An Early Warning Method for Infectious Diseases Based on Multi-Source Data

Abstract

This invention belongs to the field of influenza early warning technology and discloses an early warning method for infectious diseases based on multi-source data. By acquiring and preprocessing multi-source data on influenza cases and search indices, and combining causal relationship and lag time identification, it eliminates spurious correlation interference, ensuring the reliability of the data foundation. It breaks through the dependence of traditional technologies on case reporting, shortens the early warning lag time, and compensates for the limitations of single early warning modes by using multi-point triggering and detection early warning algorithms to analyze and integrate the results. It avoids the inadequacy of single-dimensional early warning conclusions, not only improving the accuracy and timeliness of early warning, but also providing more comprehensive early warning references for epidemic prevention departments, helping to quickly initiate emergency response, effectively block the spread of influenza, reduce the consumption of prevention and control resources, improve the emergency response efficiency of influenza prevention and control, and adapt to the actual needs of accurate early warning of influenza in complex scenarios.

Description

Technical Field

[0001] This invention belongs to the field of influenza early warning technology, specifically a method for early warning of infectious diseases based on multi-source data. Background Technology

[0002] Influenza, as a highly contagious respiratory disease, is characterized by rapid spread, wide range, high control costs, and a tendency to trigger cluster infections. Early and accurate warning is a core element in blocking the spread of the epidemic, reducing the consumption of control resources, and improving the efficiency of emergency response. With the deep integration of big data technology and the public health field, early warning technologies based on multi-source data (especially real-time dynamic data such as search indices and public opinion data) should have broken through traditional limitations due to their timeliness advantage. However, most current technologies still have significant lags, have not completely broken free from dependence on clinical case reporting, and are unable to make up for the core shortcomings of traditional monitoring models. As a result, the early warning effectiveness is greatly reduced after the technology is implemented, and upgrades and optimizations are urgently needed to adapt to the pressing needs of influenza prevention and control.

[0003] Currently, most mainstream early warning models rely on a single or limited number of data sources for construction, and their core relies on statistical analysis, causal inference, and machine learning technologies to achieve risk identification. Among these, search indices can quickly reflect the public's attention trends to influenza symptoms and prevention knowledge, and can indirectly predict the early signs of epidemic spread, thus being widely used in early warning scenarios. Algorithms such as CCM causal networks, CUSUM anomaly detection, and entropy calculation provide core technical support for data association mining, anomaly fluctuation identification, and feature complexity characterization, respectively, forming diverse early warning technology implementation paths.

[0004] However, the shortcomings of existing early warning technologies, particularly their inherent lag, have become increasingly apparent. Coupled with other limitations, they are ill-suited to meet the demands for accurate early warnings in complex scenarios. On the one hand, the fusion of multi-source data lacks a unified dual verification standard for causal and lag relationships. Simply adding data dimensions can easily introduce spurious correlations, leading to insufficient data reliability and directly affecting the accuracy of early warning results, thus exacerbating the technological lag problem. On the other hand, early warning signals are mostly single-dimensional conclusions of "whether there is risk" or "whether there is risk," lacking a design that differentiates reliability gradients. This makes it impossible to provide epidemic prevention departments with tiered and precise decision-making references. Furthermore, the time lag between signal output and the occurrence of risk further restricts the efficiency of prevention and control response, making it difficult to meet the differentiated needs of different prevention and control stages. Summary of the Invention

[0005] This invention provides an early warning method for infectious diseases based on multi-source data, which solves the problems of insufficient data reliability and time lag between signal output and risk occurrence in existing early warning technologies.

[0006] To achieve the above objectives, the present invention provides the following technical solution: An early warning method for infectious diseases based on multi-source data includes: Acquire multi-source data, including influenza case data and at least one set of search index data; The multi-source data is preprocessed to form standardized time series data; Causal relationships and lag times were identified in standardized time series data, and search index data with a true causal relationship with influenza cases were screened out and their lag times were determined. Based on the filtered data and its lag time, two types of early warning results are generated by parallel analysis using a multi-point triggering early warning algorithm and a detection early warning algorithm. Integrate and output the results of the two types of early warnings.

