Big data-based infectious disease infection transmission analysis method and system

Abstract

The invention provides an infectious disease infection transmission analysis method and system based on big data, and relates to the field of medical data processing, and the method comprises the steps: determining a plurality of first key historical time points according to the historical definite diagnosis data of a current infectious disease in a plurality of medical institutions of a to-be-analyzed region, establishing a propagation undirected graph of the current infectious disease corresponding to the first key historical time point; according to the propagation undirected graph of the current infectious disease corresponding to each first key historical time point and historical definite diagnosis data of the various infectious diseases in multiple medical institutions of the to-be-analyzed area, determining similar infectious diseases from the various infectious diseases; according to the historical definite diagnosis data of the current infectious disease in the plurality of medical institutions of the to-be-analyzed area and the historical definite diagnosis data of the similar infectious disease in the plurality of medical institutions of the to-be-analyzed area, predicting future definite diagnosis data of the current infectious disease in the plurality of medical institutions of the to-be-analyzed area, the method has the advantage of improving the prediction accuracy of the infectious disease infection propagation development trend.

Description

Technical Field

[0001] The present invention relates to the field of medical data processing, and in particular to a method and system for analyzing the infection and spread of infectious diseases based on big data. Background Art

[0002] With the rapid development of information technology, big data technology has become a crucial tool for acquiring, storing, and processing massive amounts of data, providing more data sources and more powerful computing power for infectious disease research. Infectious diseases are a global public health challenge. Understanding the patterns and trends of their spread is crucial for implementing timely and effective preventive measures. Furthermore, with the increasing maturity of data science and technology, methods such as machine learning, data mining, and statistical analysis can be used to process and analyze large amounts of data, providing powerful tools and methods for analyzing infectious disease trends. Infectious diseases are caused by various pathogens and can be transmitted between humans, animals, or both. Infectious diseases are typically highly contagious, spread rapidly, and spread widely, severely impacting human health and well-being. Establishing an infectious disease prediction and alert system that can proactively detect the development and prevalence of infectious diseases is essential to enable the timely and effective development of prevention and control strategies and mitigate the harm caused by infectious diseases.

[0003] Currently, infection and spread analysis of infectious diseases is usually based on predictive models and existing sampling data to predict the development trend of a certain infectious disease. However, existing predictive models only predict recent sampling data and do not fully consider the correlation between the development trends of infectious diseases in multiple historical periods. Therefore, they have limitations and cannot accurately predict the development trends of infectious diseases.

[0004] Therefore, it is necessary to provide an infectious disease infection and spread analysis method and system based on big data to improve the accuracy of predicting the development trend of infectious disease infection and spread. Summary of the Invention

[0005] In order to solve the technical problem in the prior art that infectious disease infection and propagation analysis has limitations and cannot accurately predict the development trend of infectious diseases, the present invention provides an infectious disease infection and propagation analysis method based on big data, including: obtaining historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed, wherein the historical confirmed data includes the number of confirmed personnel at multiple historical time points; determining multiple first key historical time points based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; for each first key historical time point, establishing a propagation undirected graph of the current infectious disease corresponding to the first key historical time point based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; obtaining historical confirmed data of multiple infectious diseases in multiple medical institutions in the area to be analyzed; determining similar infectious diseases from multiple infectious diseases based on the propagation undirected graph of the current infectious disease corresponding to each first key historical time point and the historical confirmed data of multiple infectious diseases in multiple medical institutions in the area to be analyzed; and predicting future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed and the historical confirmed data of similar infectious diseases in multiple medical institutions in the area to be analyzed.

[0006] Furthermore, based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed, multiple first key historical time points are determined, including: sampling multiple historical time points from multiple historical time points; for each sampled historical time point, based on a preset time window and the historical confirmed data of the current infectious disease in each medical institution in the area to be analyzed, calculating the discrete value of the number of confirmed personnel in each medical institution at the historical time point, and calculating the global discrete value of the historical time point based on the discrete value of the number of confirmed personnel in each medical institution at the historical time point; and determining multiple first key historical time points based on the global discrete value of each historical time point.

[0007] Furthermore, based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed, an undirected propagation graph of the current infectious disease corresponding to the first key historical time point is established, including: intercepting fragmentary historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed corresponding to the first key historical time point from the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; calculating the propagation correlation coefficient between any two medical institutions based on the fragmentary historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed corresponding to the first key historical time point; and establishing an undirected propagation graph of the current infectious disease corresponding to the first key historical time point based on the propagation correlation coefficient between any two medical institutions.

[0008] Further, the similar infectious disease is determined from the plurality of infectious diseases according to the propagation undirected graph of the current infectious disease corresponding to each first key historical time point and the historical confirmed data of the plurality of medical institutions in the region to be analyzed, including: for each infectious disease, determining a plurality of second key historical time points based on the historical confirmed data of the plurality of medical institutions in the region to be analyzed, establishing a propagation undirected graph of the infectious disease corresponding to each second key historical time point, and extracting the propagation undirected graph features of the infectious disease; extracting the propagation undirected graph features of the current infectious disease according to the propagation undirected graph of the current infectious disease corresponding to each first key historical time point; and determining the similar infectious disease from the plurality of infectious diseases according to the propagation undirected graph features of the infectious disease and the propagation undirected graph features of the current infectious disease.

[0009] Further, the propagation undirected graph features of the current infectious disease are extracted according to the propagation undirected graph of the current infectious disease corresponding to each first key historical time point, including: determining the effective edges corresponding to the current infectious disease and the mean and dispersion of the propagation correlation coefficient of each effective edge according to the propagation undirected graph of the current infectious disease corresponding to each first key historical time point.

