Pharyngeal swab pneumonia aerosol transmission early warning method and system facing confined space

Abstract

The invention discloses a throat swab pneumonia aerosol propagation early warning method and system oriented to a closed space, relates to the field of propagation early warning, and realizes dynamic classification of aerosol propagation risks in the closed space through real-time data clustering analysis and a threshold self-adaptive adjustment mechanism. The problem of response lag of a traditional static evaluation model to a high-infectious variant strain is effectively solved; direct data comparison and model mapping dual-channel early warning are combined, so that the false alarm rate of a single detection method is remarkably reduced, and the reliability of the system in a complex environment is improved; through conjoint analysis of the minimum difference value data pair and the maximum difference interval, an influence factor change interval causing risk level transition is accurately identified, and a data support is provided for targeted prevention and control; on the basis of a threshold value dynamic correction mechanism fed back by the number of infected people, a closed loop of collection-analysis-early warning-optimization is formed, and adaptation of the model to different closed scenes (such as laboratories and livestock farms) is continuously improved.

Description

Technical Field

[0001] The present invention belongs to the field of transmission warning, and more specifically, relates to a throat swab pneumonia aerosol transmission warning method and system for confined spaces. Background Art

[0002] Traditional manual detection and static threshold warnings are difficult to capture in real time the sudden changes in infection risk caused by throat swab pneumonia aerosol transmission due to confined spaces; the fixed thresholds used in traditional solutions cannot adapt to the differences in different confined spaces; in addition, traditional methods rely on post-event disposal and cannot predict and map warnings through mutation intervals, resulting in delayed prevention and control, which is not conducive to reducing the increase in the number of infections. Summary of the Invention

[0003] In response to the problems in the related art, the present invention proposes a throat swab pneumonia aerosol transmission early warning method and system for confined spaces to overcome the above-mentioned technical problems existing in the existing related art.

[0004] To solve the above technical problems, the present invention is achieved through the following technical solutions:

[0005] The present invention is a throat swab pneumonia aerosol transmission early warning method for confined spaces, comprising the following steps:

[0006] S1. Collect risk influencing factor data of the confined space to be warned, and obtain a data matrix of risk influencing factors of the confined space to be warned;

[0007] S2. Collect and classify risk level data and risk influencing factor data corresponding to several existing confined spaces to obtain a classification matrix set of existing risk influencing factors;

[0008] S3, clustering the risk influencing factor data collected in S2; when the clustering result is different from the existing risk influencing factor classification matrix set, executing S4; otherwise, saving the risk influencing factor data collected in S2; repeating S2 and S3;

[0009] S4. Extract the minimum difference data pairs of the risk influencing factor data corresponding to different risk levels in the clustering results as the risk influencing factor data intervals of the mutation, and then screen the maximum risk influencing factor data intervals corresponding to different risk levels;

[0010] S5, by comparing the data matrix of the risk influencing factors of the space to be warned with the data interval of the risk influencing factors extracted in S4, issuing a warning based on the comparison results;

[0011] S6. Map the data matrix of risk influencing factors in the warning space and issue a warning based on the mapping results; after the warning is issued, adjust the risk level warning threshold and risk level mutation threshold based on the number of new infections in the confined space to be warned.

[0012] Preferably, the S1 comprises the following steps:

[0013] S11. Set several levels of risk caused by throat swab pneumonia aerosol transmission in confined spaces to obtain a confined space risk level set; set several types of influencing factors for the risk caused by throat swab pneumonia aerosol transmission in confined spaces to obtain a confined space risk influencing factor type set;

[0014] S12. Set a confined space to be warned; based on the confined space risk influencing factor type set, collect risk influencing factor data of the confined space to be warned at multiple time points to obtain a data matrix of risk influencing factors of the confined space to be warned;

[0015] Combining key parameters such as space airtightness and ventilation efficiency, a multi-dimensional data matrix is ​​constructed, which can identify extremely high-risk scenarios such as elevators and sampling cabins in real time; through the four-level risk threshold division (such as air change rate ≥ 6 times / hour is judged as low risk), it provides decision support for differentiated epidemic prevention measures (such as flow control and forced ventilation); quantifies the multiplier effect of behaviors such as talking loudly and not wearing masks on droplet emissions, and provides targeted prompts for strengthening personal protection nodes; effectively solves the shortcomings of "over-control" or "insufficient protection" in traditional prevention and control.

