Infected case real-time monitoring system and method

Abstract

The invention discloses an infected case real-time monitoring system and method, and relates to the technical field of medical treatment, through analysis of a characteristic difference factor Tcyz, an asymptomatic infected patient and a normal healthy patient can be preliminarily distinguished, the analysis process is based on cross comparison of pathological data and behavior data, a potential asymptomatic infected patient can be identified, and the accuracy of diagnosis and treatment is improved. And preliminary estimation is provided for the infection risk of the patient, so that the patient is ensured to enter a to-be-audited mechanism before symptoms appear, and the risk of missed diagnosis is reduced. The natural language processing technology is used for analyzing infection classification of the patient to be verified, and the limitation of a traditional method in patient record processing is overcome. According to the NLP technology, a large amount of unstructured clinical text data such as medical records, medical advices and laboratory reports are automatically processed, relevant information can be rapidly extracted from the unstructured clinical text data, infection types of patients are preliminarily classified, and the infection possibility is judged based on a judgment coefficient Pdxs.

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and in particular to a real-time monitoring system and method for infection cases. Background Art

[0002] With the increase in the types of diseases, especially the continuous changes and evolution of infectious diseases and chronic diseases, the management of infection cases has become one of the important tasks of the global medical system. In the process of monitoring infection cases, real-time monitoring of changes in patients' conditions, especially monitoring of asymptomatic infections, has become a hot topic of research. Although asymptomatic infections do not show typical symptoms, such patients are still potential sources of infection. Timely monitoring and risk identification of asymptomatic patients are crucial to controlling the spread of the virus. Therefore, real-time monitoring and infection analysis of asymptomatic infected patients have become the key to ensuring the effectiveness of disease prevention and control.

[0003] The main consequences of the monitoring and risk assessment problems of asymptomatic infections are the spread of potential infections and the delay of infection control measures. Since these patients cannot be identified in the early stage, more extensive cross-infection may occur in the hospital, causing the spread of pathogens to other patients and medical staff. Especially in outbreaks of diseases such as influenza and pneumonia, asymptomatic infections that are not identified in time may become a key source of transmission. Such problems not only pose a threat to the health of patients, but also bring great pressure to the increased load of the entire medical system. In addition, delays in infection control measures will also lead to an increase in hospital infection rates, increase treatment costs and resource consumption. Therefore, in response to this situation, the development and application of more accurate real-time monitoring systems to timely identify asymptomatic infections and conduct early intervention is a key step to improve infection control efficiency and protect public health.

[0004] Currently, in the field of real-time monitoring of infection cases, although there are some monitoring and treatment methods for symptomatic patients, there are still certain limitations for the early identification and timely intervention of asymptomatic infections. The infection status of asymptomatic infections usually does not immediately manifest as clinical symptoms. Therefore, many infection cases are not discovered in time in the early stages. The main reason for this problem is that traditional monitoring systems rely on symptom manifestations, clinical test results and routine laboratory tests, while asymptomatic infections are often difficult to identify through routine symptom screening, which makes such patients easily ignored or misdiagnosed, and then miss the best time for treatment. Therefore, existing monitoring methods fail to fully consider the potential risks of asymptomatic infections, resulting in these patients continuing to spread pathogens without timely intervention. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a real-time monitoring system and method for infection cases, which solves the problems in the above-mentioned background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A real-time monitoring method for infection cases comprises the following steps: S1: Use the hospital information system to obtain relevant pathological data and behavioral data information of each patient in the hospital, and generate a personalized database after data preprocessing; S2: Based on the relevant behavioral data information in the personalized database, each patient is divided into asymptomatic and asymptomatic infected patients, and the characteristic difference analysis of the asymptomatic infected patients after division is performed based on the relevant pathological data information to obtain the characteristic difference factor Tcyz. Based on the value of the characteristic difference factor Tcyz, it is preliminarily estimated whether the corresponding asymptomatic infected patients have infection risks. If so, they enter the pending review mechanism and mark the patients for verification; S3: Monitor the patients to be verified and the patients with symptomatic infection in real time to obtain real-time patient data, and analyze the limitations of natural language processing technology in the infection classification of the patients to be verified based on the real-time patient data to obtain the judgment coefficient Pdxs. If the judgment coefficient Pdxs exceeds the preset threshold, the infection type of the patient to be verified is obtained; S4: Using deep learning technology, real-time patient data is input into the infection prediction model, and after linear normalization, the infection degree assessment index Gczs is fitted and output; S5: pre-set an infection threshold W, and compare the infection level assessment index Gczs with the infection threshold W to determine the infection status of the current patient, and take corresponding control measures based on the determination result.

[0007] Preferably, step S1 specifically includes: S11: Use a standardized interface to extract relevant pathological data information and relevant behavioral data information of each patient from the hospital information system, wherein the relevant pathological data information includes the medical records, laboratory test results and physiological indicators of each patient, and the physiological indicators include the blood oxygen saturation of each patient and cytokine levels ; The relevant behavioral data information includes whether the patient has fever, dry cough, shortness of breath, chest tightness, diarrhea and vomiting; S12: preprocessing the relevant pathological data information and relevant behavioral data information of each patient to remove duplicate values ​​and abnormal values ​​in the relevant pathological data information and relevant behavioral data information of each patient, and filling the identified missing values ​​by interpolation, and using dimensionless processing technology to unify the data format of the preprocessed data, the data format includes date format and numerical unit; The personalized database includes relevant pathological data information and relevant behavioral data information of each patient.

