Intelligent grading decision-making and automatic response method and device in infectious disease hospital, and medium

Abstract

The invention provides an intelligent grading decision-making and automatic response method and device in an infectious disease hospital, and a medium, relates to the field of public health information, and aims to carry out comprehensive calculation through a dynamic risk decision-making engine by fusing a real-time prevention and control situation and individual case characteristics, change a traditional static judgment mode based on a fixed rule or a single keyword, and improve the accuracy of judgment. Decision-making auxiliary information including disease types, quantitative risk levels and clinical guidance is actively pushed to clinicians, complex code selection and report judgment are simplified into clear risk prompting and click operation, confirmation operation of the clinicians is automatically converted into executable response operation instructions, a hospital system is driven to execute differentiated prevention and control processes, and the risk prevention and control efficiency of the clinicians is improved. A data-driven continuous optimization system is constructed, so that high-risk unresponded cases can be automatically backtracked, a traceability report is generated, and the management efficiency and pertinence are improved; the closed-loop linkage capability of medical and defense cooperation is enhanced, and the hospital is no longer an information island by simultaneously receiving a macroscopic epidemic situation monitoring platform signal and hospital diagnosis and treatment data.

Description

Technical Field

[0001] This invention relates to the intersection of public health information and smart healthcare technology, and in particular to methods, equipment, and media for intelligent hierarchical decision-making and automatic response in infectious disease hospitals. Background Technology

[0002] my country is focusing on enhancing the core sentinel role of medical institutions in the infectious disease surveillance and early warning system, and promoting the deep integration of medical treatment and prevention, as well as intelligent transformation. In this process, information technology has become a key means to improve the timeliness, accuracy, and proactivity of infectious disease reporting. Existing technical solutions mainly revolve around the automation and intelligent monitoring of data collection, and their evolution and core characteristics are as follows: Early and current technological solutions employed by many medical institutions primarily aimed to digitize the traditional manual paper-based reporting process. These systems are typically integrated into hospital information systems as modules or standalone plugins. By embedding logical checks within physician workstations (such as outpatient physician workstations), a reporting interface automatically pops up and performs mandatory verification when the diagnostic information entered by the physician matches preset infectious disease keywords, thus completing the online reporting. While this approach addresses the standardization and enforceability of the process and significantly reduces omissions due to forgetfulness or negligence, its essence remains a post-diagnosis, manually-driven reporting model. The system's functionality is limited to information entry itself, failing to deeply integrate with broader clinical data and lacking support for physician decision-making.

[0003] To overcome the problems of data silos and inconsistent standards within hospital systems, and to achieve early and macro-level epidemic awareness, a widely deployed technical solution involves proactively and in real-time capturing heterogeneous data from multiple systems within medical institutions, such as HIS, LIS, and EMR, through standardized interfaces. This data is then used to leverage intelligent algorithms such as natural language processing and syndrome analysis to achieve automatic case screening, abnormal cluster warnings, and positive result alerts. This solution marks an upgrade in infectious disease surveillance from passively receiving reports to proactive and intelligent discovery, playing a fundamental role in enhancing regional early warning capabilities for epidemics.

[0004] While both types of technical solutions have achieved significant results at their respective levels, research from frontline medical institutions reveals profound bottlenecks in their clinical implementation. An empirical analysis of existing technologies based on a large hospital's annual data indicates that even in hospitals with high levels of information technology, the quality of clinicians' reporting of legally notifiable infectious diseases remains constrained by multiple factors: for example, physicians in non-key departments face extremely high reporting risks for blood-borne and sexually transmitted diseases; chronic infectious diseases discovered during hospitalization are easily missed; and the system cannot automatically push notifications for abnormal test values, still relying on physicians to actively review them. The root of these problems lies in the fact that the functional boundaries of existing technical solutions stop at information alerts or data reporting, failing to deeply integrate into the specific clinical decision-making and treatment response loop.

[0005] Therefore, it is necessary to provide a new approach to solve the aforementioned technical problems. Summary of the Invention

[0006] To achieve the above-mentioned objectives and other advantages of the present invention, a first objective of the present invention is to provide an intelligent hierarchical decision-making and automatic response method for infectious disease hospitals, comprising the following steps: Obtain a dataset of clinical characteristics of the target cases, the dataset being derived from the macro-level epidemic monitoring platform and / or the in-hospital real-time diagnosis and treatment business system; The clinical feature dataset is input into the dynamic risk decision engine, which outputs at least one suspected infectious disease type and its associated dynamic risk level for the case; wherein the dynamic risk level is dynamically calculated and generated by the engine based on real-time prevention and control situation data and the clinical feature data; Generate and push decision support information containing the disease, its dynamic risk level and clinical guidelines to the clinical terminal; In response to the confirmation operation of the decision support information, based on the selected disease and its risk level, a response operation instruction matching the risk level is automatically generated and triggered to drive the execution of the corresponding differentiated prevention and control process.

[0007] Furthermore, the dynamic risk decision engine calculates the dynamic risk level in the following manner: Based on real-time prevention and control resource data, a prevention and control pressure coefficient is calculated to characterize the current urgency and resource carrying capacity of prevention and control within the hospital. The prevention and control resource data includes the vacancy rate of isolation wards in the infectious disease department, the inventory level of specific protective materials, and the availability of infection control personnel. Based on the clinical feature data and the pre-built infectious disease knowledge graph, the disease threat coefficient characterizing the potential transmission and harmful impact of the case is calculated. The infectious disease knowledge graph stores the transmission routes, typical symptoms and abnormal test indicators of different infectious diseases. The prevention and control pressure coefficient and the disease threat coefficient are comprehensively calculated based on a preset fusion algorithm, and a quantified risk value is output and mapped to a predefined risk level label.