[0016] Preferably, said S2 comprises the following steps:

[0017] S21. Based on the confined space risk level set and the confined space risk influencing factor type set, collect risk level data and risk influencing factor data corresponding to a plurality of existing confined spaces to obtain an existing confined space risk level data set and an existing confined space risk influencing factor data matrix;

[0018] S22. Classify the data in the existing confined space risk influencing factor data matrix according to the existing confined space risk level data set to obtain an existing risk influencing factor classification matrix set;

[0019] Based on historical data, a mapping relationship between risk levels and influencing factors is constructed (for example, KTV box data is classified as high-risk), and the core risk feature combinations of various scenarios are automatically identified through the GMM clustering algorithm (for example, a ventilation rate of less than 3 times / hour + loud talking is a typical high-risk pattern); this breaks through the limitations of manual experience and judgment, and provides a quantifiable decision-making basis for precise epidemic prevention in special scenarios such as airport security areas and temporary hospitals.

[0020] Preferably, the step S3 includes the following steps:

[0021] S31. Using a GMM model to perform a clustering operation on the data matrix of risk influencing factors of existing confined spaces to obtain a clustering matrix set of existing risk influencing factors;

[0022] S32. When the existing risk influencing factor clustering matrix set is the same as the existing risk influencing factor classification matrix set, the existing confined space risk level data set and the existing confined space risk influencing factor data matrix are saved and S21 is repeated until the existing risk influencing factor clustering matrix set is different from the existing risk influencing factor classification matrix set; otherwise, the final existing confined space risk level data set and the final existing confined space risk influencing factor data matrix are obtained and the process proceeds to S41.

[0023] By utilizing the soft clustering characteristics of the Gaussian mixture model (GMM), transitional scenarios (such as the escalation of risk due to sudden changes in humidity in semi-open corridors) can be accurately captured, improving scenario adaptability by more than 20% compared to traditional hard classification models. In addition, compared with traditional algorithms such as K-means, the Gaussian mixture model is more suitable for transitional risk scenarios with overlapping features (such as the ambiguity of risk levels due to sudden changes in humidity in semi-open corridors). In addition, the soft classification characteristics of GMM can quantify the probability of data points belonging to different risk levels, thereby improving the robustness of risk assessment. Through the probabilistic clustering capability of the GMM model, the complex distribution characteristics of risk influencing factors in confined spaces can be accurately captured, achieving refined differentiation of risk levels.

[0024] Preferably, the S4 comprises the following steps:

[0025] S41. Setting a risk level mutation threshold; obtaining, based on the existing risk influencing factor clustering matrix set, pairs of risk influencing factor data with the smallest differences between risk influencing factor data sets with different corresponding risk level data having a difference greater than or equal to the risk level mutation threshold in each existing risk influencing factor clustering matrix, to obtain an initial level mutation influencing factor data matrix set and an initial mutation risk level data pair set;

[0026] S42. Based on the initial level mutation influencing factor data matrix set and the initial mutation risk level data pair set, taking the influencing factor data pair with the largest difference between the two initial level mutation influencing factor data sets in the initial level mutation influencing factor data matrix corresponding to the same initial mutation risk level data pair as the final influencing factor data pair to obtain a final level mutation influencing factor data matrix set; and then performing a deduplication operation on the initial mutation risk level data pair set to obtain a final mutation risk level data pair set;

[0027] By analyzing the minimum difference data pairs and maximum difference features between adjacent risk levels, the range of changes in the influencing factor data that causes risk level mutations can be accurately determined, narrowing the interval for the subsequent determination of the influencing factor data corresponding to the mutation critical point; among them, standardized preprocessing (Z-Score) reduces data noise interference, making the model easier to deploy in actual engineering scenarios, and the clustering results can be directly mapped to actionable risk level labels.

[0028] Preferably, the S5 comprises the following steps:

[0029] S51, constructing a final confined space level mapping model using the final existing confined space risk level dataset and the final existing confined space risk influencing factor data matrix;

[0030] S52. When each influencing factor data in the risk influencing factor data set in the spatial risk influencing factor data matrix to be warned is located between the influencing factor data intervals in the final level mutation influencing factor data matrix set, issuing a warning; otherwise, proceeding to S61;

[0031] By integrating historical risk data sets with a real-time influencing factor matrix, the model breaks through the limitations of traditional single-dimensional evaluation and supports the comprehensive mapping of multiple parameters such as spatial structure and population density, thereby increasing the risk assessment coverage to the global dimension. A nonlinear mapping algorithm is used to automatically match risk levels. When new influencing factors are input, the model can adjust the output results in real time to avoid the risk of misjudgment caused by the static threshold method.

[0032] Preferably, the final confined space level mapping model in S51 adopts an RBF neural network model.