[0008] Preferably, step S2 specifically includes: S21: Based on the relevant behavior data information, each patient is divided into asymptomatic infection to obtain a number of symptomatic infection patients and asymptomatic infection patients; S22: Based on the relevant pathological data information, the asymptomatic infected patients are subjected to feature differentiation analysis to perform risk scoring on the asymptomatic infected patients, and after linear normalization processing, the feature difference factor Tcyz of each asymptomatic infected patient is obtained, which is specifically obtained in the following manner: ; In the formula, represents the characteristic difference factor of the i-th asymptomatic infected patient, represents the risk score of the i-th asymptomatic infected patient, represents the average risk score, represents the standard deviation of the risk score, and i is the number of asymptomatic infected patients.

[0009] Preferably, step S2 specifically includes: S221: The risk score of the i-th asymptomatic infected patient Obtained by the following formula: ; In the formula, represents the blood oxygen saturation of the i-th asymptomatic infected patient, represents the cytokine level of the i-th asymptomatic infected patient, and are weight values, Represents the first correction constant.

[0010] Preferably, step S2 further includes: S23: Compare the characteristic difference factor Tcyz of each asymptomatic infected patient with the standard deviation of three times the risk score Comparative analysis is performed to preliminarily determine whether the corresponding asymptomatic infected patients have infection risks. The specific determination contents are as follows: If the characteristic difference factor Tcyz of the corresponding asymptomatic infected patient exceeds three times the standard deviation of the risk score When the infection rate is high, the corresponding asymptomatic infected patients are preliminarily estimated to be at risk of infection; If the characteristic difference factor Tcyz of the corresponding asymptomatic infected patient does not exceed three times the standard deviation of the risk score At that time, it was preliminarily estimated that the corresponding asymptomatic infected patients had no infection risk; S24: When the corresponding asymptomatic infected patient is at risk of infection, the asymptomatic infected patient will be marked as a patient to be verified and enter the pending review mechanism.

[0011] Preferably, step S3 specifically includes: S31: Based on step S2, real-time monitoring of the patients to be verified and the patients with symptomatic infection is performed to obtain real-time patient data, wherein the real-time patient data includes the pathogen load of each patient. , cytokine concentration increase , SOFA score and serum albumin levels ; S32: Using natural language processing technology to clean and annotate text information involved in the personalized database and real-time patient data, cleaning includes removing stop words, and annotating includes word segmentation, syntactic analysis and keyword identification of text information; S33: Perform entity recognition on the personalized database and real-time patient data processed by S32 to extract the number of relevant infection entities successfully extracted from the text information involved in each patient to be verified , the number of paragraphs describing infection entities in the text information of each patient to be verified And the total number of paragraphs in the text information involved in each patient to be verified ,Entity recognition includes identifying the pathogen name and symptoms of each patient to be verified.

[0012] Preferably, step S3 further includes: S34: Based on steps S32 and S33, the limitations of natural language processing technology in the infection classification of the patient to be verified are analyzed, and after linear normalization processing, a judgment coefficient Pdxs is obtained. The judgment coefficient Pdxs is obtained by the following formula: ; In the formula, n represents the number of patients to be verified, =1, 2, 3, ..., n, represents the information adequacy score of the jth patient to be verified, represents the average information adequacy score, represents the text confidence score in the text information related to the jth patient to be verified, represents the average text confidence score, and All are weight values; Step S34 specifically includes: S341: Information adequacy score of the jth patient to be verified The specific method is as follows: ; In the formula, represents the number of related infection entities successfully extracted from the text information related to the jth patient to be verified, represents the minimum number of entities required to accurately classify the infection type; S342: Text confidence score in the text information related to the jth patient to be verified The specific method is as follows: ; In the formula, represents the number of paragraphs describing infection entities in the text information related to the jth patient to be verified, represents the total number of paragraphs in the text information involved in the jth patient to be verified; S35: By comparing the judgment coefficient Pdxs with the preset threshold, it is determined whether the infection type of the patient to be verified can be classified by the current natural language processing technology: If the judgment coefficient Pdxs exceeds the preset threshold, it is judged that the infection type of the patient to be verified can be classified by using the natural language processing technology to obtain the infection type of the patient to be verified; If the judgment coefficient Pdxs does not exceed the preset threshold, it is judged that the infection type of the patient to be verified cannot be classified using natural language processing technology at present, indicating that the patient to be verified is not infected at present; S36: At the same time, natural language processing technology is used to classify the infection types of symptomatic infected patients.

[0013] Preferably, step S4 specifically includes: S41: When the infection type of the patient to be verified is obtained, the infection prediction model is used to analyze the infection status of the patient to be verified and the symptomatic infected patient to obtain the infection degree assessment index Gczs, which is obtained specifically according to the following formula: In the formula, represents the pathogen load, Indicates that the cytokine concentration increases. represents the SOFA score, Indicates the serum albumin level, and are weight values, Represents the second correction constant.

[0014] Preferably, step S5 specifically includes: S51: By comparing the infection level assessment index Gczs with the infection threshold W, the infection status of the current patient is determined. The specific contents are as follows: If the infection level assessment index Gczs exceeds the infection threshold W, the current patient's infection is abnormal. At this time, the dosage of anti-infection drugs will be adjusted or replaced according to the patient's infection type, and the hospital's prevention and control measures will be strengthened, including personnel protection, environmental disinfection, and visitor restrictions. If the infection level assessment index Gczs does not exceed the infection threshold W, the patient's infection has not yet shown any abnormalities. At this time, the patient's physiological indicators will continue to be monitored, and medical staff will follow up on the patient regularly, paying attention to changes in symptoms and laboratory test results.