[0010] Further, the similar infectious disease is determined from the plurality of infectious diseases according to the propagation undirected graph of the current infectious disease corresponding to each first key historical time point and the historical confirmed data of the plurality of medical institutions in the region to be analyzed, including: for each infectious disease, determining a plurality of second key historical time points based on the historical confirmed data of the plurality of medical institutions in the region to be analyzed, establishing a propagation undirected graph of the infectious disease corresponding to each second key historical time point, and extracting the propagation undirected graph features of the infectious disease; extracting the propagation undirected graph features of the current infectious disease according to the propagation undirected graph of the current infectious disease corresponding to each first key historical time point; and determining the similar infectious disease from the plurality of infectious diseases according to the propagation undirected graph features of the infectious disease and the propagation undirected graph features of the current infectious disease.

[0011] Further, the future confirmed data of the current infectious disease in the plurality of medical institutions in the region to be analyzed is predicted according to the historical confirmed data of the similar infectious disease in the plurality of medical institutions in the region to be analyzed and the historical confirmed data of the current infectious disease in the plurality of medical institutions in the region to be analyzed, including: predicting the propagation undirected graph of the current infectious disease at a plurality of future time points in the region to be analyzed according to the propagation undirected graph of the similar infectious disease corresponding to each second key historical time point and the propagation undirected graph of the current infectious disease corresponding to each first key historical time point; and predicting the future confirmed data of the current infectious disease in the plurality of medical institutions in the region to be analyzed according to the historical confirmed data of the current infectious disease in the plurality of medical institutions in the region to be analyzed and the propagation undirected graph of the current infectious disease at the plurality of future time points in the region to be analyzed.

[0012] Furthermore, based on the propagation undirected graph of similar infectious diseases corresponding to each second key historical time point and the propagation undirected graph of the current infectious disease corresponding to each first key historical time point, the propagation undirected graph of the current infectious disease at multiple future time points in the area to be analyzed is predicted, including: a first graph encoder, used to generate a node embedding vector sequence corresponding to each second key historical time point of a similar infectious disease based on the propagation undirected graph of the similar infectious disease corresponding to each second key historical time point, and also used to generate a node embedding vector sequence corresponding to each first key historical time point of the current infectious disease based on the propagation undirected graph of the current infectious disease; a time encoder, used to generate time dynamic features of similar infectious diseases based on the node embedding vector sequence corresponding to each second key historical time point of similar infectious diseases, and also used to generate time dynamic features corresponding to the current infectious disease based on the node embedding vector sequence corresponding to each first key historical time point of the current infectious disease; a time decoder, used to predict the node status of the current infectious disease at multiple future time points in the area to be analyzed based on the time dynamic features of similar infectious diseases and the time dynamic features corresponding to the current infectious disease; a graph decoder, used to generate a propagation undirected graph of the current infectious disease at multiple future time points in the area to be analyzed based on the node status of the current infectious disease at multiple future time points in the area to be analyzed.

[0013] Furthermore, based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed and the undirected propagation graph of the current infectious disease at multiple future time points in the area to be analyzed, the future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed are predicted, including: a second graph encoder, used to generate a comprehensive node embedding vector sequence corresponding to the current infectious disease based on the undirected propagation graph of the current infectious disease at multiple future time points in the area to be analyzed and the undirected propagation graph of the current infectious disease corresponding to each first key historical time point; a time series encoder, used to generate propagation time dynamic features based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; a feature fusion unit, used to fuse the comprehensive node embedding vector sequence corresponding to the current infectious disease and the propagation time dynamic features to generate spatiotemporal features; an infection prediction unit, used to predict the future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed based on the spatiotemporal features.

[0014] The application provides a big data-based infectious disease infection transmission analysis system, comprising: a data acquisition module configured to acquire historical confirmed data of a current infectious disease in a plurality of medical institutions in an area to be analyzed, wherein the historical confirmed data comprises the number of confirmed persons at a plurality of historical time points; a time point screening module configured to determine a plurality of first key historical time points according to the historical confirmed data of the current infectious disease in the plurality of medical institutions in the area to be analyzed; and a transmission analysis module configured to, for each first key historical time point, establish a transmission undirected graph of the current infectious disease corresponding to the first key historical time point based on the historical confirmed data of the current infectious disease in the plurality of medical institutions in the area to be analyzed; the data acquisition module is further configured to acquire historical confirmed data of a plurality of infectious diseases in the plurality of medical institutions in the area to be analyzed; and the transmission analysis module is further configured to determine a similar infectious disease from the plurality of infectious diseases according to the transmission undirected graph of the current infectious disease corresponding to each first key historical time point and the historical confirmed data of the plurality of infectious diseases in the plurality of medical institutions in the area to be analyzed, and predict future confirmed data of the current infectious disease in the plurality of medical institutions in the area to be analyzed according to the historical confirmed data of the current infectious disease in the plurality of medical institutions in the area to be analyzed and the historical confirmed data of the similar infectious disease in the plurality of medical institutions in the area to be analyzed.