[0033] Preferably, the S6 comprises the following steps:

[0034] S61, setting a risk level warning threshold; inputting each row of data in the data matrix of the risk influencing factors of the space to be warned into the final confined space level mapping model for mapping, to obtain a risk level data set to be warned; when the risk level data in the risk level data set to be warned is greater than or equal to the risk level warning threshold, issuing a warning;

[0035] S62. Set a threshold for the number of new infections; after the warning is issued, count the number of new infections caused by aerosol transmission of pharyngeal swab pneumonia; when the number of new infections is greater than or equal to the threshold for the number of new infections, adjust the risk level warning threshold and the risk level mutation threshold, and repeat S41, S42, S51, S52, S61 and S62 until the number of new infections is less than the threshold for the number of new infections, and obtain the final risk level warning threshold and the final risk level mutation threshold;

[0036] Through the confined space level mapping model, discrete influencing factors are converted into standardized risk levels, breaking through the limitations of traditional manual experience judgment and improving the objectivity and reproducibility of assessment results; the warning threshold is used as the quantitative boundary of risk escalation to avoid response delays caused by fuzzy rules, which is particularly suitable for high-risk scenarios such as aerosol transmission that require rapid intervention; the warning threshold and mutation threshold are dynamically revised based on the number of new infections, so that the model parameters are adaptively updated as the epidemic evolves, forming a closed-loop management of "monitoring-warning-verification-optimization"; by repeatedly executing the S41-S62 process, the risk threshold range is continuously converged, and the goal of stabilizing the number of infections within the safety threshold is ultimately achieved.

[0037] Preferably, adjusting the risk level warning threshold and the risk level mutation threshold in S62 includes the following steps:

[0038] S621. Set the value intervals of the risk level warning threshold and the risk level mutation threshold to obtain the value intervals of the risk level warning threshold and the risk level mutation threshold; construct a warning mutation threshold to adjust the salp population, set the maximum number of iterations of the warning mutation threshold to adjust the salp population to b′1 and the current number of iterations to b′2, which are respectively recorded as the maximum number of iterations of the threshold adjustment and the current number of iterations of the threshold adjustment; the number of dimensions of the search space of the warning mutation threshold to adjust the salp population is 2;

[0039] S622, generating a warning mutation threshold value according to the risk level warning threshold value interval and the risk level mutation threshold value interval to adjust the initial position of each salp in the salp population to obtain an initial position matrix;

[0040] S623, constructing the fitness function of the salp population based on the early warning mutation threshold;

[0041] S624, start iteration, and before iteration, set the threshold adjustment current iteration number to 1; in the first round of iteration, use the warning mutation threshold to adjust the fitness function of the salp population to calculate the fitness value of the initial position of each salp in the initial position matrix to obtain a first fitness value set; use the maximum fitness value in the first fitness value set and the initial position of the corresponding salp as the first global optimal fitness and the first global optimal position respectively; update the initial position of each salp in the initial position matrix according to the first global optimal fitness and the first global optimal position; after the update is completed, add 1 to the threshold adjustment current iteration number and enter the next round of iteration;

[0042] In each other round of iteration, the fitness function of the early warning mutation threshold is used to adjust the salp population, and the fitness value of the position of each salp in the salp population adjusted by the early warning mutation threshold obtained in the previous round of iteration is calculated to obtain a second fitness value set; the maximum fitness value in the second fitness value set and the position of the corresponding salp are respectively used as the second global optimal fitness and the second global optimal position; the position of each salp in the salp population adjusted by the early warning mutation threshold obtained in the previous round of iteration is updated according to the second global optimal fitness and the second global optimal position; after the update is completed, the current iteration number of the threshold adjustment is increased by 1 and the next round of iteration is entered;

[0043] S625. When b2′≥b1′, stop iteration and obtain the final global optimal fitness and the final global optimal position; otherwise, continue iteration until b2′≥b1′; when the reciprocal of the final global optimal fitness is less than the threshold of the number of new infections, use the two position components of the final global optimal position as the final risk level warning threshold and the final risk level mutation threshold respectively; otherwise, return to S524 and continue iteration until the reciprocal of the final global optimal fitness is less than the threshold of the number of new infections;

[0044] The risk level warning threshold and mutation threshold are adaptively optimized through the Salp Unica population algorithm, which can dynamically adjust the threshold parameters according to real-time data to avoid the problems of insufficient sensitivity or excessive false alarm rate caused by traditional static threshold settings; a two-dimensional search space and iterative update strategy are used, combined with the fitness function to guide the population to converge to the optimal solution, while ensuring computational efficiency and realizing the global optimal solution search of the threshold combination; setting the dual termination conditions of the maximum number of iterations and the threshold of the number of new infections can not only prevent the algorithm from falling into an infinite loop, but also ensure that the output threshold meets the actual prevention and control needs.

[0045] The throat swab pneumonia aerosol transmission warning system for confined spaces includes a confined space risk level setting module, a space risk influencing factor setting module, a data collection module for risk influencing factors to be warned, an existing confined space data collection and clustering calculation module, a classification result judgment module, a mutation influencing factor interval determination module, a confined space level mapping model construction module, a first warning module, a second warning module, and a level warning mutation threshold feedback adjustment module.