[0007] Preferably, the causal relationship and lag time identification adopts a convergent cross-mapping algorithm. By setting the prediction step size interval, embedding dimension and number of neighborhood points, the time series data is transformed into a high-dimensional phase space point set for cross-prediction. The predicted skill value is calculated to determine the optimal lag time and causal direction. Furthermore, the stability of the skill value is verified by increasing the sample size, spurious correlation data is eliminated, true causal relationships are retained, and the causal strength and lag time of the effective group are output.

[0008] Preferably, the multi-point triggering early warning algorithm includes: Weights were constructed based on causal strength normalization, and case data and search index data were time-aligned. Within a sliding window, a bidirectional cross-correlation function is calculated for each group of aligned data, and a time irreversibility index is constructed based on the difference between the forward cross-correlation function and the backward cross-correlation function. The time irreversibility index sequence is dynamically accumulated and anomaly detected to construct a dynamic threshold and identify abnormal states that continuously exceed the threshold. Based on the number of search index groups that meet the anomaly detection conditions within a set time window, a warning level is generated and output.

[0009] Preferably, the time irreversibility index is calculated as follows: for each group of search indices, within a set lag time interval, the forward cross-correlation function between the case data and the lag search index, and the backward cross-correlation function between the search index and the lag case data are calculated. For each lag time, the difference between the forward cross-correlation function and the backward cross-correlation function is calculated and squared. The sum of the squared differences under all lag times is taken as the average value, and the square root of the average value is taken to obtain the time irreversibility index for that group.

[0010] Preferably, the detection and early warning algorithm includes the following steps: Weights are constructed based on causal strength normalization, and the standard deviation increment of case data is calculated within a sliding window; Calculate the absolute value of the correlation coefficient between the case data within the sliding window and each group of search indices, and normalize it to the local coupling coefficient; Calculate the causal weighted local network entropy based on causal weights and local coupling coefficients; The time irreversibility index generated by the multi-point triggering early warning algorithm is called, and the weighted time irreversibility index is calculated by combining the causal weight. A comprehensive index is constructed by integrating standard deviation increment, weighted time irreversibility index, causal weighted local network entropy, and number of search index data columns. The comprehensive index sequence is dynamically accumulated and anomaly detected to identify states that exceed the dynamic threshold and output early warning results.

[0011] Preferably, the method for constructing the comprehensive index is as follows:

[0012] in, The increment of the standard deviation of the case data. It is a weighted time-irreversible indicator. The number of columns for the search index data. The causal weighted local network entropy.

[0013] An early warning system for infectious diseases based on multi-source data includes: Data acquisition module: used to acquire multi-source data, including influenza case data and at least one set of search index data; Data processing module: used to preprocess the multi-source data to form standardized time series data; Causal lag identification module: used to identify causal relationships and lag times in standardized time series data, filter out search index data that have a real causal relationship with influenza cases, and determine their lag times; Early warning module: Based on the filtered data and its lag time, it performs parallel analysis using a multi-point triggering early warning algorithm and a detection early warning algorithm to generate two types of early warning results; Output module: Used to integrate and output the two types of early warning results.

[0014] A computer device includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement steps of a method for early warning of infectious diseases based on multi-source data.

[0015] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of an early warning method for infectious diseases based on multi-source data.

[0016] A computer program product includes a computer program that, when executed by a processor, implements steps of a method for early warning of infectious diseases based on multi-source data.