[0008] Furthermore, the prevention and control pressure coefficient is calculated through the following steps: From the aforementioned prevention and control resource data, the vacancy rate of isolation wards in infectious disease departments, the inventory level of specific protective materials, and the availability of infection control personnel were extracted as corresponding resource indicators. Each resource indicator Rx is normalized to obtain the stress sub-score Sx of that indicator; wherein, each resource indicator Rx is preset with an ideal state threshold To, an early warning threshold Tw, and a severe shortage threshold Tc, and the actual value of Rx is mapped to the stress sub-score Sx in the interval [0,1] through a piecewise sigmoid function; Assign dynamic weights to each resource indicator ;in, , The static weight of this resource indicator is set in advance. This is the current hospital-wide special treatment load coefficient calculated based on the number of outpatient visits for a specific type. For load sensitivity factor, 0≤ ≤0.3; The stress sub-scores Sx of each resource indicator are associated with their dynamic weights. Perform weighted aggregation to obtain the initial weighted sum And through convex function transformation The control pressure coefficient is obtained by performing nonlinear amplification; where τ is the pressure amplification index, 0.5 < τ < 0.8.

[0009] Furthermore, the disease threat coefficient is calculated through the following steps: Clinical characteristic data are matched with the infectious disease knowledge graph to form an evidence set consisting of etiological confirmatory evidence, clinically highly suggestive evidence, clinically general suggestive evidence, and epidemiological correlation evidence. The confidence decay aggregation algorithm is used to calculate the collective confidence of each type of evidence set. Among them, for an evidence set containing multiple evidence items, its collective confidence level... , Let λ be the confidence level of the i-th evidence item in this type of evidence set, λ be the collaboration gain coefficient, and Π be the multiplication sign. Based on the preset basic contribution weights, the collective confidence levels of each evidence set are weighted and synthesized to obtain the basic threat level. ;in, , , , , These are the pre-defined basic contribution weights for etiological confirmatory evidence, clinically highly suggestive evidence, clinically generally suggestive evidence, and epidemiologically related evidence, respectively, and they meet the following requirements: , , , , The collective confidence levels are defined as follows: etiological confirmatory evidence, clinically highly suggestive evidence, clinically generally suggestive evidence, and epidemiologically related evidence. For inhibitory factors based on negative or contradictory evidence; Based on the current urgency coefficient U of the patient's medical visit and the risk perception coefficient of the department from which the patient originates. Regarding the aforementioned basic threat level Perform gain correction to obtain the corrected threat level. ;in, , and These are the adjustment intensity coefficients for urgency and department, respectively; The modified threat level is obtained through a saturation function. Perform nonlinear calibration to output the final disease threat coefficient β; where, ξ is a saturation rate parameter that is greater than zero.

[0010] Furthermore, the fusion algorithm is configured to amplify the risk value corresponding to the same disease threat coefficient when the prevention and control pressure coefficient increases; specifically, the calculation formula for the risk value R is: Where α is the prevention and control pressure coefficient, β is the disease threat coefficient, and k is the tension amplification coefficient that is greater than zero.

[0011] Furthermore, the calculation of the dynamic risk level also incorporates a departmental risk correction factor based on the department from which the cases originated; the method further includes the following steps: After calculating the risk value R, a pre-set departmental risk profile table is queried to obtain the historical infectious disease underreporting rate corresponding to the department from which the target case originated; the departmental risk profile table records the historical infectious disease underreporting rate of each department in the hospital; Determine whether the obtained historical infectious disease underreporting rate is greater than the preset underreporting rate threshold; if so, assign a department correction factor γ greater than 1 to the case; otherwise, set the department correction factor γ=1. Multiply the risk value R by the departmental correction factor γ to obtain the corrected risk value; The revised risk value is then mapped to the final risk level label.

[0012] Furthermore, the decision support information is presented on the clinical terminal in a visual interface, where suspected infectious diseases of different risk levels are highlighted with differentiated colors, icons, or sorting methods.

[0013] Furthermore, the differentiated prevention and control process includes at least three response levels from low to high, and the response operation instructions are automatically matched to the corresponding response level according to the selected risk level, wherein: The minimum response level instruction set must include at least the generation of a standard reporting document; The intermediate response level instruction combination, based on the lowest response level, adds the sending of collaborative handling reminders to preset internal collaborative roles; The highest response level instruction combination, based on the intermediate response level, further adds instructions to synchronize information to a preset out-of-hospital collaborative data interface and applies at least one clinical process intervention to the case.

[0014] Furthermore, the clinical process intervention includes: applying a process locking instruction to the target case via an interface call to restrict the progress of some or all of its non-urgent clinical business processes.

[0015] Furthermore, it also includes a backtracking analysis step: Periodically retrieve the historical output records of the dynamic risk decision engine; Identify abnormal case records that are assessed as high-risk but fail to generate effective response instructions within a preset time. The abnormal case records are integrated to generate a high-risk underreporting tracing report, which is then pushed to the management terminal.

[0016] Furthermore, it also includes engine optimization steps: Collect historical decision records of the dynamic risk decision engine, confirmation operation results of clinical terminals, and data on the execution effect of triggered prevention and control processes to form an optimization feedback set; Based on the optimized feedback set, the evaluation logic of the dynamic risk decision engine is iteratively optimized.

[0017] A second objective of the present invention is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0018] A third objective of the present invention is to provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.

[0019] Compared with the prior art, the beneficial effects of the present invention are: Compared with the prior art, the present invention has achieved the following outstanding advantages: This invention provides a method, equipment, and medium for intelligent hierarchical decision-making and automatic response within infectious disease hospitals. It achieves a leap from static early warning to dynamic intelligent decision-making. Through a dynamic risk decision engine, it integrates real-time prevention and control situation (such as resource scarcity) with individual case characteristics for comprehensive calculation, changing the traditional static judgment mode based on fixed rules or single keywords. This allows the risk level to truly reflect the dynamic balance between disease threat and current prevention and control capabilities, making early warning more scientific and accurate, and effectively reducing false alarms and missed reports. It proactively pushes decision support information to clinicians, including disease type, quantitative risk level, and clinical guidelines, simplifying complex coding selection and reporting judgments into clear risk prompts and point-and-click operations. This greatly reduces the cognitive load and operational time of physicians, solving the problem of low reporting efficiency caused by traditional systems that only provide reminders without assistance.