[0046] The present invention has the following beneficial effects:

[0047] 1. The present invention realizes the dynamic classification of aerosol transmission risks in confined spaces through real-time data clustering analysis and threshold adaptive adjustment mechanism, effectively overcoming the response lag problem of traditional static assessment models to highly contagious mutants; combining direct data comparison and model mapping dual-channel warning, it significantly reduces the false alarm rate of a single detection method and improves the reliability of the system in complex environments; through the joint analysis of minimum difference data pairs and maximum difference intervals, it accurately identifies the change intervals of influencing factors that lead to risk level transitions, providing data support for targeted prevention and control; based on the threshold dynamic correction mechanism feedback from the number of infected people, it forms a "collection-analysis-warning-optimization" closed loop, and continuously improves the model's adaptability to different confined scenarios (such as laboratories and livestock farms).

[0048] 2. In this invention, by integrating historical risk data sets with a real-time influencing factor matrix, the model breaks through the limitations of traditional single-dimensional evaluation and supports the comprehensive mapping of multiple parameters such as spatial structure and population density, thereby increasing the risk assessment coverage to the global dimension. A nonlinear mapping algorithm is used to automatically match risk levels. When new influencing factors are input, the model can adjust the output results in real time to avoid the risk of misjudgment caused by the static threshold method.

[0049] 3. The present invention adaptively optimizes the risk level warning threshold and mutation threshold, and can dynamically adjust the threshold parameters according to real-time data, avoiding the problems of insufficient sensitivity or excessive false alarm rate caused by traditional static threshold settings; adopts a two-dimensional search space and iterative update strategy, combined with the fitness function to guide the population to converge to the optimal solution, while ensuring computational efficiency and realizing the global optimal solution search of the threshold combination; sets the dual termination conditions of the maximum number of iterations and the threshold of the number of new infections, which not only prevents the algorithm from falling into an infinite loop, but also ensures that the output threshold meets the actual prevention and control needs.

[0050] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0052] Figure 1 This is a schematic diagram of the overall process of the throat swab pneumonia aerosol transmission early warning method for confined spaces of the present invention;

[0053] Figure 2 This is a flow chart of the present invention for classifying and clustering the collected risk influencing factor data;

[0054] Figure 3 A schematic diagram of the process of determining the data interval of risk influencing factors corresponding to a risk level mutation according to the present invention;

[0055] Figure 4 This is a flow chart of adjusting the risk level warning threshold and the risk level mutation threshold according to the present invention;

[0056] Figure 5 This is a module schematic diagram of the throat swab pneumonia aerosol transmission early warning system for confined spaces of the present invention. DETAILED DESCRIPTION

[0057] The following will clearly and completely describe the technical solutions in the embodiments of the invention in conjunction with the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of the invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0058] Example 1

[0059] See also Figure 1-4 This embodiment is a throat swab pneumonia aerosol transmission early warning method for confined spaces, comprising the following steps:

[0060] S1. Collect risk influencing factor data of the confined space to be warned, and obtain a data matrix of risk influencing factors of the confined space to be warned;

[0061] Said S1 comprises the following steps:

[0062] S11. Set several levels of risk caused by the transmission of pharyngeal swab pneumonia aerosols in confined spaces to obtain a confined space risk level set; the confined space risk level set includes an extremely high risk level (applicable to completely closed and small spaces without fresh air systems (such as elevators and nucleic acid sampling cabins), where people stay for more than 15 minutes and there are positive patients talking / coughing loudly), a high risk level (covering moderately confined places such as KTV boxes and conference rooms, where the mechanical ventilation air exchange rate is less than 3 times / hour, many people gather and do not wear masks properly, and there is a risk of continuous accumulation of viral aerosols), a medium risk level (including semi-open areas such as hospital corridors and waiting areas, which have basic ventilation but dense flow of people, and droplets and aerosols are mixed). The risk factors for the transmission of pneumonia in confined spaces due to pharyngeal swabs are set to determine the risk factor type set for confined spaces. The set includes the frequency of space ventilation, per capita space volume, space confinement, internal personnel behavior intensity (such as talking / singing loudly increases droplet emissions by 10 times), the rate of missing protective measures, the average coughing frequency of infected persons, space temperature, and space humidity.

[0063] S12. Set a confined space to be warned; based on the confined space risk influencing factor type set, collect risk influencing factor data of the confined space to be warned at multiple time points to obtain a data matrix of risk influencing factors of the confined space to be warned;

[0064] S2. Collect and classify risk level data and risk influencing factor data corresponding to several existing confined spaces to obtain a classification matrix set of existing risk influencing factors;

[0065] The S2 comprises the following steps:

[0066] S21. Based on the confined space risk level set and the confined space risk influencing factor type set, collect risk level data and risk influencing factor data corresponding to a plurality of existing confined spaces to obtain an existing confined space risk level data set and an existing confined space risk influencing factor data matrix;

[0067] S22. Classify the data in the existing confined space risk influencing factor data matrix according to the existing confined space risk level data set to obtain an existing risk influencing factor classification matrix set;