[0015] Compared with the prior art, the big data-based infectious disease infection transmission analysis method and system provided by the application has at least the following beneficial effects: The plurality of first key historical time points are dynamically determined based on the historical confirmed data, and the key stages (such as the outbreak period and the turning point of the stable period) of the infectious disease transmission are focused on. The indiscriminate analysis of all historical time points is avoided, the analysis efficiency is improved, and the key transmission dynamics are ensured not to be missed. Based on the historical confirmed data of the key historical time points, the transmission undirected graph is constructed, and the transmission relationship between the medical institutions is intuitively displayed. The transmission undirected graph provides a structured spatial relationship representation for subsequent analysis, which helps to deeply understand the transmission path and the transmission mode. By introducing the historical confirmed data of a plurality of infectious diseases, a similar infectious disease similar to the transmission mode of the current infectious disease is identified. The historical transmission rule of the similar infectious disease provides a reference for the prediction of the current infectious disease, and the prediction accuracy is improved. Based on the historical confirmed data of the current infectious disease and the similar infectious disease, the future confirmed data of a plurality of medical institutions is predicted. The prediction result provides a scientific basis for the formulation of prevention and control strategies, and helps to allocate medical resources and implement prevention and control measures in advance. The historical confirmed data of the plurality of medical institutions is fully utilized, and data-driven infectious disease transmission analysis is realized, which is not dependent on the characteristics of a specific infectious disease and can be applied to the transmission analysis of a plurality of infectious diseases. BRIEF DESCRIPTION OF DRAWINGS

[0016] This specification will be further described in the form of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting, and in these embodiments, like numbers represent like structures, wherein: Figure 1 is a flowchart of a method for analyzing the spread of infectious diseases based on big data according to some embodiments of this specification; Figure 2 is a schematic diagram of a propagation undirected graph according to some embodiments of this specification; Figure 3 is a schematic diagram of a portion of the structure of a propagation prediction model according to some embodiments of this specification; Figure 4 is a schematic diagram of a portion of the structure of a propagation prediction model according to some embodiments of this specification; Figure 5 This is a module diagram of an infectious disease infection and spread analysis system based on big data according to some embodiments of this specification. DETAILED DESCRIPTION

[0017] To more clearly illustrate the technical solutions of the embodiments of this specification, the following briefly describes the drawings required for describing the embodiments. Obviously, the drawings described below are merely examples or embodiments of this specification. Those skilled in the art can apply this specification to other similar scenarios based on these drawings without inventive effort. Unless otherwise apparent from the context or otherwise noted, the same reference numerals in the figures represent the same structure or operation.

[0018] Figure 1 This is a flow chart of a method for analyzing the spread of infectious diseases based on big data according to some embodiments of this specification, such as Figure 1 As shown, the infectious disease infection and spread analysis method based on big data may include the following process.

[0019] S110, obtaining historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed.

[0020] Among them, historical confirmed data includes the number of confirmed people at multiple historical time points.

[0021] Specifically, historical confirmed data for the current infectious disease from multiple medical institutions (such as hospitals, clinics, and community health service centers) in the area being analyzed can be obtained through any method. For example, historical confirmed data for the current infectious disease from multiple medical institutions in the area being analyzed can be obtained from the medical institution's electronic medical record system or infectious disease reporting system. Alternatively, the number of confirmed cases at multiple historical time points can be obtained from a third-party platform. Confirmed cases can be counted on a daily, weekly, or monthly basis.

[0022] For example, the historical confirmed data of the current infectious disease in two medical institutions in the area to be analyzed are shown in Table 1.

[0023]

[0024] S120 , determining a plurality of first key historical time points based on historical confirmed data of the current infectious disease in a plurality of medical institutions in the area to be analyzed.

[0025] Specifically include: Sampling multiple historical time points from multiple historical time points; For each sampled historical time point, based on the preset time window and the historical confirmed data of the current infectious disease in each medical institution in the area to be analyzed, the discrete value of the number of confirmed patients in each medical institution at the historical time point is calculated. Based on the discrete value of the number of confirmed patients in each medical institution at the historical time point, the global discrete value of the historical time point is calculated; A plurality of first key historical time points are determined according to the global discrete value of each historical time point.

[0026] Specifically, a subset of time points are randomly or regularly selected from all historical time points to calculate discrete values, reducing the computational effort while retaining representative time points. This selection can be done at fixed intervals (e.g., weekly) or based on a threshold number of confirmed cases (e.g., sampling when the number of confirmed cases exceeds 100 daily).

[0027] The preset time window can be the number of sampled historical time points used to calculate the discrete value (for example, the seven sampled historical time points before and after). For each sampled historical time point, the historical time points sampled within the preset time window can be determined with the sampled historical time point as the center. The variance or standard deviation of the number of confirmed patients in the medical institution at the sampled historical time point within the preset time window is calculated as the discrete value of the number of confirmed patients in the medical institution at the sampled historical time point. The average of the discrete values ​​of the number of confirmed patients in each medical institution at the sampled historical time point is then used as the global discrete value of the historical time point.

[0028] The historical time point of the sampling when the global discrete value is greater than the global discrete value threshold is taken as the first key historical time point, wherein the global discrete value threshold can be determined according to experimental data.

[0029] As can be understood, the above-mentioned S120 can more efficiently determine the first key historical time point through sampling, discrete value calculation, and threshold screening. The historical time point where the global discrete value is greater than the threshold is used as the first key historical time point. These time points generally correspond to key stages of the spread of infectious diseases (such as outbreaks and turning points in stable periods). The key historical time points screened by the global discrete value can more accurately reflect the spread of infectious diseases, reduce the amount of data required for subsequent identification of similar infectious diseases, and improve overall efficiency.

[0030] S130 , for each first key historical time point, based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed, establish an undirected propagation graph of the current infectious disease corresponding to the first key historical time point.

