorithm calculation complexity making it difficult to be deployed on edge devices. This solution creatively adopts a lightweight improved long short-term memory network prediction method combined with parameter fluctuation characteristics to optimize the sampling scheduling of the sensing module, obtain predictive sampling strategy reference data, and achieve state-aware dynamic allocation of the home microenvironment perception frequency; for example, during the peak period of children's respiratory tract sensitivity, this solution can identify the mutation trends of specific parameters (such as TVOC, humidity), and dynamically switch the perception frequency according to the partition of "stable state - warning state - dangerous state", avoiding information delay or excessive sensor energy consumption caused by a fixed sampling frequency. At the same time, the algorithm adopts a lightweight gating mechanism, which is suitable for deployment in a home-level embedded system to ensure the balance of performance and response speed.

[0123] Example Five, refer to Figure 1 、 Figure 2 , based on the above embodiment, in step S4, the home environment intervention is used to perform targeted device control and user behavior prompts, specifically, based on the respiratory disease symptom cause characteristic data and the predictive sampling strategy reference data, perform comprehensive home environment intervention to obtain personalized respiratory protection adjustment reference data;

[0124] Preferably, the comprehensive home environment intervention can be realized through an existing smart home system, specifically including device control intervention, user prompt intervention, and intelligent output form intervention;

[0125] The device control intervention specifically generates a control instruction by calling a preset control interface based on the high-risk factors identified in the respiratory disease symptom cause characteristic data, and performs real-time adjustment of environmental factors; the preset control interface includes the switch control of the intelligent air purifier, the humidity adjustment threshold setting of the dehumidifier, and the window opening and closing control;

[0126] The user prompt intervention specifically calls the voice broadcast prompt or the mobile APP push interface to perform user operation suggestion prompt intervention based on the status level output in the predictive sampling strategy reference data;

[0127] The intelligent output form intervention specifically structures the output data of the device control intervention and the user prompt intervention into personalized respiratory protection adjustment reference data for storage and reading; the personalized respiratory protection adjustment reference data specifically includes control object parameters, target environmental factor parameters, control strategy parameters, user suggestion parameters, and timestamps.

[0128] The specific intervention execution method adopted in this embodiment belongs to a general functional component that can be directly called and implemented by those skilled in the art in the existing smart home control system. Through a unified intervention strategy generation process, it realizes the personalized and multi-factor environment linkage adjustment ability for patients with chronic respiratory diseases.

[0129] Example 6, refer to Figure 1 and Figure 2 , based on the above embodiment, a home environment intervention system for patients with chronic respiratory diseases provided by the present invention includes a sensing and acquisition module, a detection and traceability module, a sampling and regulation module, and an environment intervention module;

[0130] The sensing and acquisition module is used for constructing a sensing module. Through the construction of the sensing module, an environmental original data set is obtained, and the environmental original data set is sent to the detection and traceability module, the sampling and regulation module, and the environment intervention module;

[0131] The detection and traceability module is used for multi-node linkage traceability. Through multi-node linkage traceability, predictive sampling strategy reference data is obtained, and the predictive sampling strategy reference data is sent to the sampling and regulation module;

[0132] The sampling and regulation module is used for optimizing sampling scheduling. Through optimizing sampling scheduling, sampling scheduling strategy reference data is obtained, and the sampling scheduling strategy reference data is sent to the environment intervention module;

[0133] The environment intervention module is used for home environment intervention. Through home environment intervention, sampling scheduling strategy reference data is obtained.

[0134] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0135] Although embodiments of the present invention have been shown and described, those of ordinary skill in the art will appreciate that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention.

[0136] The present invention and its embodiments have been described above. Such description is not restrictive. What is shown in the drawings is only one of the embodiments of the present invention, and the actual structure is not limited thereto. In general, if those of ordinary skill in the art are inspired by it and, without departing from the purpose of the present invention, design similar structural forms and embodiments to this technical solution without creative efforts, they shall fall within the protection scope of the present invention.