[0068] S3, clustering the risk influencing factor data collected in S2; when the clustering result is different from the existing risk influencing factor classification matrix set, executing S4; otherwise, saving the risk influencing factor data collected in S2; repeating S2 and S3;

[0069] The S3 includes the following steps:

[0070] S31, performing a clustering operation on the existing confined space risk influencing factor data matrix using a GMM model to obtain an existing risk influencing factor clustering matrix set;

[0071] The GMM model includes:

[0072] Probability distribution component: Each cluster center corresponds to a Gaussian distribution, and the data distribution characteristics are described by the mean vector (μ) and covariance matrix (Σ); the EM algorithm is used to iteratively optimize the parameters, including the E step (calculating the posterior probability) and the M step (updating the distribution parameters)

[0073] Input data processing: The existing classification matrix of confined space risk influencing factors (such as seven-dimensional features such as air change rate and occupant density) needs to be standardized and pre-processed (Z-Score processing), and then the optimal number of clusters (such as four categories corresponding to risk level sets) is automatically determined using the BIC criterion or silhouette coefficient.

[0074] Output layer: Outputs the probability of each data point belonging to each cluster center (soft classification) rather than hard classification; the resulting cluster center matrix can be directly mapped to risk levels (e.g., μ = [low air change rate, high occupancy density] → very high risk class);

[0075] S32. When the existing risk influencing factor clustering matrix set is the same as the existing risk influencing factor classification matrix set, the existing confined space risk level data set and the existing confined space risk influencing factor data matrix are saved and S21 is repeated until the existing risk influencing factor clustering matrix set is different from the existing risk influencing factor classification matrix set; otherwise, the final existing confined space risk level data set and the final existing confined space risk influencing factor data matrix are obtained and the process proceeds to S41.

[0076] S4. Extract the minimum difference data pairs of the risk influencing factor data corresponding to different risk levels in the clustering results as the risk influencing factor data intervals of the mutation, and then screen the maximum risk influencing factor data intervals corresponding to different risk levels;

[0077] The S4 comprises the following steps:

[0078] S41, set the risk level mutation threshold; according to the existing risk influencing factor clustering matrix set, obtain the risk influencing factor data pairs with the smallest difference between the risk level data sets with corresponding difference values ​​greater than or equal to the risk level mutation threshold in each existing risk influencing factor clustering matrix, and obtain the initial level mutation influencing factor data matrix set a1={a 11 ,...,a 1i ,...,a 1a′} and the initial mutation risk level data set a 1i represents the initial level mutation influencing factor data matrix of the combination of the two risk influencing factor data sets with the smallest Euclidean distance between the i-th risk influencing factor data sets with different corresponding risk level data in each existing risk influencing factor clustering matrix obtained, and a′ represents the total number of initial level mutation influencing factor data matrices obtained; Respectively represent a 1i The low and high values ​​of the risk level data corresponding to the two risk influencing factor data sets;

[0079] S42, according to the initial level mutation influencing factor data matrix set and the initial mutation risk level data pair set, taking the influencing factor data pair with the largest difference between the two initial level mutation influencing factor data sets in the initial level mutation influencing factor data matrix corresponding to the same initial mutation risk level data pair as the final influencing factor data pair, to obtain the final level mutation influencing factor data matrix set Then perform deduplication on the initial mutation risk level data set to obtain the final mutation risk level data set. as follows,

[0080]

[0081] in, Represents the obtained data matrix of the i-th final level mutation influencing factors, Indicates the total number of final grade mutation influencing factor data matrices obtained; Respectively The low and high values ​​of the risk level data corresponding to the two risk influencing factor data sets; Respectively The j-th type of risk influencing factor data corresponding to the medium, low, and high risk level data; b represents the total number of the set confined space risk influencing factor types;

[0082] S5, by comparing the data matrix of the risk influencing factors of the space to be warned with the data interval of the risk influencing factors extracted in S4, issuing a warning based on the comparison results;

[0083] The S5 comprises the following steps:

[0084] S51, constructing a final confined space level mapping model using the final existing confined space risk level dataset and the final existing confined space risk influencing factor data matrix;

[0085] The S51 includes the following steps:

[0086] S511. Construct an initial RBF neural network model and set a training data ratio; divide the final existing confined space risk level data set and the final existing confined space risk influencing factor data matrix according to the training data ratio to obtain a final existing confined space risk influencing factor training data matrix, a final existing confined space risk level training data set, a final existing confined space risk influencing factor test data matrix, and a final existing confined space risk level test data set;

[0087] S512. Setting a training error threshold; inputting the final existing confined space risk influencing factor training data matrix as training data and the final existing confined space risk level training data set as training labels into the initial RBF neural network model for training; during the training process, when the training error is less than the training error threshold, stopping the training to obtain a trained RBF neural network model; otherwise, continuing the training until the training error is less than the training error threshold;