[0020] This invention automatically transforms physician confirmation actions into a series of executable response commands, driving the hospital system to execute differentiated prevention and control procedures. This completely changes the previous disconnected model of manual reminders, communication, and implementation, achieving second-level automatic response and significantly shortening the time window from risk detection to prevention and control initiation. It automatically matches different levels of prevention and control commands based on dynamic risk levels, such as from simple reporting to bed locking, infection control notification, and synchronized disease control, avoiding waste or inadequacy of prevention and control resources. This ensures that limited infection control resources are prioritized and accurately allocated to the highest-risk cases and processes, improving prevention and control efficiency and safety.

[0021] This invention constructs a data-driven continuous optimization system that can automatically trace back high-risk unresponsive cases and generate source tracing reports. This enables management departments to shift from a crude mode of full-scale manual screening to a data-based precise verification mode, greatly improving management efficiency and targeting. By collecting feedback data on decision-making results and prevention and control effects, the core assessment logic can be iteratively optimized, allowing its risk judgment capability to continuously improve with actual operational data. It has a self-learning ability that becomes more accurate with use, fundamentally solving the problem of traditional system rules being rigid and difficult to adapt to new outbreaks or localized scenarios.

[0022] This invention strengthens the closed-loop linkage capability of medical and preventive collaboration. By simultaneously receiving signals from the macro-level epidemic monitoring platform and in-hospital diagnosis and treatment data, hospitals are no longer information silos. Hospitals can act as intelligent nodes, quickly responding to early warnings from higher authorities and synchronizing internal risks externally. This truly achieves vertical integration and horizontal collaboration of monitoring, early warning, decision-making, and response within the region, enhancing the resilience and efficiency of overall infectious disease prevention and control in the region.

[0023] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Specific embodiments of the present invention are given in detail below with reference to the accompanying drawings. Attached Figure Description

[0024] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 Flowchart of intelligent hierarchical decision-making and automatic response method within infectious disease hospitals; Figure 2 Flowchart for calculating dynamic risk levels for the dynamic risk decision engine; Figure 3 Flowchart for backtracking analysis; Figure 4 Optimize the flowchart for the dynamic risk decision engine; Figure 5 A diagram of the intelligent recommendation module for risk diseases in the clinical terminal visualization interface; Figure 6 A diagram of the intelligent operation panel module in the clinical terminal visualization interface; Figure 7 This is a schematic diagram of a computer device. Figure 8 This is a schematic diagram of a computer-readable storage medium. Detailed Implementation

[0025] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0026] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0027] The drawing numbers in this application are only used to distinguish the steps in the scheme and are not used to limit the execution order of the steps. The specific execution order is as described in the specification.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0029] This invention, based on receiving macro-level early warning signals, provides clinicians with intelligent decision support based on real-time risk assessment. It can automatically and accurately extend the act of reporting information into a series of executable and traceable tiered prevention and control responses within the hospital, thereby truly achieving intelligent integration from risk perception to closed-loop management. The specific solution is as follows: Example 1

[0030] A method for intelligent hierarchical decision-making and automatic response in infectious disease hospitals, such as Figure 1 As shown, it includes the following steps: S100. Obtain a dataset of clinical characteristics of the target case, wherein the dataset is derived from the macro-epidemic monitoring platform and / or the in-hospital real-time diagnosis and treatment business system; In this embodiment, a data sensing and adaptation module deployed in the hospital information network proactively acquires or receives risk signals related to the target case from two dimensions. One dimension comes from data in the hospital's internal business flow. This module monitors events generated in the hospital's medical business systems (such as HIS, LIS, and EMR) in real time. For example, when an outpatient doctor's workstation submits a diagnosis containing complaints such as fever and rash, or when the LIS system issues a report of a positive Vibrio cholerae nucleic acid test, the module immediately captures these structured or unstructured medical events and extracts key features, such as diagnostic descriptions, test items and results, and keywords of examination conclusions, to form a preliminary case feature vector.

[0031] Another dimension is the early warning signals from external public health monitoring data sources. This module receives early warning information pushed by the superior intelligent monitoring and early warning platform through standardized data interfaces such as APIs. The system performs spatiotemporal correlation matching between this early warning and the information of patients visiting the hospital. If a related patient is found, the patient is marked as an external early warning related case, and information such as the early warning disease type and early warning level is incorporated into the patient's feature vector.

[0032] Finally, the aggregated and cleaned data is integrated into a structured clinical feature dataset, which serves as input for subsequent intelligent assessment. For example, the clinical feature dataset may include fields such as unique case identifiers, data aggregation time, and primary triggering sources.

[0033] Optionally, real-time data aggregation can be achieved through a combination of asynchronous event listening and proactive data retrieval. The core mechanisms for data acquisition include Mode A (Event-driven): listening for state change events on specific interfaces, and triggering data collection upon event arrival; Mode B (Condition-triggered): when preset composite logical conditions are met, the system proactively initiates query requests to relevant data sources; and Mode C (Periodic Synchronization): maintaining a dynamic cache pool for critical external situational information and matching it with entity information as needed.

[0034] To obtain macro-level situational awareness data, we can first define the data interface abstraction, including the interface identifier: external public health situational awareness interface, the data format: standardized regional risk situation notification messages, and the transmission protocol: secure data push or query based on HTTPS.

[0035] The specific acquisition logic is as follows: The system needs to determine whether the target entity is affected by external regional risks. An external public health situation awareness interface pushes a new situational announcement, including a geographic range code, risk level label, code for the type of pathogen of concern, and an effective time window. The system acquires the spatiotemporal attributes of the target entity, including the residence / treatment location code, which can be extracted from the location attribute field of the internal business system, and the timestamp, i.e., the entity's current treatment time. Then, a matching algorithm is executed: when the target entity's location attributes are completely within the geographic area defined by the situational announcement, and the current time is within the announcement's effective time window, the system determines it as a complete match, and the external risk association strength is assigned a value of 1.0. If the target entity's location attributes are not directly within the core geographic area, but are located in its adjacent or related surrounding areas, the system calculates an association strength value between 0 and 1 based on a preset spatial decay function. The output value of this function monotonically decreases with increasing geographic distance. In this case, the match is still considered successful, but the association strength is less than 1.0.