[0017] Compared with existing technologies, the present invention has the following beneficial effects: The present invention provides an early warning method for infectious diseases based on multi-source data. By acquiring and preprocessing multi-source data of influenza cases and search indices, and combining causal relationship and lag time identification, it eliminates spurious correlation interference, ensures the reliability of the data foundation, breaks through the dependence of traditional technologies on case reporting, shortens the early warning lag time, and compensates for the limitations of single early warning modes by using multi-point triggering and detection early warning algorithms to analyze and integrate the results. It avoids the inadequacy of single-dimensional early warning conclusions, not only improving the accuracy and timeliness of early warning, but also providing more comprehensive early warning references for epidemic prevention departments, helping to quickly initiate emergency response, effectively block the spread of influenza, reduce the consumption of prevention and control resources, improve the emergency response efficiency of influenza prevention and control, and adapt to the actual needs of accurate early warning of influenza in complex scenarios. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of an early warning method for infectious diseases based on multi-source data, according to an embodiment of the present invention. Figure 2 This is a flowchart of the multi-point triggering early warning algorithm according to an embodiment of the present invention; Figure 3 This is a flowchart of the detection and early warning algorithm according to an embodiment of the present invention; Figure 4 This is a block diagram of an early warning system for infectious diseases based on multi-source data, according to an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0021] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0022] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0023] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0024] The inventors discovered in their research that existing influenza early warning technologies are limited by algorithm design and data processing logic, resulting in numerous core shortcomings that make it difficult to meet the actual needs of prevention and control for accurate, timely, and reliable early warnings, as follows: 1. The early warning is delayed: Most traditional algorithms rely too much on case monitoring data, which needs to be reported and summarized at each level, resulting in a natural time lag; moreover, the algorithms are not good at capturing early hidden transmission signals, and often trigger early warnings only after the number of cases has increased significantly, thus missing the best window of opportunity for prevention and control.

[0025] 2. Skipping the causal relationship verification step: In order to simplify the calculation process, most algorithms directly build the association model based on data correlation, skipping the causal relationship identification and verification steps. This makes them susceptible to interference from spurious correlation data, resulting in distortion of core calculation parameters. They cannot accurately define the intrinsic relationship between search index and influenza spread, and the reliability of early warning is greatly reduced.

[0026] 3. Poor integration of multi-source data: Existing technologies only integrate multi-source data (such as case data and search indexes) at a superficial level, lacking targeted adaptation and integration mechanisms, and failing to fully explore the complementary value of different data sources.

[0027] 4. Poor algorithm interpretability: Some early warning algorithms rely on black-box models, only outputting early warning results without clearly explaining the core logic behind the early warning trigger. Prevention and control personnel find it difficult to understand the intrinsic connection between the data and the early warning results, which hinders the verification and judgment of early warning results and affects the trust and efficiency of prevention and control decisions.

[0028] like Figure 1 As shown, this embodiment of the invention provides an early warning method for infectious diseases based on multi-source data, including: S1: Acquire multi-source data, which includes influenza case data and at least one set of search index data; S2: Preprocess the multi-source data to form standardized time series data; S3: Identify causal relationships and lag times in standardized time series data, filter out search index data that have a true causal relationship with influenza cases, and determine their lag times; S4: Based on the filtered data and its lag time, the multi-point triggering early warning algorithm and the detection early warning algorithm are analyzed in parallel to generate two types of early warning results; S5: Integrate and output the results of the two types of early warnings.

[0029] By acquiring and preprocessing multi-source data on influenza cases and search indices, and combining causal relationship and lag time identification, false correlation interference is eliminated to ensure the reliability of the data foundation. This breaks through the reliance of traditional technologies on case reporting, shortens the early warning lag time, and compensates for the limitations of single early warning modes by using multi-point triggering and detection early warning algorithms to analyze and integrate the results. It also avoids the inadequacy of single-dimensional early warning conclusions, which not only improves the accuracy and timeliness of early warnings, but also provides more comprehensive early warning references for epidemic prevention departments, helps to quickly initiate emergency response, effectively block the spread of influenza, reduce the consumption of prevention and control resources, improve the emergency response efficiency of influenza prevention and control, and adapt to the actual needs of accurate early warning of influenza in complex scenarios.

[0030] S1: Acquire multi-source data, which includes influenza case data and at least one set of search index data; S2: Preprocess the multi-source data to form standardized time series data; (1) Data collection: Two types of core data are collected: one is the daily number of new influenza cases. The data comes from two sources: first, the influenza reporting system of disease control agencies, with data dimensions including date and number of newly confirmed cases, in a structured data table format, collected daily; and second, multi-source search index data. ( (The data is a positive integer and can be adjusted flexibly. It is sourced from search engine interfaces or third-party platforms, and is selected around keywords related to influenza symptoms, diagnosis, and prevention. The data dimensions include date, keywords, and index values. It is collected once a day and synchronized with the time dimension of case data.)