[0088] S513, setting a test accuracy threshold; inputting the final existing confined space risk influencing factor test data matrix as test data and the final existing confined space risk level test data set as test labels into the trained RBF neural network model for testing; after the test is completed, obtaining test accuracy data; when the test accuracy data is greater than or equal to the test accuracy threshold, using the trained RBF neural network model as the final confined space level mapping model; otherwise, returning to S512 to continue training until the test accuracy data is greater than or equal to the test accuracy threshold;

[0089] The initial RBF neural network model includes:

[0090] Input layer: receives a multi-dimensional feature vector whose dimension is consistent with the number of types of factors affecting confined space risks;

[0091] Hidden layer (radial base layer): number of neurons: usually 10%-20% of the number of input samples;

[0092] Hidden layer activation function: Gaussian radial basis function (RBF) to achieve nonlinear mapping:

[0093] Fully connected layer: linearly weighted combination of hidden layer outputs, number of neurons = number of risk level categories;

[0094] Output layer activation function: Softmax outputs the probability distribution of risk level;

[0095] S52. When each influencing factor data in the risk influencing factor data set in the spatial risk influencing factor data matrix to be warned is located between the influencing factor data intervals in the final level mutation influencing factor data matrix set, issuing a warning; otherwise, proceeding to S61;

[0096] S6. Map the data matrix of risk influencing factors in the space to be warned and issue a warning based on the mapping results. After the warning is issued, adjust the risk level warning threshold and risk level mutation threshold based on the number of new infections in the confined space to be warned.

[0097] The S6 comprises the following steps:

[0098] S61, setting a risk level warning threshold; inputting each row of data in the data matrix of the risk influencing factors of the space to be warned into the final confined space level mapping model for mapping, to obtain a risk level data set to be warned; when the risk level data in the risk level data set to be warned is greater than or equal to the risk level warning threshold, issuing a warning;

[0099] S62. Set a threshold for the number of new infections; after the warning is issued, count the number of new infections caused by aerosol transmission of pharyngeal swab pneumonia; when the number of new infections is greater than or equal to the threshold for the number of new infections, adjust the risk level warning threshold and the risk level mutation threshold, and repeat S41, S42, S51, S52, S61 and S62 until the number of new infections is less than the threshold for the number of new infections, and obtain the final risk level warning threshold and the final risk level mutation threshold;

[0100] Adjusting the risk level warning threshold and the risk level mutation threshold in S62 includes the following steps:

[0101] S621: Set the risk level warning threshold and the risk level mutation threshold value interval to obtain the risk level warning threshold value interval. And the risk level mutation threshold value range Respectively represent the lower limit and upper limit of the risk level warning threshold; Respectively represent the lower limit and upper limit of the risk level mutation threshold; construct an early warning mutation threshold to adjust the salp population, set the maximum number of iterations of the early warning mutation threshold to adjust the salp population to b′1 and the current number of iterations to b′2, which are respectively recorded as the maximum number of iterations of the threshold adjustment and the current number of iterations of the threshold adjustment; the search space dimension of the early warning mutation threshold to adjust the salp population is 2;

[0102] S622, generating a warning mutation threshold value based on the risk level warning threshold value interval and the risk level mutation threshold value interval to adjust the initial position of each salp in the salp population to obtain an initial position matrix; the generation formulas are as follows:

[0103]

[0104] Where, They represent the position components of the initial position of the i-th salp in the salp population adjusted by the warning mutation threshold in the risk level warning threshold and the risk level mutation threshold dimensions; rand j1 、rand j2 Respectively for Generates a random number between 0 and 1; ceil represents the rounding function;

[0105] S623, constructing the fitness function c of the early warning mutation threshold to adjust the salp population; as follows,

[0106]

[0107] Where c′ represents the number of new infections caused by aerosol transmission of pneumonia from pharyngeal swabs;

[0108] S624, start iteration, and before iteration, set the threshold adjustment current iteration number to 1; in the first round of iteration, use the warning mutation threshold to adjust the fitness function of the salp population to calculate the fitness value of the initial position of each salp in the initial position matrix to obtain a first fitness value set; use the maximum fitness value in the first fitness value set and the initial position of the corresponding salp as the first global optimal fitness and the first global optimal position respectively; update the initial position of each salp in the initial position matrix according to the first global optimal fitness and the first global optimal position; after the update is completed, add 1 to the threshold adjustment current iteration number and enter the next round of iteration;

[0109] In each other round of iteration, the fitness function of the early warning mutation threshold is used to adjust the salp population, and the fitness value of the position of each salp in the salp population adjusted by the early warning mutation threshold obtained in the previous round of iteration is calculated to obtain a second fitness value set; the maximum fitness value in the second fitness value set and the position of the corresponding salp are respectively used as the second global optimal fitness and the second global optimal position; the position of each salp in the salp population adjusted by the early warning mutation threshold obtained in the previous round of iteration is updated according to the second global optimal fitness and the second global optimal position; after the update is completed, the current iteration number of the threshold adjustment is increased by 1 and the next round of iteration is entered;

[0110] S625. When b2′≥b1′, stop iteration and obtain the final global optimal fitness and the final global optimal position; otherwise, continue iteration until b2′≥b1′; when the reciprocal of the final global optimal fitness is less than the threshold of the number of new infections, the two position components of the final global optimal position are used as the final risk level warning threshold and the final risk level mutation threshold respectively; otherwise, return to S524 and continue iteration until the reciprocal of the final global optimal fitness is less than the threshold of the number of new infections.