[0015] Real-time monitoring system for infected cases, including data acquisition module, potential risk analysis module, determination module and comprehensive feedback module; The data acquisition module is used to obtain relevant pathological data information and relevant behavioral data information of each patient in the hospital using the hospital information system, and generate a personalized database after data preprocessing; The potential risk analysis module is used to classify each patient into asymptomatic infection according to the relevant behavioral data information in the personalized database, and perform characteristic difference analysis on the asymptomatic infection patients after classification based on the relevant pathological data information to obtain the characteristic difference factor Tcyz. Based on the numerical value of the characteristic difference factor Tcyz, it is preliminarily estimated whether the corresponding asymptomatic infection patient has an infection risk. If so, it enters the pending review mechanism and marks the patient to be verified; The determination module is used to monitor the patients to be verified and the patients with symptomatic infection in real time to obtain real-time patient data, and analyze the limitations of natural language processing technology in the infection classification of the patients to be verified based on the real-time patient data to obtain the judgment coefficient Pdxs. If the judgment coefficient Pdxs exceeds a preset threshold, the infection type of the patient to be verified is obtained; The comprehensive feedback module is used to use deep learning technology to input real-time patient data into the infection prediction model, and after linear normalization processing, fit the output infection level assessment index Gczs; pre-set the infection threshold W, and compare the infection level assessment index Gczs with the infection threshold W to judge the current patient's infection status, and take corresponding control measures based on the judgment result.

[0016] The present invention provides a real-time monitoring system and method for infection cases, which has the following beneficial effects: By using the morbid and behavioral data obtained from the hospital information system and establishing a personalized database, the monitoring system can more accurately identify the specific health status and behavioral patterns of each patient. The comprehensiveness and accuracy of this data enhances the ability to identify asymptomatic infections and can capture potential infection risks in advance when there are no obvious clinical symptoms. Through the analysis of the characteristic difference factor Tcyz, it is possible to preliminarily distinguish asymptomatic infected patients from normal healthy patients. This analysis process is based on the cross-comparison of morbid data and behavioral data, which can identify potential asymptomatic infections and provide a preliminary estimate of their infection risk, ensuring that patients enter the pending review mechanism before symptoms appear, reducing the risk of missed diagnosis. In step S3, the infection classification of the verified patients is analyzed using natural language processing technology, which helps to overcome the limitations of traditional methods in processing patient records. NLP technology can quickly extract relevant information from a large amount of unstructured clinical text data, such as medical records, doctor's orders, laboratory reports, etc., by automatically processing a large amount of unstructured clinical text data, such as medical records, doctor's orders, laboratory reports, etc., and make a preliminary classification of the patient's infection type, and judge the possibility of infection based on the judgment coefficient Pdxs, thereby reducing errors and delays in manual operations. In step S4, the infection prediction model based on deep learning technology can accurately evaluate the patient's infection level assessment index Gczs through the input of real-time patient data. The deep learning model has the ability to automatically extract complex features from a large amount of data, so that the model can learn rules from dynamically changing conditions and obtain more accurate infection risk predictions. This indicator provides clinicians with a scientific basis based on data to help them make more accurate treatment decisions. Comparing the infection level assessment index Gczs with the preset infection threshold W can achieve real-time monitoring of the patient's infection status.

[0017] The system accurately distinguishes asymptomatic infected patients from general infected patients, providing a clear patient classification for subsequent analysis. Avoid missing potential infection cases: Asymptomatic infected people are often not discovered in time due to the lack of obvious symptoms. Early classification helps to include them in the monitoring scope and reduce the risk of missed diagnosis. Identify the potential risks of asymptomatic infection: Through differentiated analysis of patient characteristics, potential infection risk factors can be found in asymptomatic infected people, further avoiding ignoring minor pathological changes.

[0018] The quality of NLP classification is ensured by comprehensively considering the number of infection entities in the text information of the patient to be verified and the confidence of their expression. The scoring mechanism can help the system accurately evaluate the completeness and credibility of the text information, thereby reducing the risk of misclassification. Automated infection type classification judgment: The calculated judgment coefficient Pdxs can automatically evaluate whether NLP technology is sufficient to classify the infection type of the patient to be verified. Whether the infection type can be classified also depends on the patient's own infection level. If the coefficient is high, it means that the patient's infection information is sufficient and credible, and NLP classification can be accurately performed; otherwise, it means that the infection information of the patient to be verified is insufficient and cannot be effectively classified. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic diagram of the process of the real-time monitoring method of infection cases of the present invention; Figure 2 It is a structural block diagram of the real-time monitoring system for infection cases of the present invention. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] Example 1: Please refer to Figure 1 The present invention provides a method for real-time monitoring of infection cases, comprising the following steps: S1: Use the hospital information system (HIS) to obtain relevant pathological data and behavioral data information of each patient in the hospital, and generate a personalized database after data preprocessing; S2: Based on the relevant behavioral data information in the personalized database, each patient is divided into asymptomatic and asymptomatic infected patients, and the characteristic difference analysis of the asymptomatic infected patients after division is performed based on the relevant pathological data information to obtain the characteristic difference factor Tcyz. Based on the value of the characteristic difference factor Tcyz, it is preliminarily estimated whether the corresponding asymptomatic infected patients have infection risks. If so, they enter the pending review mechanism and mark the patients for verification; S3: Monitor the patients to be verified and the patients with symptomatic infection in real time to obtain real-time patient data, and analyze the limitations of natural language processing (NLP) technology in the infection classification of the patients to be verified based on the real-time patient data to obtain the judgment coefficient Pdxs. If the judgment coefficient Pdxs exceeds the preset threshold, the infection type of the patient to be verified is obtained; S4: Using deep learning technology, real-time patient data is input into the infection prediction model, and after linear normalization, the infection degree assessment index Gczs is fitted and output; S5: pre-set an infection threshold W, and compare the infection level assessment index Gczs with the infection threshold W to determine the infection status of the current patient, and take corresponding control measures based on the determination result.