[0031] (2) Filtering and cleaning: Remove items with missing dates, empty values, and... Abnormal extreme value data identified by the criteria; missing daily data are supplemented using linear interpolation to ensure the continuity of the time series and guarantee the stability of subsequent calculations.

[0032] (3) Standardization: Calculate the historical data mean for each data series. with standard deviation Through the formula (original data) Obtain standardized data (t) is transformed into dimensionless data with a mean of 0 and a standard deviation of 1, thus eliminating dimensional differences.

[0033] (4) Window configuration: Set window length With step size , The shortest transmission cycle of influenza needs to be covered to ensure sufficient sample size. Set to update daily, and the window will be updated daily with the latest version. Data calculations enable dynamic real-time monitoring.

[0034] S3: Identify causal relationships and lag times in standardized time series data, filter out search index data that have a true causal relationship with influenza cases, and determine their lag times; 1. Data Input: Import standardized cases and time-series data of search indices for each group, ensuring dimensional alignment and temporal continuity; 2. CCM parameter configuration: Set the prediction step size interval, embedding dimension, and number of neighborhood points to adapt to the lag characteristics and data complexity of influenza. 3. Causal calculation: Transform the high-dimensional phase space point set, calculate the predicted skill value through cross-prediction, and determine the optimal lag time and causal direction; 4. Causal verification screening: Increasing the sample size to verify the stability of skill values, eliminating spurious correlations and retaining true causal relationships; Output results: Output the causal strength and lag time maxLag[j], j=1,...,n of the effective group, and simultaneously pass them into the subsequent two types of algorithm modules to provide unified basic parameters for the calculation of the time irreversibility index of Algorithm 1, the construction of the causal network and the calculation of the tRCE index of Algorithm 2.

[0035] S4: Based on the filtered data and its lag time, the multi-point triggering early warning algorithm and the detection early warning algorithm are analyzed in parallel to generate two types of early warning results; Multi-point triggering early warning algorithm: The core of this algorithm is to accurately capture early hidden anomalies through CCM causal relationship construction, time irreversibility index quantification, dynamic anomaly detection, and multi-point trigger aggregation. The algorithm flow is as follows: The first step is to call and construct causal parameters. This involves calling the causal strength g output by the causal lag identification module. j (Predicted skill value) and lag time Normalize the causal strength into weights w j This establishes a basic causal relationship system, providing reliable support for subsequent indicator calculations. (Weight) Calculation method: Weight of each group = absolute value of causal strength of that group / sum of absolute values ​​of causal strength of all groups.

[0036] The second step is data input and alignment. Input standardized case data. Effective search index data after causal filtering Combined with the obtained lag time Time alignment was performed on each set of data, using a lag. The search index for each day is correlated with current case data to ensure the timeliness of causal relationships.

[0037] The third step involves sliding window traversal and cross-correlation function calculation. Using the sliding window parameters of W=28 days and a step size of 1 day, the window range is updated daily; for each window... The alignment data within, Calculate the bidirectional cross-correlation function within an interval: Forward cross-correlation function (Case and Lag) The correlation coefficient of the search index and the back cross-correlation function. (Search index and lag) The correlation coefficient of daily cases fully characterizes the bidirectional association.

[0038] The fourth step is to calculate the time irreversibility index. For each... Value, calculate the difference between the forward and backward cross-correlation functions and square it, then apply this to all The time irreversibility index is obtained by summing the squared differences under the given values, taking the average, and then taking the square root of the average. The degree to which the quantitative system deviates from the equilibrium state is updated daily to form a continuous indicator sequence.