[0111] Example 2

[0112] See also Figure 5 This embodiment discloses a throat swab pneumonia aerosol transmission early warning system for confined spaces. The system can implement the method of the above embodiment, including a confined space risk level setting module, a space risk influencing factor setting module, a risk influencing factor data collection module for warning, an existing confined space data collection and clustering calculation module, a classification result determination module, a mutation influencing factor interval determination module, a confined space level mapping model construction module, a first warning module, a second warning module, and a level warning mutation threshold feedback adjustment module;

[0113] The confined space risk level setting module sets several levels of risk caused by throat swab pneumonia aerosol transmission in the confined space to obtain a confined space risk level set;

[0114] The space risk influencing factor setting module sets several types of influencing factors of confined space risk to obtain a confined space risk influencing factor type set;

[0115] The risk influencing factor data collection module for the to-be-warned risk factors collects the risk influencing factor data of the to-be-warned confined space according to the confined space risk influencing factor type set to obtain a data matrix of the risk influencing factors of the to-be-warned space;

[0116] The existing confined space data collection and clustering calculation module collects risk level data and risk influencing factor data corresponding to a number of existing confined spaces and classifies them according to the confined space risk level set and the confined space risk influencing factor type set to obtain an existing risk influencing factor classification matrix set;

[0117] The classification result determination module clusters the risk influencing factor data collected in S2; when the clustering result is different from the existing risk influencing factor classification matrix set, execute S4; otherwise, save the risk influencing factor data collected in S2; repeat S2 and S3;

[0118] The mutation influencing factor interval determination module sets a risk level mutation threshold, extracts the minimum difference data pair of risk influencing factor data corresponding to different risk levels in the clustering results according to the risk level mutation threshold as the mutation risk influencing factor data interval, and then screens the maximum risk influencing factor data interval corresponding to different risk levels to obtain the final level mutation influencing factor data matrix set and the final mutation risk level data pair set;

[0119] The confined space level mapping model construction module uses the risk influencing factor data collected in S2 and stored in S3 to construct a final confined space level mapping model;

[0120] The first warning module compares the data matrix of the spatial risk influencing factors to be warned with the data matrix set of the final level mutation influencing factors, and issues a warning based on the comparison result;

[0121] The second warning module inputs the data matrix of the risk influencing factors of the space to be warned into the final confined space level mapping model for mapping, and issues a warning based on the mapping result;

[0122] After the warnings in S5 and S6 are issued, the level warning mutation threshold feedback adjustment module compares the number of new infections in the confined space to be warned with the preset threshold and adjusts the risk level warning threshold and the risk level mutation threshold according to the comparison results.

[0123] Example 3

[0124] This embodiment discloses a storage medium having a program stored thereon. When the program is executed by a processor, the method of the above embodiment is implemented.

[0125] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0126] The preferred embodiments of the invention disclosed above are intended only to help illustrate the invention. These preferred embodiments do not exhaust all details, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations are possible based on the content of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A throat swab pneumonia aerosol transmission early warning method for confined spaces, characterized by: The following steps are involved: S1. Collect risk influencing factor data of the confined space to be warned, and obtain a data matrix of risk influencing factors of the confined space to be warned; S2. Collect and classify risk level data and risk influencing factor data corresponding to several existing confined spaces to obtain a classification matrix set of existing risk influencing factors; S3, clustering the risk influencing factor data collected in S2; When the clustering result is different from the existing risk influencing factor classification matrix set, execute S4; otherwise, save the risk influencing factor data collected in S2; repeat S2 and S3; S4. Extract the minimum difference data pairs of the risk influencing factor data corresponding to different risk levels in the clustering results as the risk influencing factor data intervals of the mutation, and then screen the maximum risk influencing factor data intervals corresponding to different risk levels; S5, by comparing the data matrix of the risk influencing factors of the space to be warned with the data interval of the risk influencing factors extracted in S4, issuing a warning based on the comparison results; S6. Map the data matrix of risk influencing factors in the warning space and issue a warning based on the mapping results; after the warning is issued, adjust the risk level warning threshold and risk level mutation threshold based on the number of new infections in the confined space to be warned.