[0022] In this embodiment, based on the establishment of a personalized database, the pathological data and behavioral information of each patient are accurately summarized and analyzed. By distinguishing patients according to symptoms and asymptomatic infections and analyzing their characteristic differences, potential asymptomatic infections can be effectively identified. The introduction of the characteristic difference factor Tcyz makes the risk assessment of asymptomatic infected patients more accurate, and can effectively warn asymptomatic infections in the early stage, thereby improving the accuracy of early screening. Real-time patient data monitoring and analysis enable the dynamic changes of patients to be verified and symptomatic infected patients to be grasped in a timely manner, and analyze the infection type of patients to be verified through natural language processing (NLP) technology, thereby improving the ability to identify various clinical information. The judgment coefficient Pdxs, as a key evaluation indicator, can make sensitive judgments between different pathological changes and infection types, ensuring that effective infection classification can still be performed when clinical data is incomplete or complex. The combination of deep learning technology and infection prediction model makes the assessment of the degree of infection of patients more accurate. By inputting real-time patient data into the model and performing linear normalization, the infection degree assessment index Gczs is generated to provide a basis for doctors to help decide on the patient's treatment plan. The patient's personalized data and historical records make the prediction more consistent with the actual condition, improve the accuracy and reliability of the prediction, and ensure that the patient's treatment plan can be updated and optimized in real time. By comparing the infection threshold W with the infection degree assessment index Gczs, the patient's infection status can be judged in real time, and abnormal cases can be responded to quickly. In the case of asymptomatic infection, this method can capture potential asymptomatic infections in a timely manner, avoid hospital cross-infection caused by missed diagnosis, and ensure that infection prevention and control in the hospital can be implemented. This method reduces the time and energy required for traditional manual monitoring and judgment through automated data analysis and monitoring, and significantly improves the efficiency of infection case management.

[0023] Embodiment 2: Specifically, step S1 specifically includes: S11: Use standardized interfaces (such as API, database query) to extract relevant pathological data information and relevant behavioral data information of each patient from the hospital information system (HIS), wherein the relevant pathological data information includes the medical records of each patient, laboratory test results (blood routine, imaging examination) and physiological indicators, and the physiological indicators include the blood oxygen saturation of each patient and cytokine levels ; The relevant behavioral data information includes whether the patient has fever, dry cough, shortness of breath, chest tightness, diarrhea and vomiting; S12: preprocessing the relevant pathological data information and relevant behavioral data information of each patient to remove duplicate values ​​and abnormal values ​​in the relevant pathological data information and relevant behavioral data information of each patient, and filling the identified missing values ​​by interpolation, and using dimensionless processing technology to unify the data format of the preprocessed data, the data format includes date format and numerical unit; The personalized database includes relevant pathological data information and relevant behavioral data information of each patient.

[0024] In this embodiment, by using standardized interfaces (such as API, database query), the relevant morbid data and behavioral data of patients can be fully extracted from the hospital information system (HIS), especially in terms of data types, including medical records, laboratory test results, blood oxygen saturation, cytokine levels, and clinical symptoms (such as fever, dry cough, shortness of breath, chest tightness, etc.), covering multiple dimensions that affect the health status of patients. This comprehensive data information can provide rich and accurate data support for subsequent infection risk analysis and prediction, and help to better identify potential infection cases, especially asymptomatic infections. Data preprocessing is an important step to ensure data quality and consistency. In step S12, the processing method of removing duplicate values ​​and outliers can significantly reduce data noise and ensure that the input data is a valid and analyzable sample. By filling missing values ​​with interpolation, the system can avoid deviations in analysis results caused by incomplete data. In addition, the use of dimensionless processing technology to standardize the data ensures that data from different units can be processed uniformly, making subsequent analysis and model training more accurate and reliable. This data preprocessing and standardization process reduces the need for manual intervention, improves data consistency, and further enhances the accuracy of data-driven decision-making. After steps S11 and S12, the formed personalized database contains detailed pathological and behavioral data of each patient. The data of each patient is processed into a standardized and uniform format, which can ensure that the system can make personalized analysis and predictions for each patient's real-time data. This personalized database provides the necessary data basis for the accurate monitoring and intervention of infection cases, and helps to identify patients with special pathological manifestations or risks, especially early asymptomatic infections. The personalized database enables the monitoring system to more flexibly adapt to the personalized needs of different patients, provide accurate infection risk assessment and intervention measures, and improve the overall effect of the system and the treatment effect of patients. The use of automated data extraction and preprocessing methods reduces the workload of manual data collation and analysis and improves work efficiency.

[0025] Example 3: Please refer to Figure 1 Specifically, step S2 specifically includes: S21: Based on the relevant behavior data information, each patient is divided into asymptomatic infection to obtain a number of symptomatic infection patients and asymptomatic infection patients; S22: Based on the relevant pathological data information, the asymptomatic infected patients are subjected to feature differentiation analysis to perform risk scoring on the asymptomatic infected patients, and after linear normalization processing, the feature difference factor Tcyz of each asymptomatic infected patient is obtained, which is specifically obtained in the following manner: ; In the formula, represents the characteristic difference factor of the i-th asymptomatic infected patient, represents the risk score of the i-th asymptomatic infected patient, represents the average risk score, represents the standard deviation of the risk score, and i is the number of asymptomatic infected patients.

[0026] Step S22 includes: S221: The risk score of the i-th asymptomatic infected patient Obtained by the following formula: ; In the formula, represents the blood oxygen saturation of the i-th asymptomatic infected patient, represents the cytokine level of the i-th asymptomatic infected patient, and are weight values, represents the first correction constant, where and The specific value is set by the user according to the situation.