[0039] Step 5, Dynamic CUSUM Anomaly Detection. Input The indicator series and sliding window parameters are initialized with the following CUSUM cumulative statistic for each group: ;time Through formula Calculate the statistic, where , The mean and standard deviation of the current window indicator. The sensitivity parameters are trained based on historical data, capturing only continuous upward trends. Simultaneously, based on the median and MAD of the CUSUM statistic within the window, [the following is used / implied]... Constructing dynamic thresholds ( (for training calibration factor), compared to C and An alert is only confirmed and the corresponding group alert signal is triggered when the threshold is exceeded for two or more consecutive time periods.

[0040] Step 6: 7-day window for multi-point triggering and result output. Set a 7-day statistical time window, iterate through the time series daily, and count the number of search index groups that meet the warning conditions within the window. Each group is counted as only one valid trigger, regardless of how many times it is triggered within a single window. Map the warning level and color according to the number of triggers: 1 type corresponds to green (low risk), 2 types correspond to yellow (low to medium risk), 3 types correspond to orange (medium to high risk), and 4 or more types correspond to red (high risk).

[0041] Detection and early warning algorithm: The core of this algorithm is to construct the tRCE index and perform dynamic anomaly detection through CCM causal filtering, multi-feature fusion, and the execution process is as follows: The first step is causal parameter invocation and network construction. This involves invoking the effective causal strength and lag time output by the causal lag identification module. Focusing on the correlation between the daily number of new cases and the effective search indexes of each group, and based on verified real causal relationships, the causal strength is normalized into weights. This involves constructing a targeted causal network to eliminate spurious correlations and lay a reliable foundation for subsequent index construction. Weights Calculation method: Weight of each group = absolute value of causal strength of that group / sum of absolute values ​​of causal strength of all groups.

[0042] The second step is data input and sliding window traversal. Input standardized case data. Effective search index data Using the sliding window parameters, the window range is updated daily. Various statistical calculations are performed based on the data within the window, ensuring consistency with the calculation benchmark of Algorithm 1.

[0043] The third step is to calculate the statistics. Within the current sliding window, the standard deviation of the case sequence is calculated. ,pass Calculate the standard deviation increment to accurately capture the fluctuation trend of case data and quantify the dynamic change characteristics of case sequences.

[0044] The fourth step is local correlation characteristic analysis. Based on the constructed causal network, the absolute value of the correlation coefficient between the case data within the window and each group of search indices is calculated. and normalize it to local coupling coefficients. (Each group's coefficient = the correlation coefficient of that group / the sum of the correlation coefficients of all groups), representing the real-time correlation between cases and each search index.

[0045] Step 5: Calculate the causal weighted entropy value. This is done using the formula... Calculate the causal weighted local network entropy ,in For causal weights, The local coupling coefficient quantifies the structural uncertainty and disorder of associations within a causal network.

[0046] Step 6: Weighted calculation of time irreversibility index. This involves calling the time irreversibility index generated synchronously by Algorithm 1. Combined with the causal weights of this group Through formula Calculate weighted index By integrating bidirectional correlation characteristics, the ability of indicators to characterize system anomalies is optimized.

[0047] Step 7 Index Construction. Based on the aforementioned multi-dimensional features, the index is constructed using the formula... Build The index, where N is the number of columns in the search data and log is the natural logarithm, comprehensively integrates dynamic fluctuations, causal relationships, network entropy, and the irreversible nature of time to quantify the correlation fluctuations between cases and the search index. It is updated daily to form a continuous index sequence.

[0048] Step 8, Dynamic CUSUM Anomaly Detection. Input Exponential series, initial CUSUM cumulative statistic Time t is obtained through the formula

[0049] Calculate the statistic Current window (Exponential mean and standard deviation); construct a dynamic threshold based on the median and MAD of the CUSUM statistic within the window. ,when An alert is triggered when the threshold is exceeded.

[0050] Step 9: Output the results. Record the alert trigger status for each group of search indices and... Key characteristic values ​​of the index.

[0051] S5: Integrate and output the results of the two types of early warnings; The warning results output by the multi-point triggering warning algorithm and the detection warning algorithm are integrated and displayed in the visualization module.