2. The method for early warning of pneumonia aerosol transmission by pharyngeal swabs in confined spaces according to claim 1, characterized in that: Said S1 comprises the following steps: S11. Set several levels of risk caused by throat swab pneumonia aerosol transmission in confined spaces to obtain a confined space risk level set; set several types of influencing factors for the risk caused by throat swab pneumonia aerosol transmission in confined spaces to obtain a confined space risk influencing factor type set; S12. Set a confined space to be warned; based on the confined space risk influencing factor type set, collect risk influencing factor data of the confined space to be warned at multiple time points to obtain a data matrix of risk influencing factors of the confined space to be warned.

3. The method for early warning of pneumonia aerosol transmission by pharyngeal swabs in confined spaces according to claim 2, characterized in that: The S2 comprises the following steps: S21. Based on the confined space risk level set and the confined space risk influencing factor type set, collect risk level data and risk influencing factor data corresponding to a plurality of existing confined spaces to obtain an existing confined space risk level data set and an existing confined space risk influencing factor data matrix; S22. Classify the data in the existing confined space risk influencing factor data matrix according to the existing confined space risk level data set to obtain an existing risk influencing factor classification matrix set.

4. The method for early warning of pneumonia aerosol transmission by pharyngeal swabs in confined spaces according to claim 3, characterized in that: The S3 includes the following steps: S31. Using a GMM model to perform a clustering operation on the data matrix of risk influencing factors of existing confined spaces to obtain a clustering matrix set of existing risk influencing factors; S32. When the existing risk influencing factor clustering matrix set is the same as the existing risk influencing factor classification matrix set, the existing confined space risk level data set and the existing confined space risk influencing factor data matrix are saved and S21 is repeated until the existing risk influencing factor clustering matrix set is different from the existing risk influencing factor classification matrix set; otherwise, the final existing confined space risk level data set and the final existing confined space risk influencing factor data matrix are obtained and enter S41.

5. The method for early warning of pneumonia aerosol transmission by pharyngeal swabs in confined spaces according to claim 4, characterized in that: The S4 comprises the following steps: S41. Setting a risk level mutation threshold; obtaining, based on the existing risk influencing factor clustering matrix set, pairs of risk influencing factor data with the smallest differences between risk influencing factor data sets with different corresponding risk level data having a difference greater than or equal to the risk level mutation threshold in each existing risk influencing factor clustering matrix, to obtain an initial level mutation influencing factor data matrix set and an initial mutation risk level data pair set; S42. According to the initial level mutation influencing factor data matrix set and the initial mutation risk level data pair set, the influencing factor data pair with the largest difference between the two initial level mutation influencing factor data sets in the initial level mutation influencing factor data matrix corresponding to the same initial mutation risk level data pair is taken as the final influencing factor data pair to obtain the final level mutation influencing factor data matrix set; and then the initial mutation risk level data pair set is deduplicated to obtain the final mutation risk level data pair set.

6. The method for early warning of pneumonia aerosol transmission by pharyngeal swabs in confined spaces according to claim 5, characterized in that: The S5 comprises the following steps: S51, constructing a final confined space level mapping model using the final existing confined space risk level dataset and the final existing confined space risk influencing factor data matrix; S52. When each influencing factor data in the risk influencing factor data set in the spatial risk influencing factor data matrix to be warned is located between the influencing factor data intervals in the final level mutation influencing factor data matrix set, an early warning is issued; otherwise, enter S61.

7. The throat swab pneumonia aerosol transmission early warning method for confined spaces according to claim 6, characterized in that: The final confined space level mapping model in S51 adopts the RBF neural network model.

8. The method for early warning of pneumonia aerosol transmission by pharyngeal swabs in confined spaces according to claim 7, characterized in that: The S6 comprises the following steps: S61, setting a risk level warning threshold; inputting each row of data in the data matrix of the risk influencing factors of the space to be warned into the final confined space level mapping model for mapping, to obtain a risk level data set to be warned; when the risk level data in the risk level data set to be warned is greater than or equal to the risk level warning threshold, issuing a warning; S62. Set a threshold for the number of new infections; after the warning is issued, count the number of new infections caused by aerosol transmission of pharyngeal swab pneumonia; when the number of new infections is greater than or equal to the threshold for the number of new infections, adjust the risk level warning threshold and the risk level mutation threshold, and repeat S41, S42, S51, S52, S61 and S62 until the number of new infections is less than the threshold for the number of new infections, and obtain the final risk level warning threshold and the final risk level mutation threshold.

9. A system for implementing the throat swab pneumonia aerosol transmission early warning method for confined spaces as described in any one of claims 1 to 8.

10. A storage medium, characterized in that: A program is stored thereon, and when the program is executed by the processor, the throat swab pneumonia aerosol transmission early warning method for confined spaces as described in any one of claims 1 to 8 is implemented.

Images