[0027] Step S2 also includes: S23: Compare the characteristic difference factor Tcyz of each asymptomatic infected patient with the standard deviation of three times the risk score Comparative analysis is performed to preliminarily determine whether the corresponding asymptomatic infected patients have infection risks. The specific determination contents are as follows: If the characteristic difference factor Tcyz of the corresponding asymptomatic infected patient exceeds three times the standard deviation of the risk score When the infection rate is high, the corresponding asymptomatic infected patients are preliminarily estimated to be at risk of infection; If the characteristic difference factor Tcyz of the corresponding asymptomatic infected patient does not exceed three times the standard deviation of the risk score At that time, it was preliminarily estimated that the corresponding asymptomatic infected patients had no infection risk; S24: When the corresponding asymptomatic infected patient is at risk of infection, the asymptomatic infected patient will be marked as a patient to be verified and enter the pending review mechanism.

[0028] Among them, 3 standard deviations (99.73% coverage) means that the values ​​outside this range account for only about 0.27%, which is suitable for reducing the false alarm rate (false positive) and paying attention to relatively extreme anomalies as much as possible to reduce false alarms while still capturing significantly abnormal individuals.

[0029] In this embodiment, step S21 can ensure that the system accurately distinguishes symptomatic and asymptomatic infections by dividing patients into asymptomatic and asymptomatic infections based on relevant behavioral data information. Asymptomatic infections are often difficult to be discovered by traditional symptom recognition mechanisms. This step can effectively identify potential asymptomatic infected patients, thereby providing basic data for subsequent risk assessment. The addition of behavioral data (such as fever, cough and other symptoms) enables the system to consider the clinical manifestations of patients more comprehensively, rather than relying solely on a single physiological data, thereby improving the accuracy of the division. The risk score calculation in step S22, based on key physiological data such as blood oxygen saturation and cytokine levels, combined with the set weight value, can quantify the infection risk of each asymptomatic infected patient. This quantitative method can provide a clear reference for subsequent decision-making and make the risks between different patients comparable. The characteristic difference factor Tcyz obtained after linear normalization further ensures the standardization of the data, so that the risk scores and characteristic differences of different patients can be compared on the same scale, improving the feasibility and accuracy of cross-patient analysis. Step S23 can effectively screen out potential high-risk patients by comparing the characteristic difference factor Tcyz of each patient with three times the standard deviation of the risk score. If the characteristic difference factor Tcyz exceeds three times the standard deviation, it means that the patient's characteristics deviate significantly from the normal range, indicating that there may be a higher risk of infection. This standard deviation-based judgment method can better handle the noise in the sample, avoid misjudgment due to individual outliers, and ensure the stability and reliability of risk assessment. When step S23 identifies that an asymptomatic infected person has a high risk of infection, the patient will be marked as a patient to be verified and will be put into the pending review mechanism. This mechanism can promptly feedback high-risk patients to the clinical team, facilitate subsequent verification checks and further diagnosis, especially in the monitoring and management of asymptomatic infected persons. It is of great significance. The mechanism marked as a patient to be verified can increase attention to potential asymptomatic infected persons, avoid missed diagnosis or delayed diagnosis, especially during epidemic prevention and control. This early identification helps to effectively curb the spread of the epidemic.

[0030] Example 4: Please refer to Figure 1 Specifically, step S3 includes: S31: Based on step S2, real-time monitoring of the patients to be verified and the patients with symptomatic infection is performed to obtain real-time patient data, wherein the real-time patient data includes the pathogen load of each patient. , cytokine concentration increase , SOFA score and serum albumin levels ; S32: Use natural language processing (NLP) technology to clean and annotate text information involved in the personalized database and real-time patient data. Cleaning includes removing stop words, and annotation includes word segmentation, syntactic analysis and keyword identification of text information. Keywords include but are not limited to "asymptomatic infection", "viral load", "bacterial culture" and "fungal infection"; Among them, text information refers to structured and unstructured text information. Structured text information: This type of information is usually organized according to a certain format or rules, such as tables, database records, etc. For example, name, age, address and phone number are all structured text information. Unstructured text information: This type of information has no strict organizational structure and is usually presented in the form of free text. For example, articles, medical records, comments, emails and chat records.

[0031] S33: Perform entity recognition on the personalized database and real-time patient data processed by S32 to extract the number of relevant infection entities successfully extracted from the text information involved in each patient to be verified , the number of paragraphs describing infection entities in the text information of each patient to be verified And the total number of paragraphs in the text information involved in each patient to be verified ,Entity recognition includes identifying the pathogen name and symptoms of each patient to be verified.

[0032] In this embodiment, by collecting key physiological indicators and infection-related data in real time, the patient's infection status can be accurately monitored, especially the dynamic changes of asymptomatic and symptomatic patients. Timely discover the changing trend of infection and optimize the timing of treatment intervention. Multi-dimensional comprehensive evaluation: It integrates multi-dimensional monitoring data such as pathogen load increase and cytokine concentration increase to provide a comprehensive patient health assessment for the clinic and ensure a comprehensive analysis of the infection situation. Identify infection risks: Using real-time data to dynamically assess infection risks can help clinicians make timely and scientific decisions and provide targeted treatment for high-risk patients. Improve text data quality: By removing stop words, marking segmentation, syntactic parsing and other operations, the quality of input data is ensured, making text information more accurate, orderly and easy to analyze, which provides a high-quality data foundation for subsequent text analysis and feature extraction. Enhance data availability: Through the cleaning and annotation of structured and unstructured texts, effective information can be better extracted from unstructured texts such as medical records and diagnostic reports, improving the efficiency of data utilization. Entity recognition can automatically extract key infection-related information (such as pathogen names, symptoms, etc.) from texts such as patients' medical records and diagnostic reports, further reducing the workload of manual screening and extraction and improving work efficiency. Accurately capture infection information: By automatically identifying and annotating infection-related entities in the text, the patient's pathological information can be accurately captured to avoid missing important information, such as unmentioned viral loads, bacterial culture results, etc., to ensure the comprehensiveness of the data. Structured data output: By identifying infection entities and the number of paragraphs, the system can generate structured data output based on the patient's medical records and diagnostic reports, providing a clear and accurate basis for subsequent analysis and judgment. Support clinical decision-making: By identifying and extracting infection entities, it can help doctors more clearly understand the patient's infection type, pathogens, and related symptoms, thereby providing data support for the development of personalized treatment plans and intervention measures.