[0052] Based on the aforementioned technical solution, this invention provides a formative solution to the core deficiencies of existing influenza early warning technologies through algorithm design, data processing, and mechanism innovation. Its core advantages are as follows: 1. Accurately Addressing the Pain Point of Delayed Early Warning and Enhancing Reliability Through Mutual Verification: Addressing the core shortcomings of traditional algorithms—over-reliance on delayed case data reported at each level and difficulty in capturing early, hidden transmission signals, leading to delayed warnings and missed early intervention windows—this invention employs two core algorithms to achieve precise early risk perception with significantly improved effectiveness. Algorithm 1 innovatively integrates the CCM causal network and time irreversibility indicators, deeply mining early, hidden influenza risk characteristics from multi-source data and incorporating a multi-point triggering approach for tiered early warning, thus eliminating reliance on delayed case data. Algorithm 2, based on CCM causal screening, constructs a tRCE index that integrates multi-dimensional features, achieving precise quantification of the correlation fluctuations between cases and the search index, ultimately realizing early risk identification. The results of both algorithms mutually verify each other, effectively eliminating the random errors and abnormal interference of single algorithms, resulting in a qualitative improvement in the accuracy and reliability of early warnings compared to traditional technologies, buying valuable time for prevention and control deployment and resource allocation.

[0053] 2. Complete the causal verification process and improve the reliability of early warning data: The independent causal lag identification module solves the problems of skipping causal verification and data correlation distortion in traditional technologies by eliminating false correlations and outputting standardized parameters. It filters out true causal relationships, eliminates false correlation data, and outputs standardized causal strength and lag time parameters. Both algorithms are based on this unified causal foundation, which significantly improves the credibility of early warning results.

[0054] 3. Optimize the multi-source data fusion logic and overcome the limitations of superficial integration: In response to the shortcomings of traditional technologies that fail to achieve good results in multi-source data fusion and only remain at the surface level, this invention constructs a multi-level fusion system. First, it achieves accurate screening of data associations through the causal module. Then, it integrates causal weights, dynamic statistics, network entropy, and other multi-dimensional features through the tRCE index of Algorithm 2 to deeply mine the complementary value of case data and search index, thus solving the problems of chaotic multi-source data integration and insufficient value mining.

[0055] 4. Enhance algorithm interpretability and adapt to prevention and control decision-making needs: In response to the shortcomings of traditional algorithms that rely on black box models and have poor interpretability, this invention retains the entire early warning logic traceability chain—from causal relationship identification and indicator calculation to early warning triggering, each link is supported by clear parameters and calculation rules; at the same time, it displays the early warning level, trigger group and core basis through visual output, so that prevention and control personnel can clearly understand the intrinsic relationship between data and early warning results, thereby improving their trust in early warning results and execution efficiency.

[0056] like Figure 4 As shown, this embodiment of the invention provides an early warning system for infectious diseases based on multi-source data, including: Data acquisition module: used to acquire multi-source data, including influenza case data and at least one set of search index data; Data processing module: used to preprocess the multi-source data to form standardized time series data; Causal lag identification module: used to identify causal relationships and lag times in standardized time series data, filter out search index data that have a real causal relationship with influenza cases, and determine their lag times; Early warning module: Based on the filtered data and its lag time, it performs parallel analysis using a multi-point triggering early warning algorithm and a detection early warning algorithm to generate two types of early warning results; Output module: Used to integrate and output the two types of early warning results.

[0057] A computer device is provided according to an embodiment of the present invention. This computer device includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps in the various method embodiments described above. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the various device embodiments described above.

[0058] The computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention.

[0059] The computer device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device may include, but is not limited to, a processor and memory.

[0060] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0061] The memory can be used to store the computer program and / or module, and the processor implements various functions of the computer device by running or executing the computer program and / or module stored in the memory, and by calling the data stored in the memory.

[0062] If the modules / units integrated into the computer device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory, random access memory, electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0063] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or basic characteristics. Therefore, the embodiments should be considered exemplary and non-limiting in all respects.

[0064] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can be appropriately combined to form other embodiments that can be understood by those skilled in the art. The above content is only for illustrating the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made based on the technical concept proposed in this invention shall fall within the scope of protection of the claims of this invention.