[0036] The structured fields (risk level, pathogen type) in the successfully matched announcements are extracted, and the external risk association factors, i.e. the association strength, are calculated. This information is stored as a macro-level feature in the entity's clinical feature dataset.

[0037] To achieve real-time acquisition of in-hospital diagnostic and treatment business data, the data source is first abstracted and defined. The system interacts with the in-hospital system through the following four logical adapters: Adapter A (Patient Flow): connects to the core diagnostic and treatment business management system to obtain entity status, location, and schedule; Adapter B (Laboratory Flow): connects to the laboratory diagnostic object status management system to obtain test and examination results; Adapter C (Text Knowledge Flow): connects to the free text diagnostic and treatment record library to extract semantic entities through an NLP engine; Adapter D (Resource Status Flow): connects to the real-time status system of medical resources to obtain the status of physical and human resources.

[0038] Taking Mode B: Conditional Trigger as an example, the specific acquisition logic is as follows: The trigger condition is that the detection status of a key pathogenic marker of the entity reported by Adapter B changes to positive or significantly abnormal. The system response sequence includes, after the condition is triggered, the system immediately initiates collaborative query requests to other adapters: It requests Adapter A to obtain the entity's current treatment context, with the returned data format being {Department: Type K, Status: Inpatient, Plan: Elective Operation, Primary Diagnosis: Non-infectious Disease}; it requests Adapter C to parse the semantics related to the code of the pathogen of concern in the entity's recent records, with the returned data format being {Historical Mentions: "Carrying Status", Recent Symptom Description: None}; and it requests Adapter D to obtain the current status of dedicated resources for responding to the code of the pathogen of concern, with the returned data format being {Isolation Unit Occupancy Rate: 85%, Specific Protective Equipment Inventory: Sufficient}.

[0039] The system automatically compares the data returned by different adapters: Contradiction 1: Adapter B (positive marker) compared with adapter C (no recent symptom description), generating a data consistency warning mark; Contradiction 2: Adapter A (non-infectious disease department & elective operation) compared with adapter B (positive marker), generating a clinical scenario risk mark.

[0040] The raw data is transformed into feature vectors: from adapter B data, biological threat intensity features are extracted based on marker type and value; from adapter A data, diagnosis and treatment scenario features are extracted, such as department type, inpatient / outpatient, and operational urgency; from adapter C data, semantic consistency features are extracted, namely the text matching degree between history and current status; and from adapter D data, resource constraint features are extracted, namely the tension of prevention and control resources for this pathogen.

[0041] All acquired data is appended with a uniform event time and system reception time, and converted to a standard time zone, such as UTC. A version number is generated for the entity's clinical feature dataset, and any subsequent data updates will generate a new version.

[0042] Record the genealogy constructed for this dataset, including the triggering mode (A / B / C), the response status and latency of each adapter, and the data quality verification results, such as the field missing rate.

[0043] S200. Input the clinical feature dataset into the dynamic risk decision engine and output at least one suspected infectious disease type and its associated dynamic risk level for the case; wherein, the dynamic risk level is dynamically calculated and generated by the engine based on real-time prevention and control situation data and the clinical feature data; In some embodiments, after receiving the dataset output from step S100, a set of dynamic evaluation logic is executed to output a quantified risk assessment result. Specifically, such as... Figure 2 As shown, the dynamic risk decision engine calculates the dynamic risk level in the following way: S210. Based on real-time prevention and control resource data, calculate the prevention and control pressure coefficient, which represents the current urgency and resource carrying capacity of prevention and control in the hospital. The prevention and control resource data includes the vacancy rate of isolation wards in the infectious disease department, the inventory level of specific protective materials, and the availability of infection control personnel. This embodiment does not assess the disease in isolation, but rather incorporates the hospital's real-time capacity. It periodically collects a set of dynamic parameters from the hospital's resource management system, such as the current vacancy rate of negative pressure isolation wards in the infectious disease department, the inventory turnover days of specific levels of protective equipment, and the on-duty status of infection control personnel. Through a pre-set normalization and weighted calculation model, these parameters are integrated and calculated into a value between 0 and 1, which serves as the prevention and control pressure coefficient α. A higher α value indicates a more strained current prevention and control resources and a lower capacity to withstand pressure at the hospital.

[0044] Optionally, the normalization and weighted calculation model is as follows: From the aforementioned epidemic prevention and control resource data, the vacancy rate of isolation wards in infectious disease departments, the inventory level of specific protective materials, and the availability of infection control personnel are extracted as corresponding resource indicators. For single-indicator normalization, three thresholds are defined for each indicator Rx: the ideal state threshold To, the warning threshold Tw, and the severe shortage threshold Tc. The actual value is mapped to a tension sub-score Sx within the interval [0,1] using a piecewise sigmoid function. When Rx is better than To, Sx≈0; when Rx is near Tw, Sx smoothly rises to 0.5; when Rx is worse than Tc, Sx rapidly approaches 1. This function is first-order differentiable, avoiding abrupt changes in results caused by hard thresholds, making the pressure coefficient change smoother and more stable.

[0045] The ideal state threshold To and the warning threshold Tw are mainly determined based on infection control guidelines, hospital accreditation standards, and best practices from the hospital's historical operational data. For example, the To for the vacancy rate of negative pressure wards in Department A can be set by referring to the recommendations on the proportion of spare beds in the hospital's isolation technical guidelines and combining them with the average vacancy rate of the same period in previous years.

[0046] The severe shortage threshold Tc is primarily determined based on the resource safety stock theory, emergency response plans, and expert assessments. It signifies that the level of resource scarcity has reached a critical point where an emergency response must be initiated. For example, the Tc for protective equipment is determined jointly by the hospital's infection control department and materials management department based on the equipment's procurement cycle and average daily consumption, to ensure a minimum inventory level for safe operation.