[0031] Specifically include: Extracting fragments of historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed corresponding to the first key historical time point from the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; Calculate the transmission correlation coefficient between any two medical institutions based on the fragmented historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed corresponding to the first key historical time point; Based on the transmission correlation coefficient of any two medical institutions, an undirected transmission graph of the current infectious disease corresponding to the first key historical time point is established.

[0032] Specifically, the fragmented historical confirmation data of the current infectious disease corresponding to the first key historical time point in multiple medical institutions in the area to be analyzed may include the number of confirmed personnel in multiple medical institutions at historical time points between the first key historical time point and the number of confirmed personnel in multiple medical institutions at the first key historical time point.

[0033] For any two medical institutions, the fragmented historical confirmed data of the current infectious disease in the two medical institutions in the area to be analyzed corresponding to the first key historical time point can be substituted into the calculation formula of the correlation coefficient (for example, the Pearson correlation coefficient, the Spearman rank correlation coefficient, etc.) to calculate the transmission correlation coefficient of the two medical institutions.

[0034] Figure 2 is a schematic diagram of a propagation undirected graph according to some embodiments of this specification, such as Figure 2 As shown, the propagation undirected graph may include nodes representing multiple medical institutions in the area to be analyzed, and edges connecting any two medical institutions, where the weight of the edge is the propagation correlation coefficient of the two medical institutions. For example, Figure 2In the data, the transmission correlation coefficient between medical institution A and medical institution B at a certain point in time is 0.6, the transmission correlation coefficient between medical institution B and medical institution C at the same time point is 0.7, and the transmission correlation coefficient between medical institution A and medical institution D at the same time point is 0.67, etc.

[0035] The undirected graph of the spread of the current infectious disease corresponding to the first key historical time point can be established according to the following process: Create an undirected graph G =( V , E ),in V For medical institutions, E is the edge set; For any two medical institutions A and B , calculate the propagation correlation coefficient r AB , if∣ r AB ∣ Exceeds the absolute value threshold (such as 0.5), then A and B Add edges between them, and the weight is the propagation correlation coefficient r AB ,wherein, the absolute value threshold can be determined based on manual or experimental data; Until all edges are established.

[0036] As you can understand, by calculating the transmission correlation coefficient between any two medical institutions, an undirected transmission graph is constructed, visually displaying the transmission relationship between medical institutions. By setting an absolute value threshold (such as 0.5), edges with low transmission correlation coefficients are filtered out, while edges with strong transmission relationships are retained. This reduces noise interference, highlights key transmission relationships, reduces the amount of data required to subsequently identify similar infectious diseases, and improves overall efficiency.

[0037] S140, obtains historical confirmed data of various infectious diseases (such as influenza, hand, foot and mouth disease, etc.) from multiple medical institutions in the area to be analyzed.

[0038] S150, determining similar infectious diseases from multiple infectious diseases based on the undirected propagation graph of the current infectious disease corresponding to each first key historical time point and the historical confirmed data of multiple infectious diseases in multiple medical institutions in the area to be analyzed.

[0039] Specifically include: For each infectious disease, based on the historical confirmed data of the infectious disease in multiple medical institutions in the area to be analyzed, multiple second key historical time points are determined, an undirected spread graph of the infectious disease corresponding to each second key historical time point is established, and the features of the undirected spread graph of the infectious disease are extracted; Extract the undirected graph features of the current infectious disease according to the undirected graph of the current infectious disease corresponding to each first key historical time point; According to the undirected graph characteristics of the spread of infectious diseases and the undirected graph characteristics of the spread of current infectious diseases, similar infectious diseases are determined from multiple infectious diseases.

[0040] Specifically, the method of determining multiple second key historical time points is similar to the method of determining multiple first key historical time points. The method of establishing an undirected propagation graph of an infectious disease corresponding to each second key historical time point is similar to the method of establishing an undirected propagation graph of the current infectious disease corresponding to the first key historical time point, which will not be repeated here.

[0041] In some embodiments, extracting features of the undirected graph of the current infectious disease according to the undirected graph of the current infectious disease corresponding to each first key historical time point includes: According to the undirected propagation graph of the current infectious disease corresponding to each first key historical time point, the valid edges corresponding to the current infectious disease and the mean and discrete values ​​of the propagation correlation coefficients of each valid edge are determined.

[0042] Specifically, for any two medical institutions, it is possible to determine whether an edge exists in the undirected propagation graph between the two medical institutions at each first key historical time point. The ratio of the first key historical time point at which the edge exists to the total number of first key historical time points is calculated. If the ratio is greater than a ratio threshold (e.g., 0.5), the edge connecting the two medical institutions is determined to be a valid edge. The ratio threshold can be determined based on manual or experimental data. The mean of the propagation correlation coefficient of the valid edge is calculated as the mean of the weight of the undirected propagation graph of the valid edge at each first key historical time point, and the variance or standard deviation of the weight of the undirected propagation graph of the valid edge at each first key historical time point is calculated as the discrete value of the propagation correlation coefficient of the valid edge.