[0033] Example 5: Please refer to Figure 1 Specifically, step S3 further includes: S34: Based on steps S32 and S33, the limitations of natural language processing (NLP) technology in the infection classification of the patient to be verified are analyzed, and after linear normalization processing, a judgment coefficient Pdxs is obtained. The judgment coefficient Pdxs is obtained by the following formula: ; In the formula, n represents the number of patients to be verified, =1, 2, 3, ..., n, represents the information adequacy score of the jth patient to be verified, represents the average information adequacy score, represents the text confidence score in the text information related to the jth patient to be verified, represents the average text confidence score, and ,in, and The specific value is set by the user according to the situation; Step S34 specifically includes: S341: Information adequacy score of the jth patient to be verified The specific method is as follows: ; In the formula, represents the number of related infection entities successfully extracted from the text information related to the jth patient to be verified, represents the minimum number of entities required to accurately classify the infection type; Minimum number of entities required to accurately classify infection type It is the minimum number of key entities that the system can identify and requires, used to support the lower limit of infection type classification. The method of obtaining it is: preset based on medical experience or infection type classification standards.

[0034] S342: Text confidence score in the text information related to the jth patient to be verified The specific method is as follows: ; In the formula, represents the number of paragraphs describing infection entities in the text information related to the jth patient to be verified, represents the total number of paragraphs in the text information involved in the jth patient to be verified; S35: By comparing the judgment coefficient Pdxs with the preset threshold, it is determined whether the infection type of the patient to be verified can be classified by the current natural language processing (NLP) technology: If the judgment coefficient Pdxs exceeds the preset threshold, it is judged that the infection type of the patient to be verified can be classified by using natural language processing (NLP) technology to obtain the infection type of the patient to be verified; If the judgment coefficient Pdxs does not exceed the preset threshold, it is judged that the infection type of the patient to be verified cannot be classified using the natural language processing (NLP) technology, indicating that the patient to be verified is not infected at present; S36: At the same time, natural language processing (NLP) technology is used to classify the infection types of symptomatic infected patients.

[0035] Infection types include bacterial infection, viral infection and fungal infection; In this embodiment, the classification accuracy is improved: by comprehensively considering the adequacy and confidence of the text information, this step can evaluate whether the information of the patient to be verified is sufficient to support accurate infection classification. If the judgment coefficient Pdxs exceeds the threshold, the system can be confident that the infection type of the patient to be verified can be effectively classified by NLP technology. Strengthen the credibility of the text: by calculating the text confidence score, the system can judge the reliability of the text information, ensure that the information used is valuable, avoid excessive reliance on low-quality or irrelevant text, and thus ensure the accuracy of the classification decision. Step S35 can identify which patients to be verified cannot be accurately classified by existing NLP technology. This is especially important for asymptomatic infected persons in clinical practice, because asymptomatic infected persons often have no obvious clinical symptoms, and traditional methods may find it difficult to identify their infection type. Through the judgment of this step, the system can provide an early warning and require further verification.

[0036] Example 6: Please refer to Figure 1 Specifically, step S4 includes: S41: When the infection type of the patient to be verified is obtained, the infection prediction model is used to analyze the infection status of the patient to be verified and the symptomatic infected patient to obtain the infection degree assessment index Gczs, which is obtained specifically according to the following formula: In the formula, represents the pathogen load, Indicates that the cytokine concentration increases. represents the SOFA score, Indicates the serum albumin level, and are weight values, represents the second correction constant, where and The specific value is set by the user according to the situation.

[0037] The SOFA score is used to assess the severity of the patient's condition and prognosis, with a score range of 0 to 24 points. The higher the score, the more severe the organ dysfunction and the worse the patient's prognosis. The score consists of the following components: respiratory system (arterial oxygen partial pressure / oxygen concentration ratio, PaO2 / FiO2); coagulation system (platelet count); liver function (bilirubin level); cardiovascular system (blood pressure and use of vasoactive drugs); central nervous system (Glasgow Coma Scale, GCS); and renal function (serum creatinine or urine volume). Each system is scored from 0 to 4 points according to severity, and the total score of the six systems is the final SOFA score.

[0038] Step S5 specifically includes: S51: By comparing the infection level assessment index Gczs with the infection threshold W, the infection status of the current patient is determined. The specific contents are as follows: If the infection level assessment index Gczs exceeds the infection threshold W, the current patient's infection is abnormal. At this time, the dosage of anti-infection drugs will be adjusted or replaced according to the patient's infection type, and the hospital's prevention and control measures will be strengthened, including personnel protection, environmental disinfection, and visitor restrictions. If the infection level assessment index Gczs does not exceed the infection threshold W, the patient's infection has not yet shown any abnormalities. At this time, the patient's physiological indicators, such as body temperature, blood oxygen saturation, cytokine levels, and pathogen load, will continue to be monitored, and medical staff will follow up on the patient regularly, paying attention to changes in symptoms and laboratory test results.

[0039] In this embodiment, early detection of potential risks: by integrating the patient's physiological parameters and pathological data, patients with a high degree of infection, especially asymptomatic infections, can be identified at an early stage, further avoiding the spread or deterioration of the infection. Dynamic tracking of infection: based on real-time data to update the infection level assessment index Gczs, doctors can dynamically understand the patient's infection changes and provide a basis for subsequent treatment adjustments. Targeted treatment plan: The infection level assessment index Gczs can help doctors understand the patient's infection type and severity, so as to develop a more personalized and targeted treatment plan to achieve the effect of early intervention and timely treatment.