[0047] During system initialization, these thresholds are set as configurable parameters by the hospital administrator based on the above criteria.

[0048] Then, dynamic weight allocation is performed, assigning weights to each type of resource indicator. It consists of two parts: ,in, This refers to the static weights pre-set based on the fundamental importance of these resource indicators in standard prevention and control procedures. For example, At its core, isolation space is the most basic and most difficult resource to expand instantly. Secondly, supplies can be procured urgently, but this takes time. and Secondly. The current special treatment load coefficient of the entire hospital, such as the proportion of fever / diarrhea clinics to daily outpatient volume, is normalized to 0-1. For load sensitivity factor (0≤ ≤0.3), used to appropriately increase the weight of materials and human resources when the load increases.

[0049] Among them, static weights The criteria for setting these limits are based on the irreplaceable nature of various prevention and control resources in standard prevention and control procedures and the difficulty of emergency expansion. For example, the vacancy rate of isolation wards... The highest priority is given to physical isolation spaces, as they are fundamental to infection control and cannot be expanded instantaneously; the next highest priority is the stockpile of protective equipment, as it can be replenished through emergency procurement; and the next highest priority is the availability of personnel. This arrangement aligns with the relevant regulations' requirements for prioritizing isolation, protection, and personnel allocation.

[0050] The aforementioned weights are preset by the implementer during system initialization, based on the above logic and the actual situation of the institution, in the form of configuration files or management backend parameters.

[0051] Finally, weighted aggregation and nonlinear amplification are performed. First, the preliminary weighted sum is calculated: ,in To highlight the shortcomings and synergistic effects of increased pressure when multiple resources are simultaneously strained, a convex function transformation is introduced: ,in, The pressure amplification index is a constant between 0.5 and 0.8, for example, 0.65. When all When both are low (resources are abundant), Small, Even smaller, the square root effect further reduces it; when multiple When resources are generally scarce, Increase Power functions with an exponent less than 1 approach 1 at a much faster rate. This precisely simulates reality: a shortage of a single resource can be managed, but a simultaneous shortage of multiple resources will cause the system's resilience to decline exponentially.

[0052] S220. Based on the clinical feature data and the pre-set infectious disease knowledge graph, calculate the disease threat coefficient that characterizes the potential transmission and harmful impact of the case. The infectious disease knowledge graph stores the transmission routes, typical symptoms and abnormal test indicators of different infectious diseases. This embodiment utilizes a pre-built infectious disease attribute knowledge base. This knowledge base not only includes disease names and codes but also associates attributes such as transmission routes (droplets, contact, blood, etc.), average incubation period, typical symptom sets, and common abnormal test indicator combinations in a graph format. The system matches clinical feature data with the infectious disease knowledge graph to form an evidence set consisting of etiological confirmatory evidence, clinically highly suggestive evidence, clinically generally suggestive evidence, and epidemiological correlation evidence. For example, for a case with abnormal liver function (ALT>200U / L) and HBsAg positivity, the knowledge base will point to viral hepatitis (hepatitis B) and, based on its transmission routes, chronicity characteristics, and individual information such as the case's age and gestational age, output a disease threat coefficient β between 0 and 1 using a threat calculation function. The higher the β value, the greater the potential transmission risk and health hazard if the case is not intervened.

[0053] Optionally, the threat level calculation function is calculated as follows: first, evidence sets are grouped and confidence levels are aggregated; for each type of evidence set... If it contains multiple pieces of evidence, they are etiological confirmatory evidence. Clinical evidence strongly suggests Clinical evidence generally suggests Epidemiological link evidence Negative / Contradictory Evidence Its collective confidence level Calculation using confidence decay aggregation: ,in, Let λ be the confidence level of the i-th piece of evidence in this class, and λ be the collaboration gain coefficient, such as 0.2. This is the multiplication symbol.

[0054] Then, the base threat level is synthesized. The calculation is as follows: in, , , , These are the pre-defined basic contribution weights for evidence confirming the pathogen, evidence strongly suggestive of clinical findings, evidence generally suggestive of clinical findings, and evidence related to epidemiology. The weight allocation follows these principles: ; , , , These are the collective confidence levels for etiological confirmatory evidence, clinically highly suggestive evidence, clinically generally suggestive evidence, and epidemiologically relevant evidence. As an inhibitory factor based on negative or contradictory evidence, This acts as a suppressor of contradictory evidence. If strong refutational evidence exists, such as valid proof of immunity, It can approach 0; if there is no contradictory evidence or the contradictory evidence is very weak, .

[0055] Among them, the basic contribution weight , , , The weighting of etiological confirmatory evidence is based on the inherent strength of different medical evidence in clinical diagnosis. The highest weighting is based on its diagnostic significance; clinical evidence is next; and epidemiological evidence is the weakest. This weighting follows the principle of evidence hierarchy established in infectious disease diagnostic criteria and related clinical practice guidelines.

[0056] The aforementioned weights are preset by the implementer during system initialization, based on the above logic and the actual situation of the institution, in the form of configuration files or management backend parameters.

[0057] Then, clinical context gain adjustment is performed, modifying the baseline threat level based on the patient's current urgency and the perceived risk within the department:

[0058] U represents the urgency coefficient for seeking medical attention (0-1). Emergency and resuscitation patients have a high U value, such as 0.8; elective hospitalization patients have a low U value, such as 0.1. The departmental risk perception coefficient (0-1) is derived from the departmental risk profile database, representing departments with a historically high rate of underreporting of infectious diseases. The value is low. , The correction intensity coefficients (0≤η≤0.3) are for urgency and department, respectively, to control the correction range.

[0059] For emergency patients or patients in departments with a high rate of historical underreporting of infectious diseases, their perceived threat level should be appropriately amplified to compensate for the underestimation of risk that may occur in clinical decision-making due to time constraints or lack of experience.