[0043] In some embodiments, determining similar infectious diseases from multiple infectious diseases based on the undirected graph characteristics of the infectious disease and the undirected graph characteristics of the current infectious disease includes: For each infectious disease, the common effective edges are determined based on the effective edges corresponding to the current infectious disease and the effective edges corresponding to the infectious disease. The edge overlap corresponding to the infectious disease is determined based on the common effective edges. The difference in the propagation correlation coefficient of each common effective edge is determined based on the discrete value of the propagation correlation coefficient of each effective edge corresponding to the current infectious disease and the mean and discrete value of the propagation correlation coefficient of each effective edge corresponding to the infectious disease. The similarity between the infectious disease and the current infectious disease is calculated based on the edge overlap corresponding to the infectious disease and the difference in the propagation correlation coefficient of each common effective edge. Similar infectious diseases are identified from multiple infectious diseases based on the similarity between each infectious disease and the current infectious disease.

[0044] Specifically, the shared valid edge may be a valid edge that exists in both the current infectious disease and other infectious diseases.

[0045] Edge overlap is the ratio of the number of shared valid edges to the number of valid edges for the current infectious disease, reflecting the similarity between the two transmission networks. The higher the edge overlap, the more similar the transmission network structure between the current infectious disease and the other infectious disease is.

[0046] For each shared effective edge, the absolute value of the difference between the mean of the transmission correlation coefficient of the shared effective edge corresponding to the current infectious disease and the mean of the transmission correlation coefficient of the shared effective edge corresponding to the infectious disease can be calculated, and the absolute value of the difference between the discrete value of the transmission correlation coefficient of the shared effective edge corresponding to the current infectious disease and the discrete value of the transmission correlation coefficient of the infectious disease corresponding to the shared effective edge can be calculated. According to the absolute value of the mean difference and the absolute value of the discrete value difference, the difference value of the transmission correlation coefficient of the shared effective edge can be determined.

[0047] For example, the difference in propagation correlation coefficients of shared valid edges can be calculated using the following formula: in, is the difference value of the propagation correlation coefficient of the i-th common effective edge, is the absolute value of the mean difference of the i-th common effective edge, is the absolute value of the difference between the discrete values ​​of the i-th common valid edge, and is the preset weight, and is greater than 0, and ,For example, is 0.5, is 0.5.

[0048] As you can understand, by weighting the absolute value of the mean difference and the absolute value of the dispersion difference, the formula can comprehensively consider the mean and dispersion of transmission characteristics, providing a more comprehensive differential assessment. The preset weights can be adjusted according to specific needs to adapt to the differential assessment requirements of different infectious diseases or different analysis scenarios.

[0049] The similarity between an infectious disease and the current infectious disease can be calculated based on the edge overlap corresponding to the infectious disease and the difference in the propagation correlation coefficient of each common valid edge according to the following formula: Among them, S is the similarity between the infectious disease and the current infectious disease, P is the edge overlap corresponding to the infectious disease, and I is the total number of valid edges. and is the preset weight, and is greater than 0, and ,For example, is 0.5, is 0.5.

[0050] As you can understand, this formula is used to calculate the similarity between an infectious disease and the current infectious disease. Its core concept is to use a weighted average to comprehensively consider the combined effects of edge overlap and the difference in propagation correlation coefficients of shared valid edges. Edge overlap represents the ratio of edge overlap between the undirected propagation graph corresponding to the infectious disease and the undirected propagation graph of the current infectious disease. The higher the edge overlap, the greater the similarity in the transmission paths of the two infectious diseases. The sum of the differences in the transmission correlation coefficients of all shared valid edges represents the overall difference in the transmission characteristics of two infectious diseases. By combining the weighted average edge overlap and the difference in transmission correlation coefficients, the formula comprehensively considers the similarities in the transmission paths and characteristics of infectious diseases, providing a more comprehensive similarity assessment. The preset weights can be adjusted to meet specific needs to adapt to the similarity assessment requirements of different infectious diseases or different analysis scenarios.

[0051] Infectious diseases with the greatest similarity can be regarded as similar infectious diseases.

[0052] S160 , predicting future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed based on historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed and historical confirmed data of similar infectious diseases in multiple medical institutions in the area to be analyzed.

[0053] Specifically include: Predicting the undirected spread graph of the current infectious disease at multiple future time points in the area to be analyzed based on the undirected spread graph of similar infectious diseases corresponding to each second key historical time point and the undirected spread graph of the current infectious disease corresponding to each first key historical time point; Based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed and the undirected propagation graph of the current infectious disease in the area to be analyzed at multiple future time points, the future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed are predicted.

[0054] Specifically, the propagation prediction model can be used to predict the future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed and the historical confirmed data of similar infectious diseases in multiple medical institutions in the area to be analyzed.

[0055] As can be understood, this method predicts the undirected propagation graph of the current infectious disease at multiple future time points in the area to be analyzed based on the undirected propagation graph of similar infectious diseases corresponding to each second key historical time point and the undirected propagation graph of the current infectious disease corresponding to each first key historical time point. The undirected propagation graph can intuitively display the propagation relationship between medical institutions. By predicting the future undirected propagation graph, the potential propagation path and intensity of infectious diseases between medical institutions can be understood. By combining the historical data of similar infectious diseases with the current data of the current infectious disease, and using the undirected propagation graph to capture the propagation relationship, this prediction method can more comprehensively consider the propagation characteristics of infectious diseases, thereby improving the accuracy of the prediction.

[0056] Next, the team combines historical confirmed cases data from multiple medical institutions in the region with the predicted future undirected spread graph to further predict future confirmed cases for the current infectious disease. Historical confirmed cases provide information about the actual spread of the infectious disease among medical institutions, while the future undirected spread graph provides information about potential spread trends. By integrating these two pieces of information, we can more accurately predict future confirmed cases.