Claims

1. A method for family environment intervention for patients with chronic respiratory diseases, characterized in that: The method includes the following steps: Step S1: Sensing module construction to obtain an original environmental data set; Step S2: Multi-node linkage traceability. Using a spatio-temporal correlation parameter mutation analysis method that combines an improved propagation consistency index and parameter gradient changes to perform multi-node linkage traceability to obtain respiratory disease symptom cause characteristic data, including the following steps: Step S21: Node data preprocessing; Step S22: Propagation consistency index modeling. By introducing a directional fluctuation gradient and an exponential time decay parameter, the correlation coefficient model is improved; Step S23: Parameter gradient change identification; Step S24: Spatio-temporal interaction path modeling; Step S25: Potential health risk identification; Step S26: Symptom alignment linkage traceability; Step S3: Sampling schedule optimization. Using a lightweight improved long short-term memory network prediction method that combines parameter fluctuation characteristics to perform sampling schedule optimization of the sensing module to obtain predictive sampling strategy reference data, including the following steps: Step S31: Sampling state classification data preprocessing; Step S32: Parameter fluctuation partitioning; Step S33: Lightweight prediction network improvement. Using a lightweight long short-term memory neural network and combining a custom parameter fluctuation partitioning gating function to perform lightweight prediction network improvement; Step S34: Sampling strategy generation; Step S35: Dynamic adjustment of the sampling schedule strategy; Step S4: Home environment intervention to obtain personalized respiratory protection adjustment reference data.

2. The family environment intervention method for patients with chronic respiratory diseases according to claim 1, characterized in that: The sensing module construction is used to construct a multi-modal micro-sensor module required for the home environment of chronic respiratory disease patients and realize real-time data collection. Specifically, a multi-modal micro-sensing module including particulate matter detection, gas detection, temperature and humidity detection, and dust mite allergen detection is constructed in sequence. And through the construction of the multi-modal micro-sensing module, home environment sensing data is collected to obtain an original environmental data set.

3. The method for family environment intervention for patients with chronic respiratory diseases according to claim 2, wherein: In step S2, the multi-node linkage traceability is used for spatio-temporal correlation analysis of node data and identification of hidden health threats. Specifically, based on the original environmental data set, a spatio-temporal correlation parameter mutation analysis method that combines an improved propagation consistency index and parameter gradient changes is used to perform multi-node linkage traceability to obtain respiratory disease symptom cause characteristic data, including the following steps: Step S21: Node data preprocessing. Specifically, based on the original environmental data set, by sequentially performing timestamp interpolation, Z-score normalization, and unit conversion on the asynchronous data of the sensing module, a multi-dimensional time series matrix is constructed to obtain a multi-node basic data set; Step S22: Propagation consistency index modeling, which is used to measure the synchronization of factor fluctuation propagation between different nodes. Specifically, based on the multi-node basic data set, by introducing a directional fluctuation gradient and an exponential time decay parameter, the correlation coefficient model is improved to construct a propagation consistency index model; Step S23: Parameter gradient change identification. Specifically, based on the multi-node basic data set, by constructing a locally adaptive parameter gradient mutation index based on a sliding window, an abnormal change model of environmental factors is established to obtain environmental parameter gradient mutation index data; Step S24: Spatiotemporal interaction path modeling, which is used to dynamically construct the propagation path of a suspected pollution source from the starting point to the symptom node. Specifically, based on the propagation consistency index model and the environmental parameter gradient mutation index data, spatiotemporal interaction path modeling is carried out by sequentially constructing temporal propagation graph data and screening the environmental factor pollution propagation path, so as to obtain spatiotemporal interaction path reference data; Step S25: Potential health risk identification. Specifically, based on the spatiotemporal interaction path reference data, the health risk score of each spatiotemporal interaction path is calculated by using an improved health risk weighting function, and the spatiotemporal interaction path is screened according to the health risk score to obtain potential risk identification reference data; Step S26: Symptom alignment and linkage tracing. Specifically, by constructing a disease risk rule matching table, rule mapping modeling between symptom types and environmental factors is carried out, and through causal weight scoring and incentive feature screening, the incentive feature data of respiratory disease symptoms is obtained.