[0060] Finally, nonlinear calibration and output are performed, and the final threat coefficient β is calibrated using a saturation function to simulate the evidence saturation effect. ,in, For saturation rate parameters, ,like .when When smaller, Nearly linear growth; when When it is large, The growth rate slows down and gradually approaches 1, avoiding the infinite accumulation of evidence that could lead to an infinite amplification of the threat level, which aligns with clinical cognitive logic.

[0061] S230. The prevention and control pressure coefficient and the disease threat coefficient are comprehensively calculated according to the preset fusion algorithm, and a quantified risk value is output and mapped to a predefined risk level label.

[0062] After obtaining the prevention and control pressure coefficient α and the disease threat coefficient β, based on the context-aware fusion function The calculation is performed. The fusion function is configured such that when resources are highly strained (i.e., α is high), diseases of the same threat level (β) should be assigned a higher risk value to prompt the system to take more decisive intervention measures to prevent resource exhaustion. For example, the fusion function is: risk value... , where k is a tension amplification factor greater than 0. The calculated risk value R (e.g., 0.85) will be mapped to a preset level label, such as high (R≥0.7), medium (0.4≤R<0.7), and low (R<0.4).

[0063] To improve the accuracy of the assessment, this invention also introduces a departmental risk perception mechanism. Specifically, the calculation of the dynamic risk level also incorporates a departmental risk correction factor based on the department from which the case originates; wherein, for cases from departments with a preset historical infectious disease underreporting rate higher than a set threshold, the calculated risk value is increased according to the correction factor. Specifically, the method further includes the following steps: After calculating the risk value R, a pre-set departmental risk profile table is queried to obtain the historical infectious disease underreporting rate corresponding to the department from which the target case originated; the departmental risk profile table records the historical infectious disease underreporting rate of each department in the hospital; Determine whether the obtained historical infectious disease underreporting rate is greater than the preset underreporting rate threshold; if so, assign a department correction factor γ greater than 1 to the case; otherwise, set the department correction factor γ=1. Multiply the risk value R by the departmental correction factor γ to obtain the corrected risk value; The revised risk value is then mapped to the final risk level label.

[0064] This embodiment maintains a departmental risk profile table, recording the historical reporting sensitivity of non-infectious disease departments to infectious diseases (especially bloodborne and sexually transmitted diseases). When a case comes from a department with a historical infectious disease underreporting rate higher than the underreporting rate threshold, a departmental correction factor γ greater than 1 (e.g., γ=1.3) is multiplied by the final risk value to proactively increase the risk output, compensating for potential insufficient risk awareness among the department's medical staff and ensuring that high-risk cases are not missed.

[0065] S300. Generate and push decision support information containing the disease type, its dynamic risk level and clinical guidelines to the clinical terminal; further, the decision support information is presented on the clinical terminal in a visual interface, wherein suspected infectious diseases of different risk levels are highlighted with differentiated colors, icons or sorting methods.

[0066] In this embodiment, after the assessment is completed, a contextualized decision support information is generated. This decision support information includes at least: a list of recommended diseases (arranged in descending order of risk value); a dynamic risk level label for each disease (indicated visually by red, yellow, and green); and key clinical guidelines and reporting prompts, such as high-risk cholera: immediate gastrointestinal isolation is required, and the reporting time limit is 2 hours.

[0067] This decision support information is pushed to the attending physician's terminal via the hospital's internal communication system, such as integration into the physician's workstation or mobile office app. The interface design uses a high-contrast visualization scheme to ensure that physicians can focus on the highest-risk items within seconds.

[0068] like Figure 5 , Figure 6 As shown, the clinical terminal visualization interface abandons the form-filling mindset of traditional infectious disease reporting systems and adopts a risk-driven information architecture. It allocates the largest screen space and the strongest visual contrast to risk level and recommended disease, rather than basic patient information. The interface's goal is not to have doctors fill out a card, but to assist doctors in completing a complete decision chain of risk identification, disease confirmation, and response triggering within 10 seconds.

[0069] The interface organizes information according to a logical framework of risk level, disease name, key evidence, and clinical guidelines, avoiding information overload. Structured data such as patient information and test results are automatically extracted and highlighted by the system, allowing doctors to focus solely on confirming the most critical risk level. Each action button clearly indicates its consequences, such as immediate intervention or routine reporting, eliminating operational uncertainty. For example, the patient information overview displays the patient's core diagnostic and treatment information, including case ID, name, department, and relevant triggers for the visit.

[0070] The interface implements a risk level visualization coding system, and the sorting of the recommended disease list follows a multi-factor weighted algorithm. For high-risk operations, a confirmation layer is forced to pop up, clearly listing the four operations that will be automatically executed: generating a report card, notifying infection control, synchronizing with disease control, and applying clinical intervention.

[0071] The high-risk confirmation button has a 500ms click delay to avoid triggering a critical response immediately due to accidental touch. All operations can be undone in the operation history panel within 5 minutes, and a reason for cancellation must be filled in.

[0072] The interface adopts a responsive design, automatically adjusting the layout on the doctor's workstation (large screen), mobile ward round PAD (medium screen), and mobile phone (small screen). When the network is interrupted, the interface can cache the 10 most recent high-risk alerts and support offline confirmation. It will automatically synchronize after the network is restored. In scenarios where hand operation is restricted, such as operating rooms and ICUs, it supports voice commands such as confirming high risks and viewing the next alert.

[0073] The average time from the pop-up interface to the doctor completing the operation is less than 15 seconds. The number of clicks / inputs required to complete a standard report has been reduced from 8-12 steps in the traditional system to 1-3 steps. After the doctor selects excluded cases, the system reviews them and triggers an early warning reminder within 24 hours.

[0074] S400: In response to the confirmation operation of the decision support information, based on the selected disease and its risk level, automatically generate and trigger response operation instructions that match the risk level to drive the execution of the corresponding differentiated prevention and control process.