[0057] Figure 3 is a partial structural diagram of a propagation prediction model according to some embodiments of this specification, such as Figure 3 As shown, in some embodiments, based on the undirected propagation graph of a similar infectious disease corresponding to each second key historical time point and the undirected propagation graph of the current infectious disease corresponding to each first key historical time point, predicting the undirected propagation graph of the current infectious disease at multiple future time points in the area to be analyzed includes: A first graph encoder is configured to generate a sequence of node embedding vectors corresponding to each second key historical time point of a similar infectious disease based on an undirected propagation graph corresponding to each second key historical time point of a similar infectious disease, and further configured to generate a sequence of node embedding vectors corresponding to each first key historical time point of the current infectious disease based on an undirected propagation graph corresponding to each first key historical time point of the current infectious disease; A time encoder is used to generate temporal dynamic features of similar infectious diseases based on the node embedding vector sequence corresponding to each second key historical time point of similar infectious diseases, and is also used to generate temporal dynamic features corresponding to the current infectious disease based on the node embedding vector sequence corresponding to each first key historical time point of the current infectious disease; A temporal decoder is used to predict the node states of the current infectious disease at multiple future time points in the area to be analyzed based on the temporal dynamic characteristics of similar infectious diseases and the temporal dynamic characteristics corresponding to the current infectious disease; The graph decoder is used to generate an undirected propagation graph of the current infectious disease at multiple future time points in the area to be analyzed based on the node states of the current infectious disease at multiple future time points in the area to be analyzed.

[0058] Specifically, the first graph encoder can aggregate the information of the neighbor nodes through a multi-layer graph neural network, update the embedding vectors of the nodes, and directly affect the information contribution in the aggregation process based on the edge weights determined by the propagation correlation coefficients. The node embedding vector sequence corresponding to each second key historical time point of the similar infectious disease can represent the position and propagation relationship of each node in the propagation undirected graph. The way of generating the node embedding vector sequence corresponding to each first key historical time point of the current infectious disease is similar to the way of generating the node embedding vector sequence corresponding to each second key historical time point of the similar infectious disease, which will not be described here.

[0059] The time encoder captures the change rule of the node embedding vector sequence over time to generate global time features reflecting the dynamic of the infectious disease propagation. The global pooling (such as average pooling or maximum pooling) is performed on the node embedding vector sequence of each second key historical time point to generate a global feature vector at the time point level. The time encoder can capture the time dependency of the global feature vector sequence through LSTM (Long Short-Term Memory), and take the hidden state of the last time point as the time dynamic feature. The way of generating the time dynamic feature of the current infectious disease is similar to the way of generating the time dynamic feature of the similar infectious disease, which will not be described here.

[0060] Through the time decoder, the historical propagation rule (time dynamic feature) of the similar infectious disease and the current propagation state (time dynamic feature) of the current infectious disease are combined to predict the node state at the future time point. The time dynamic features of the similar infectious disease and the current infectious disease are spliced or weightedly fused to generate a joint time feature, and the encoder of the LSTM-Seq2Seq time decoder is used to process the joint time feature, and the decoder gradually generates the node state at the future time point, wherein the node state can include the propagation correlation coefficient of any two nodes. The attention mechanism is introduced in the decoder to dynamically allocate the contribution of the time dynamic features of the similar infectious disease and the current infectious disease to the prediction of the future node state.

[0061] The graph decoder determines the correlation coefficient matrix of the current infectious disease at multiple future time points according to the output of the time encoder, and any element of the correlation coefficient matrix represents the propagation correlation coefficient of any two medical institutions at the future time point. Based on the correlation coefficient matrix of the current infectious disease at multiple future time points, the graph neural network generates the propagation undirected graph of the current infectious disease at multiple future time points in the region to be analyzed.

[0062] Figure 4 is a partial structure schematic diagram of the propagation prediction model according to some embodiments of the present specification, as shown in Figure 4As shown, in some embodiments, based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed and the undirected propagation graph of the current infectious disease in the area to be analyzed at multiple future time points, predicting the future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed includes: A second graph encoder is configured to generate a sequence of comprehensive node embedding vectors corresponding to the current infectious disease based on an undirected propagation graph of the current infectious disease at multiple future time points in the area to be analyzed and an undirected propagation graph of the current infectious disease corresponding to each first key historical time point; A time series encoder is used to generate dynamic characteristics of the spread time based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; Feature fusion unit, used to fuse the comprehensive node embedding vector sequence corresponding to the current infectious disease and the dynamic characteristics of propagation time to generate spatiotemporal features; The infection prediction unit is used to predict the future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed based on temporal and spatial characteristics.

[0063] Specifically, the second graph encoder uses a graph neural network to encode the undirected propagation graph of the current infectious disease at multiple future time points in the analyzed area and the undirected propagation graph corresponding to each first key historical time point. This generates a sequence of node embedding vectors corresponding to each future time point and a sequence of node embedding vectors corresponding to each first key historical time point. These node embedding vector sequences for each future time point and the node embedding vector sequences corresponding to each first key historical time point are concatenated in chronological order to generate a comprehensive node embedding vector sequence corresponding to the current infectious disease, reflecting the propagation characteristics of the current infectious disease in spatial dimensions (e.g., across different medical institutions).

[0064] The time series encoder uses the Transformer model to capture the dependencies between different time points in the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed through the self-attention mechanism. Specifically, the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed are first converted into sequence embeddings through a fully connected layer, and position encoding is generated using sine and cosine functions. The position encoding is added to the sequence embedding, and the multi-head self-attention mechanism is used to capture diverse dependencies. The output of each time point is nonlinearly transformed through a feedforward neural network, and the output of the last time point is taken as the dynamic feature of the propagation time, reflecting the propagation trend and change law of the current infectious disease in the time dimension.