Claims

1. A method for early warning of infectious diseases based on multi-source data, characterized in that, include: Acquire multi-source data, including influenza case data and at least one set of search index data; The multi-source data is preprocessed to form standardized time series data; Causal relationships and lag times were identified in standardized time series data, and search index data with a true causal relationship with influenza cases were screened out and their lag times were determined. Based on the filtered data and its lag time, two types of early warning results are generated by parallel analysis using a multi-point triggering early warning algorithm and a detection early warning algorithm. Integrate and output the results of the two types of early warnings.

2. The method for early warning of infectious diseases based on multi-source data according to claim 1, characterized in that, The causal relationship and lag time identification adopts a convergent cross-mapping algorithm. By setting the prediction step size interval, embedding dimension and number of neighborhood points, the time series data is transformed into a high-dimensional phase space point set for cross-prediction. The predicted skill value is calculated to determine the optimal lag time and causal direction. Furthermore, the stability of the skill value is verified by increasing the sample size, spurious correlation data is eliminated, true causal relationships are retained, and the causal strength and lag time of the effective group are output.

3. The method for early warning of infectious diseases based on multi-source data according to claim 1, characterized in that, The multi-point triggering early warning algorithm includes: Weights were constructed based on causal strength normalization, and case data and search index data were time-aligned. Within a sliding window, a bidirectional cross-correlation function is calculated for each group of aligned data, and a time irreversibility index is constructed based on the difference between the forward cross-correlation function and the backward cross-correlation function. The time irreversibility index sequence is dynamically accumulated and anomaly detected to construct a dynamic threshold and identify abnormal states that continuously exceed the threshold. Based on the number of search index groups that meet the anomaly detection conditions within a set time window, a warning level is generated and output.

4. The method for early warning of infectious diseases based on multi-source data according to claim 3, characterized in that, The time irreversibility index is calculated as follows: For each set of search indices, within the set lag time interval, calculate the forward cross-correlation function between the case data and the lag search index, and the backward cross-correlation function between the search index and the lag case data. For each lag time, calculate the difference between the forward cross-correlation function and the backward cross-correlation function and square it. Sum the squared differences under all lag times and take the average value. Then take the square root of the average value to obtain the time irreversibility index for that set.

5. The method for early warning of infectious diseases based on multi-source data according to claim 1, characterized in that, The detection and early warning algorithm includes the following steps: Weights are constructed based on causal strength normalization, and the standard deviation increment of case data is calculated within a sliding window; Calculate the absolute value of the correlation coefficient between the case data within the sliding window and each group of search indices, and normalize it to the local coupling coefficient; Calculate the causal weighted local network entropy based on causal weights and local coupling coefficients; The time irreversibility index generated by the multi-point triggering early warning algorithm is called, and the weighted time irreversibility index is calculated by combining the causal weight. A comprehensive index is constructed by integrating standard deviation increment, weighted time irreversibility index, causal weighted local network entropy, and number of search index data columns. The comprehensive index sequence is dynamically accumulated and anomaly detected to identify states that exceed the dynamic threshold and output early warning results.

6. The method for early warning of infectious diseases based on multi-source data according to claim 5, characterized in that, The specific method for constructing the composite index is as follows: in, The increment of the standard deviation of the case data. It is a weighted time-irreversible indicator. The number of columns for the search index data. The causal weighted local network entropy.

7. An early warning system for infectious diseases based on multi-source data, characterized in that, include: Data acquisition module: used to acquire multi-source data, including influenza case data and at least one set of search index data; Data processing module: used to preprocess the multi-source data to form standardized time series data; Causal lag identification module: used to identify causal relationships and lag times in standardized time series data, filter out search index data that have a real causal relationship with influenza cases, and determine their lag times; Early warning module: Based on the filtered data and its lag time, it performs parallel analysis using a multi-point triggering early warning algorithm and a detection early warning algorithm to generate two types of early warning results; Output module: Used to integrate and output the two types of early warning results.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method for early warning of infectious diseases based on multi-source data as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method for early warning of infectious diseases based on multi-source data as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the method for early warning of infectious diseases based on multi-source data as described in any one of claims 1-6.

Images