[0075] The differentiated prevention and control process includes at least three response levels from low to high. The response operation instructions are automatically matched to the corresponding response level based on the selected risk level. The minimum response level instruction set must include at least the generation of a standard reporting document; The intermediate response level instruction combination, based on the lowest response level, adds the sending of collaborative handling reminders to preset internal collaborative roles; The highest response level instruction combination, based on the intermediate response level, further adds instructions to synchronize information to a preset out-of-hospital collaborative data interface and applies at least one clinical process intervention to the case.

[0076] Specifically, the clinical process intervention includes: applying a process locking instruction to the target case through an interface call to restrict the progress of some or all of its non-urgent clinical business processes.

[0077] Optionally, when the decision engine determines that clinical process intervention needs to be implemented on the target entity, the system will generate a set of standardized business lock instructions and send them to the core business execution system of the medical institution through a preset business process control interface. This allows temporary logical blocking to be applied to non-urgent and delayable process nodes without interrupting emergency and life-saving treatment, so as to ensure that infection control measures are implemented first.

[0078] A unified control interface for coordinating clinical business processes is implemented, based on synchronous / asynchronous calls via message queues or APIs, and employing a two-way certificate authentication and operation token mechanism. Each lock command includes a unique identifier, a globally unique command number automatically generated by the system. Its encoding rules typically integrate the lock action identifier, the unique identity code of the target entity, and the command generation timestamp, used to accurately track the lifecycle of the command across the entire system; a target entity identifier, which explicitly specifies the specific business object targeted by this command, i.e., the unique identifier of the target case or patient requiring process intervention within the information system; an intervention operation type, which specifies the specific category of clinical business process to be locked in a predefined coded form; a lock reason, which provides a concise and standardized description of the fundamental reason for implementing the lock, consisting of a predefined, fixed expression indicating that the lock can only be released after risk assessment and completion of specific prerequisite processes; and unlock conditions, which define the prerequisites for the lock command to automatically invalidate or be allowed to be released, using logical expressions or explicit conditions, such as meeting preset system automatic review rules or completing preset associated process nodes (e.g., the associated infectious disease report card has been approved).

[0079] In this embodiment, physicians review decision support information and make selections on the terminal, such as clicking to confirm the highest-risk disease. This confirmation operation acts as a trigger, activating the automated response orchestration engine. This engine has pre-built, executable response script templates strictly bound to different risk levels. Responses are divided into three levels: a low-level response automatically calls the electronic medical record system interface to fill in and generate a draft infectious disease report card, and pushes it to a preset internal review process queue; a medium-level response, while completing the report, automatically sends a collaborative handling reminder to preset internal collaborative roles through the hospital's internal messaging platform, with the message including key case information and suggested measures; a high-level response, building on the above, adds two core automated operations: encrypting and synchronizing key early warning information through the regional health information platform interface, and automatically sending a treatment process intervention request to the hospital's HIS system. This request, through the HIS's open interface, applies a soft lock to the target case's electronic file to restrict the progress of some or all non-urgent clinical business processes. This intervention ensures that the physical movement of high-risk cases is effectively controlled before proper treatment, forming a true closed loop of monitoring and interception.

[0080] To focus management attention directly on the very few abnormal cases most likely to be missed and requiring intervention, and to identify situations where the system has issued high-risk alerts but clinical terminals have not confirmed them, design flaws in the system's alerting mechanisms (such as inconspicuous pop-ups or unreachable push notifications) or usage obstacles in specific work environments (such as operating rooms or emergency departments) may be exposed. In some embodiments, such as Figure 3As shown, it also includes S500 and backtracking analysis steps: S510. Periodically retrieve the historical output records of the dynamic risk decision engine; S520, Identify abnormal case records that are assessed as high-risk but have not generated valid response operation instructions within a preset time. S530. Integrate the abnormal case records to generate a high-risk underreporting tracing report and push it to the management terminal.

[0081] This embodiment designs a background analysis process that runs a source tracing analyzer periodically (e.g., daily). This analyzer retrieves historical output records from the dynamic risk decision engine. Using pattern recognition rules, it specifically screens out cases assessed as high-risk, but for which no corresponding valid response operation command generation records were found within a subsequent period (e.g., 12 hours). These cases are identified as high-risk missing cases. The analyzer summarizes their information, the context of the assessment at the time, and possible attending physicians, generating a high-risk underreporting source tracing analysis report. This report is automatically pushed to the command dashboard of hospital public health management personnel, guiding them to conduct precise offline verification and intervention, thus forming a closed loop of management and supervision.

[0082] To continuously perceive the gap between its own judgment and real-world clinical feedback, it fundamentally addresses the problem of the system's disconnect from clinical practice. In some embodiments, such as Figure 4 As shown, it also includes S600 and engine optimization steps: S610. Collect the historical decision records of the dynamic risk decision engine, the confirmation operation results of the clinical terminal, and the execution effect data of the triggered prevention and control process to form an optimization feedback set; This embodiment also includes an offline optimization module, which periodically (e.g., weekly) collects feedback datasets. These datasets include: historical risk levels (R), physician confirmation behavior (whether confirmation was made, which disease was confirmed, etc.), and the actual execution status and effectiveness evaluation of subsequent prevention and control procedures, such as whether isolation was timely and whether reporting was accurate.

[0083] S620. Based on the optimized feedback set, iteratively optimize the evaluation logic of the dynamic risk decision engine.

[0084] Using this data, the offline optimization module fine-tunes the parameters of the fusion function F mentioned in step S200, the departmental correction factor γ, and even the disease threat weights in the knowledge base through feedback learning algorithms, such as gradient descent-based online learning or reinforcement learning frameworks. The goal of optimization is to ensure that the risk level output by the system is as consistent as possible with the urgency of the treatment ultimately recognized by clinicians and the actual necessity of that treatment.

[0085] Optionally, let the system's risk assessment output for the i-th case be... The mapped risk levels are, for example, high = 3, medium = 2, low = 1, while the clinical consensus label for this case is... It is generated by fusing feedback data from the following three parts: physician confirmation label. The actual treatment level selected by the physician at the terminal, such as high risk = 3, medium risk = 2, exclusion = 0; prevention and control implementation label. The actual level of the subsequent prevention and control process triggered, such as executing a high-level response = 3; effect verification tag. In retrospect (e.g., 24 hours later), whether the case was confirmed as a reportable infectious disease, yes = 1, no = 0.