[0065] The feature fusion unit weights the comprehensive node embedding vector sequence and propagation time dynamic features corresponding to the current infectious disease through learnable parameters to generate spatiotemporal features, which integrate the characteristics of the spatial (medical institutions) and temporal (propagation trend) information of the current infectious disease spread.

[0066] The spread of infectious diseases is spatiotemporally coupled (for example, a confirmed case at one medical institution may affect related medical institutions, and the rate of spread varies over time). The infection prediction unit captures this dependency through spatiotemporal features, improving prediction accuracy. Using a spatiotemporal graph neural network, the infection prediction unit performs graph convolution and time series modeling based on spatiotemporal features to generate future confirmed cases of the current infectious disease at multiple medical institutions in the area being analyzed.

[0067] The loss function used to train the transmission prediction model can be related to the prediction accuracy of the transmission undirected graph and the prediction accuracy of future confirmed data. Specifically, the loss function can include the prediction loss of the transmission undirected graph and the prediction loss of future confirmed data. The prediction loss of the transmission undirected graph measures the difference between the predicted transmission undirected graph and the true transmission undirected graph and is calculated based on the mean squared error loss function. The prediction loss of future confirmed data measures the difference between the predicted future confirmed data and the true confirmed data and is calculated based on the mean absolute error loss function.

[0068] Figure 5 is a module diagram of an infectious disease infection and spread analysis system based on big data according to some embodiments of this specification, such as Figure 5 As shown, the infectious disease infection and propagation analysis system based on big data may include a data acquisition module, a time point screening module and a propagation analysis module.

[0069] A data acquisition module is used to obtain historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed, where the historical confirmed data includes the number of confirmed patients at multiple historical time points; A time point screening module is used to determine multiple first key historical time points based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; A propagation analysis module is used to establish, for each first key historical time point, an undirected propagation graph of the current infectious disease corresponding to the first key historical time point based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; The data acquisition module is also used to obtain historical confirmed data of various infectious diseases from multiple medical institutions in the area to be analyzed; The propagation analysis module is also used to determine similar infectious diseases from multiple infectious diseases based on the propagation undirected graph of the current infectious disease corresponding to each first key historical time point and the historical confirmed data of multiple infectious diseases in multiple medical institutions in the area to be analyzed, and to predict the future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed and the historical confirmed data of similar infectious diseases in multiple medical institutions in the area to be analyzed.

[0070] The infectious disease infection and spread analysis system based on big data can be used to execute the infectious disease infection and spread analysis method based on big data, which will not be described in detail here.

[0071] Finally, it should be understood that the embodiments described in this specification are intended only to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, by way of example and not limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly described and illustrated in this specification.

Claims

1. A method for analyzing the spread of infectious diseases based on big data, characterized in that: include: Obtain historical confirmed data of the current infectious disease from multiple medical institutions in the area to be analyzed, wherein the historical confirmed data includes the number of confirmed patients at multiple historical time points; Determine multiple first key historical time points based on historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; For each first key historical time point, an undirected spread graph of the current infectious disease corresponding to the first key historical time point is established based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; Obtain historical confirmed data on multiple infectious diseases from multiple medical institutions in the area to be analyzed; Identify similar infectious diseases from multiple infectious diseases based on the undirected spread graph of the current infectious disease corresponding to each first key historical time point and the historical confirmed data of multiple infectious diseases in multiple medical institutions in the area to be analyzed; Based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed and the historical confirmed data of similar infectious diseases in multiple medical institutions in the area to be analyzed, the future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed are predicted.

2. The infectious disease infection and spread analysis method based on big data according to claim 1, characterized in that: Based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed, multiple key historical time points are determined, including: Sampling multiple historical time points from multiple historical time points; For each sampled historical time point, based on the preset time window and the historical confirmed data of the current infectious disease in each medical institution in the area to be analyzed, the discrete value of the number of confirmed patients in each medical institution at the historical time point is calculated. Based on the discrete value of the number of confirmed patients in each medical institution at the historical time point, the global discrete value of the historical time point is calculated; A plurality of first key historical time points are determined according to the global discrete value of each historical time point.

3. The infectious disease infection and spread analysis method based on big data according to claim 1, characterized in that: Based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed, an undirected spread graph of the current infectious disease corresponding to the first key historical time point is established, including: Extracting fragments of historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed corresponding to the first key historical time point from the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; Calculate the transmission correlation coefficient between any two medical institutions based on the fragmented historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed corresponding to the first key historical time point; Based on the transmission correlation coefficient of any two medical institutions, an undirected transmission graph of the current infectious disease corresponding to the first key historical time point is established.

4. The infectious disease infection and spread analysis method based on big data according to any one of claims 1 to 3, characterized in that: Based on the undirected spread graph of the current infectious disease corresponding to each first key historical time point and the historical confirmed data of multiple infectious diseases in multiple medical institutions in the area to be analyzed, similar infectious diseases are identified from multiple infectious diseases, including: For each infectious disease, based on the historical confirmed data of the infectious disease in multiple medical institutions in the area to be analyzed, multiple second key historical time points are determined, an undirected spread graph of the infectious disease corresponding to each second key historical time point is established, and the features of the undirected spread graph of the infectious disease are extracted; Extract the undirected graph features of the current infectious disease according to the undirected graph of the current infectious disease corresponding to each first key historical time point; According to the undirected graph characteristics of the spread of infectious diseases and the undirected graph characteristics of the spread of current infectious diseases, similar infectious diseases are determined from multiple infectious diseases.