[0040] Example 7: Please refer to Figure 2 ,Specifically, the real-time monitoring system for infected cases includes a data acquisition module, a potential risk ,analysis module, a determination module and a comprehensive feedback module; The data acquisition module is used to obtain relevant pathological data information and relevant behavioral data information of each patient in the hospital using the hospital information system, and generate a personalized database after data preprocessing; The potential risk analysis module is used to classify each patient into asymptomatic infection according to the relevant behavioral data information in the personalized database, and perform characteristic difference analysis on the asymptomatic infection patients after classification based on the relevant pathological data information to obtain the characteristic difference factor Tcyz. Based on the numerical value of the characteristic difference factor Tcyz, it is preliminarily estimated whether the corresponding asymptomatic infection patient has an infection risk. If so, it enters the pending review mechanism and marks the patient to be verified; The determination module is used to monitor the patients to be verified and the patients with symptomatic infection in real time to obtain real-time patient data, and analyze the limitations of natural language processing technology in the infection classification of the patients to be verified based on the real-time patient data to obtain the judgment coefficient Pdxs. If the judgment coefficient Pdxs exceeds a preset threshold, the infection type of the patient to be verified is obtained; The comprehensive feedback module is used to use deep learning technology to input real-time patient data into the infection prediction model, and after linear normalization processing, fit the output infection level assessment index Gczs; pre-set the infection threshold W, and compare the infection level assessment index Gczs with the infection threshold W to judge the current patient's infection status, and take corresponding control measures based on the judgment result.

[0041] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for real-time monitoring of infection cases, characterized in that: The following steps are involved: S1: Use the hospital information system to obtain relevant pathological data and behavioral data information of each patient in the hospital, and generate a personalized database after data preprocessing; S2: Based on the relevant behavioral data information in the personalized database, each patient is divided into asymptomatic and asymptomatic infected patients, and the characteristic difference analysis of the asymptomatic infected patients after division is performed based on the relevant pathological data information to obtain the characteristic difference factor Tcyz. Based on the value of the characteristic difference factor Tcyz, it is preliminarily estimated whether the corresponding asymptomatic infected patients have infection risks. If so, they enter the pending review mechanism and mark the patients for verification; S3: Monitor the patients to be verified and the patients with symptomatic infection in real time to obtain real-time patient data, and analyze the limitations of natural language processing technology in the infection classification of the patients to be verified based on the real-time patient data to obtain the judgment coefficient Pdxs. If the judgment coefficient Pdxs exceeds the preset threshold, the infection type of the patient to be verified is obtained; S4: Using deep learning technology, real-time patient data is input into the infection prediction model, and after linear normalization, the infection degree assessment index Gczs is fitted and output; S5: pre-set an infection threshold W, and compare the infection level assessment index Gczs with the infection threshold W to determine the infection status of the current patient, and take corresponding control measures based on the determination result.

2. The method for real-time monitoring of infection cases according to claim 1, characterized in that: Step S1 specifically includes: S11: Extract relevant pathological data information and relevant behavioral data information of each patient from the hospital information system using a standardized interface, wherein the relevant pathological data information includes the medical records, laboratory test results and physiological indicators of each patient, and the physiological indicators include the blood oxygen saturation of each patient and cytokine levels ; The relevant behavioral data information includes whether the patient has fever, dry cough, shortness of breath, chest tightness, diarrhea and vomiting; S12: preprocessing the relevant pathological data information and relevant behavioral data information of each patient to remove duplicate values ​​and abnormal values ​​in the relevant pathological data information and relevant behavioral data information of each patient, and filling the identified missing values ​​by interpolation, and using dimensionless processing technology to unify the data format of the preprocessed data, the data format includes date format and numerical unit; The personalized database includes relevant pathological data information and relevant behavioral data information of each patient.

3. The method for real-time monitoring of infection cases according to claim 2, characterized in that: Step S2 specifically includes: S21: Based on the relevant behavior data information, each patient is divided into asymptomatic infection to obtain a number of symptomatic infection patients and asymptomatic infection patients; S22: Based on the relevant pathological data information, the asymptomatic infected patients are subjected to feature differentiation analysis to perform risk scoring on the asymptomatic infected patients, and after linear normalization processing, the feature difference factor Tcyz of each asymptomatic infected patient is obtained, which is specifically obtained in the following manner: ; In the formula, represents the characteristic difference factor of the i-th asymptomatic infected patient, represents the risk score of the i-th asymptomatic infected patient, represents the average risk score, represents the standard deviation of the risk score, and i is the number of asymptomatic infected patients.

4. The method for real-time monitoring of infection cases according to claim 3, characterized in that: Step S22 includes: S221: The risk score of the i-th asymptomatic infected patient Obtained by the following formula: ; In the formula, represents the blood oxygen saturation of the i-th asymptomatic infected patient, represents the cytokine level of the i-th asymptomatic infected patient, and are weight values, Represents the first correction constant.

5. The method for real-time monitoring of infection cases according to claim 3, characterized in that: Step S2 also includes: S23: Compare the characteristic difference factor Tcyz of each asymptomatic infected patient with the standard deviation of three times the risk score Comparative analysis is performed to preliminarily determine whether the corresponding asymptomatic infected patients have infection risks. The specific determination contents are as follows: If the characteristic difference factor Tcyz of the corresponding asymptomatic infected patient exceeds three times the standard deviation of the risk score When the infection rate is high, the corresponding asymptomatic infected patients are preliminarily estimated to be at risk of infection; If the characteristic difference factor Tcyz of the corresponding asymptomatic infected patient does not exceed three times the standard deviation of the risk score At that time, it was preliminarily estimated that the corresponding asymptomatic infected patients had no infection risk; S24: When the corresponding asymptomatic infected patient is at risk of infection, the asymptomatic infected patient will be marked as a patient to be verified and enter the pending review mechanism.