4. A method for family environment intervention for patients with chronic respiratory diseases according to claim 3, characterized in that: In step S2, the multi-node basic data set includes particulate matter detection data, gas detection data, temperature and humidity detection data, and dust mite allergen detection data.

5. A method for family environment intervention for patients with chronic respiratory diseases according to claim 4, characterized in that: In step S3, the sampling scheduling optimization is used to adjust the sensor sampling frequency. Specifically, based on the incentive feature data of respiratory disease symptoms and the environmental original data set, a lightweight improved long short-term memory network prediction method combined with parameter fluctuation characteristics is adopted to optimize the sampling scheduling of the sensing module, so as to obtain predictive sampling strategy reference data, including the following steps: Step S31: Sampling state classification data preprocessing. Specifically, based on the incentive feature data of respiratory disease symptoms and the environmental original data set, the sampling state category data is labeled to obtain sampling state classification data; Step S32: Parameter fluctuation partitioning. Specifically, by constructing a parameter fluctuation attention index, sampling attention partitioning is carried out to obtain parameter fluctuation partitioning data; Step S33: Lightweight prediction network improvement. Specifically, a lightweight long short-term memory neural network is adopted and combined with a custom parameter fluctuation partitioning gating function to improve the lightweight prediction network. The prediction network model is trained according to the sampling state classification data to obtain a lightweight prediction network model, and by using the lightweight prediction network model, based on the incentive feature data of respiratory disease symptoms and the environmental original data set, parameter distribution prediction is carried out to obtain sampling node prediction state data; Step S34: Sampling strategy generation. Specifically, based on the sampling node prediction state data, the sampling frequency is adjusted by constructing a scoring function to obtain node sampling frequency reference data; Step S35: Dynamic adjustment of sampling scheduling strategy. Specifically, the multi-modal micro sensing module is dynamically adjusted according to the node sampling frequency reference data to obtain sampling scheduling strategy reference data.

6. The family environment intervention method for patients with chronic respiratory diseases according to claim 5, characterized in that: In step S31, the sampling state classification data includes a stable state, a warning state, and a dangerous state; In step S32, the parameter fluctuation attention index is specifically calculated by weighting the parameter sharp jump score, the propagation distance normalization value, and the sensitivity score of environmental factors; The parameter sharp jump score is specifically calculated by combining the sliding window method with threshold setting.

7. A method for family environment intervention of patients with chronic respiratory diseases according to claim 6, characterized in that: In step S4, the home environment intervention is used to perform targeted device control and user behavior prompts. Specifically, based on the respiratory disease symptom inducement characteristic data and the predictive sampling strategy reference data, comprehensive home environment intervention is carried out to obtain personalized respiratory protection adjustment reference data.

8. A home environment intervention system for patients with chronic respiratory diseases, which is used to implement a home environment intervention method for patients with chronic respiratory diseases as described in any one of claims 1-7, characterized in that: It includes a sensing and acquisition module, a detection and traceability module, a sampling regulation module, and an environment intervention module.

9. The family environment intervention system for patients with chronic respiratory diseases according to claim 8, characterized in that: The sensing and acquisition module is used for constructing a sensing module. Through the construction of the sensing module, an original environmental data set is obtained, and the original environmental data set is sent to the detection and traceability module, the sampling regulation module, and the environment intervention module. The detection and traceability module is used for multi-node linked traceability. Through multi-node linked traceability, predictive sampling strategy reference data is obtained, and the predictive sampling strategy reference data is sent to the sampling regulation module. The sampling regulation module is used for optimizing sampling scheduling. Through optimizing sampling scheduling, sampling scheduling strategy reference data is obtained, and the sampling scheduling strategy reference data is sent to the environment intervention module. The environment intervention module is used for home environment intervention. Through home environment intervention, sampling scheduling strategy reference data is obtained.

Images