5. The infectious disease infection and spread analysis method based on big data according to claim 4, characterized in that: According to the undirected propagation graph of the current infectious disease corresponding to each first key historical time point, the undirected propagation graph features of the current infectious disease are extracted, including: According to the undirected propagation graph of the current infectious disease corresponding to each first key historical time point, the valid edges corresponding to the current infectious disease and the mean and discrete values ​​of the propagation correlation coefficients of each valid edge are determined.

6. The infectious disease infection and spread analysis method based on big data according to claim 5, characterized in that: Based on the undirected graph characteristics of the spread of infectious diseases and the undirected graph characteristics of the spread of current infectious diseases, similar infectious diseases are identified from multiple infectious diseases, including: For each infectious disease, the common effective edges are determined based on the effective edges corresponding to the current infectious disease and the effective edges corresponding to the infectious disease. The edge overlap corresponding to the infectious disease is determined based on the common effective edges. The difference in the propagation correlation coefficient of each common effective edge is determined based on the discrete value of the propagation correlation coefficient of each effective edge corresponding to the current infectious disease and the mean and discrete value of the propagation correlation coefficient of each effective edge corresponding to the infectious disease. The similarity between the infectious disease and the current infectious disease is calculated based on the edge overlap corresponding to the infectious disease and the difference in the propagation correlation coefficient of each common effective edge. Similar infectious diseases are identified from multiple infectious diseases based on the similarity between each infectious disease and the current infectious disease.

7. The infectious disease infection and spread analysis method based on big data according to claim 6, characterized in that: Based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed and the historical confirmed data of similar infectious diseases in multiple medical institutions in the area to be analyzed, the future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed is predicted, including: Predicting the undirected spread graph of the current infectious disease at multiple future time points in the area to be analyzed based on the undirected spread graph of similar infectious diseases corresponding to each second key historical time point and the undirected spread graph of the current infectious disease corresponding to each first key historical time point; Based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed and the undirected propagation graph of the current infectious disease in the area to be analyzed at multiple future time points, the future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed are predicted.

8. The infectious disease infection and spread analysis method based on big data according to claim 7, characterized in that: Based on the undirected propagation graph of similar infectious diseases corresponding to each second key historical time point and the undirected propagation graph of the current infectious disease corresponding to each first key historical time point, the undirected propagation graph of the current infectious disease at multiple future time points in the area to be analyzed is predicted, including: A first graph encoder is configured to generate a sequence of node embedding vectors corresponding to each second key historical time point of a similar infectious disease based on an undirected propagation graph corresponding to each second key historical time point of a similar infectious disease, and further configured to generate a sequence of node embedding vectors corresponding to each first key historical time point of the current infectious disease based on an undirected propagation graph corresponding to each first key historical time point of the current infectious disease; A time encoder is used to generate temporal dynamic features of similar infectious diseases based on the node embedding vector sequence corresponding to each second key historical time point of similar infectious diseases, and is also used to generate temporal dynamic features corresponding to the current infectious disease based on the node embedding vector sequence corresponding to each first key historical time point of the current infectious disease; A temporal decoder is used to predict the node states of the current infectious disease at multiple future time points in the area to be analyzed based on the temporal dynamic characteristics of similar infectious diseases and the temporal dynamic characteristics corresponding to the current infectious disease; The graph decoder is used to generate an undirected propagation graph of the current infectious disease at multiple future time points in the area to be analyzed based on the node states of the current infectious disease at multiple future time points in the area to be analyzed.

9. The infectious disease infection and spread analysis method based on big data according to claim 8, characterized in that: Based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed and the undirected propagation graph of the current infectious disease at multiple future time points in the area to be analyzed, the future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed is predicted, including: A second graph encoder is configured to generate a sequence of comprehensive node embedding vectors corresponding to the current infectious disease based on an undirected propagation graph of the current infectious disease at multiple future time points in the area to be analyzed and an undirected propagation graph of the current infectious disease corresponding to each first key historical time point; A time series encoder is used to generate dynamic characteristics of the spread time based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; Feature fusion unit, used to fuse the comprehensive node embedding vector sequence corresponding to the current infectious disease and the dynamic characteristics of propagation time to generate spatiotemporal features; The infection prediction unit is used to predict the future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed based on temporal and spatial characteristics.

10. The infectious disease infection and spread analysis system based on big data is characterized by: include: A data acquisition module is used to obtain historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed, wherein the historical confirmed data includes the number of confirmed patients at multiple historical time points; A time point screening module is used to determine multiple first key historical time points based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; A propagation analysis module is used to establish, for each first key historical time point, an undirected propagation graph of the current infectious disease corresponding to the first key historical time point based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed; The data acquisition module is also used to obtain historical confirmed data of multiple infectious diseases in multiple medical institutions in the area to be analyzed; The propagation analysis module is also used to determine similar infectious diseases from multiple infectious diseases based on the propagation undirected graph of the current infectious disease corresponding to each first key historical time point and the historical confirmed data of multiple infectious diseases in multiple medical institutions in the area to be analyzed, and to predict the future confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed based on the historical confirmed data of the current infectious disease in multiple medical institutions in the area to be analyzed and the historical confirmed data of similar infectious diseases in multiple medical institutions in the area to be analyzed.

Images