6. The method for real-time monitoring of infection cases according to claim 5, characterized in that: Step S3 specifically includes: S31: Based on step S2, real-time monitoring of the patients to be verified and the patients with symptomatic infection is performed to obtain real-time patient data, wherein the real-time patient data includes the pathogen load of each patient. , cytokine concentration increase , SOFA score and serum albumin levels ; S32: Using natural language processing technology to clean and annotate text information involved in the personalized database and real-time patient data, cleaning includes removing stop words, and annotating includes word segmentation, syntactic analysis and keyword identification of text information; S33: Perform entity recognition on the personalized database and real-time patient data processed by S32 to extract the number of relevant infection entities successfully extracted from the text information involved in each patient to be verified , the number of paragraphs describing infection entities in the text information of each patient to be verified And the total number of paragraphs in the text information involved in each patient to be verified ,Entity recognition includes identifying the pathogen name and symptoms of each patient to be verified.

7. The method for real-time monitoring of infection cases according to claim 6, characterized in that: Step S3 also includes: S34: Based on steps S32 and S33, the limitations of natural language processing technology in the infection classification of the patient to be verified are analyzed, and after linear normalization processing, a judgment coefficient Pdxs is obtained. The judgment coefficient Pdxs is obtained by the following formula: ; In the formula, n represents the number of patients to be verified, =1, 2, 3, ..., n, represents the information adequacy score of the jth patient to be verified, represents the average information adequacy score, represents the text confidence score in the text information related to the jth patient to be verified, represents the average text confidence score, and All are weight values; Step S34 specifically includes: S341: Information adequacy score of the jth patient to be verified The specific method is as follows: ; In the formula, represents the number of related infection entities successfully extracted from the text information related to the jth patient to be verified, represents the minimum number of entities required to accurately classify the infection type; S342: Text confidence score in the text information related to the jth patient to be verified The specific method is as follows: ; In the formula, represents the number of paragraphs describing infection entities in the text information related to the jth patient to be verified, represents the total number of paragraphs in the text information involved in the jth patient to be verified; S35: By comparing the judgment coefficient Pdxs with the preset threshold, it is determined whether the infection type of the patient to be verified can be classified by the current natural language processing technology: If the judgment coefficient Pdxs exceeds the preset threshold, it is judged that the infection type of the patient to be verified can be classified by using the natural language processing technology to obtain the infection type of the patient to be verified; If the judgment coefficient Pdxs does not exceed the preset threshold, it is judged that the infection type of the patient to be verified cannot be classified using natural language processing technology at present, indicating that the patient to be verified is not infected at present; S36: At the same time, natural language processing technology is used to classify the infection types of symptomatic infected patients.

8. The method for real-time monitoring of infection cases according to claim 6, characterized in that: Step S4 specifically includes: S41: When the infection type of the patient to be verified is obtained, the infection prediction model is used to analyze the infection status of the patient to be verified and the symptomatic infected patient to obtain the infection degree assessment index Gczs, which is obtained specifically according to the following formula: In the formula, represents the pathogen load, Indicates that the cytokine concentration increases. represents the SOFA score, Indicates the serum albumin level, and are weight values, Represents the second correction constant.

9. The method for real-time monitoring of infection cases according to claim 1, characterized in that: Step S5 specifically includes: S51: By comparing the infection level assessment index Gczs with the infection threshold W, the infection status of the current patient is determined. The specific contents are as follows: If the infection level assessment index Gczs exceeds the infection threshold W, the current patient's infection is abnormal. At this time, the dosage of anti-infection drugs will be adjusted or replaced according to the patient's infection type, and the hospital's prevention and control measures will be strengthened, including personnel protection, environmental disinfection, and visitor restrictions. If the infection level assessment index Gczs does not exceed the infection threshold W, the patient's infection has not yet shown any abnormalities. At this time, the patient's physiological indicators will continue to be monitored, and medical staff will follow up on the patient regularly, paying attention to changes in symptoms and laboratory test results.

10. A real-time monitoring system for infected cases, used to implement the real-time monitoring method for infected cases according to any one of claims 1 to 9, characterized in that: It includes data acquisition module, potential risk analysis module, determination module and comprehensive feedback module; The data acquisition module is used to obtain relevant pathological data information and relevant behavioral data information of each patient in the hospital using the hospital information system, and generate a personalized database after data preprocessing; The potential risk analysis module is used to classify each patient into asymptomatic infection according to the relevant behavioral data information in the personalized database, and perform characteristic difference analysis on the asymptomatic infection patients after classification based on the relevant pathological data information to obtain the characteristic difference factor Tcyz. Based on the numerical value of the characteristic difference factor Tcyz, it is preliminarily estimated whether the corresponding asymptomatic infection patient has an infection risk. If so, it enters the pending review mechanism and marks the patient to be verified; The determination module is used to monitor the patients to be verified and the patients with symptomatic infection in real time to obtain real-time patient data, and analyze the limitations of natural language processing technology in the infection classification of the patients to be verified based on the real-time patient data to obtain the judgment coefficient Pdxs. If the judgment coefficient Pdxs exceeds a preset threshold, the infection type of the patient to be verified is obtained; The comprehensive feedback module is used to use deep learning technology to input real-time patient data into the infection prediction model, and after linear normalization processing, fit the output infection level assessment index Gczs; pre-set the infection threshold W, and compare the infection level assessment index Gczs with the infection threshold W to judge the current patient's infection status, and take corresponding control measures based on the judgment result.

Images