[0086] The ultimate optimization objective is to minimize the following loss function: Where θ is the set of parameters to be optimized, the weights of the fusion function F, the departmental correction factor γ, the disease threat weight, etc. , , The weighting coefficients for the losses of each objective are used to balance the importance of different feedback signals; This is a regularization term to prevent overfitting.

[0087] Extract quadruples from historical logs ,in This represents the original feature vector of the case at that time, such as clinical characteristics and resource status. This refers to the risk level calculated by the system at that time based on the old set of parameters to be optimized.

[0088] This embodiment uses a reinforcement learning framework as an example. It is modeled as a Markov Decision Process (MDP), whose states... The feature vector of case t In addition to the current state of hospital resources, actions Rewards for the risk level (low, medium, high) selected by the system The strategy is to calculate immediate rewards based on subsequent feedback. The risk assessment model is determined by the parameter θ.

[0089] Reward function guides model learning of clinical consensus and the necessity of prevention and control: in, , , These are preset weighting coefficients used to adjust the relative importance of different feedback dimensions. As a consensus reward for clinicians, To implement matching rewards for epidemic prevention and control, To verify the reward for the final outcome.

[0090] The clinician consensus reward measures the consistency between the system's risk assessment and the clinician's actual judgment. The highest reward is +2.0, awarded when the risk level automatically assessed by the system is completely consistent with the treatment level ultimately selected by the doctor on the terminal. This indicates a high degree of alignment between the system's judgment and the clinician's immediate understanding. A moderate penalty is -1.0, triggered when the system's assessment level differs from the doctor's selected level by one level (e.g., the system assesses high risk, but the doctor selects risk), indicating an acceptable deviation between the system's judgment and the doctor's assessment. A severe penalty is -3.0, triggered when the system's assessment level differs from the doctor's selected level by two levels or more, or when the doctor directly selects exclusion, assuming it is not an infectious disease. This indicates that the system may have issued a serious false alarm, interfering with clinical work.

[0091] The prevention and control execution matching reward measures the degree of matching between the automatic response triggered by the system and the subsequent actual prevention and control measures. A positive reward of +1.0 is obtained when the risk level assessed by the system and the actual level of the prevention and control process that is triggered are perfectly matched, indicating that the system's early warning accurately drives the appropriate response. A mild penalty of -0.5 is triggered when the system's assessment level is higher than the actual level of prevention and control implemented, which means that the system may be over-warning, resulting in unnecessary consumption of alarm resources or disruption of clinical processes. A severe penalty of -1.5 is triggered when the system's assessment level is lower than the actual level of prevention and control implemented, which means that the system's early warning is insufficient, its risk assessment fails to fully reflect the actual urgency of prevention and control, and may lead to response delays or inadequate measures.

[0092] The final outcome verification reward is based on the final truth of the case, measuring the objective accuracy of the system's risk assessment. A significant positive reward of +3.0 is awarded when the case is ultimately confirmed as a reportable infectious disease, and the system's initial assessment was medium or high risk. This is the highest affirmation of the system's successful identification of a real threat. A significant penalty of -3.0 is triggered when the case is ultimately confirmed as an infectious disease, but the system initially assessed it as low risk. This indicates a serious underreporting by the system, a situation that must be avoided during the optimization process. A moderate positive reward of +1.0 is awarded when the case is ultimately ruled out as an infectious disease, and the system initially assessed it as low risk. This affirms the system's correct judgment of non-infectious disease cases. A moderate penalty of -1.0 is triggered when the case is ultimately ruled out as an infectious disease, but the system initially assessed it as medium or high risk. This indicates an over-warning by the system, which, while not as serious as underreporting, still results in wasted resources.

[0093] The policy gradient method is employed, which guides parameter updates by calculating the gradient of the objective function with respect to the model parameters. (Gradient) ,in, For the cumulative discount reward starting from time t, the parameters are updated as follows: ,in, For the new set of parameters to be optimized, For the old set of parameters to be optimized, This is the learning rate.

[0094] Taking the optimization of the department correction factor for Department A as an example, based on the historical infectious disease underreporting rate recorded in the department's risk profile table, Department A is initialized as a candidate that requires a higher risk assessment (i.e., initial γ>1). Backtracking analysis revealed that a large number of cases in Department A were rated as high-risk by the system, but doctors frequently excluded them, resulting in a low final diagnosis rate. 100 cases from Department A were extracted from historical data, and the average reward under the current strategy was calculated. Lower Mainly affected and Negative impact. Calculate the partial derivative of the loss function L with respect to γ. Due to excessively high risk leading to negative rewards, the gradient direction is to decrease γ. Parameter update: After several iterations, γ may converge to 1.1 or lower. After the update, the system's risk assessment of similar cases in Department A is reduced, the consistency with doctors' judgments is improved, and there is no increase in missed diagnoses, resulting in an overall increase in rewards.

[0095] In the fusion function In this study, α had a higher weight, significantly increasing risk during periods of resource scarcity. When beds were scarce, the system frequently issued high-risk alerts even for moderate-threat cases, triggering over-response and consuming infection control resources. Analysis of cases during periods of resource scarcity revealed… When the reward is low and the system's warning level is higher than the actual execution level, the reinforcement learning framework will try to reduce the weight of alpha when encountering resource constraints, observing whether it can achieve higher alpha without affecting the final outcome. Rewards. The system learns to balance disease threats with resource pressure, allowing resource pressure to have a significant impact only when the disease threat itself is high enough, thus avoiding the exclusion of high-risk slots simply due to resource issues.

[0096] For a newly emerging pathogen, the initial threat weight in the knowledge base is low. The first few cases of infection with this new pathogen are assessed as low or medium risk by the system due to their low threat weight. However, clinicians, based on professional vigilance, manually confirm these cases as high risk and trigger a response, which is subsequently proven to be correct. The system records (…). =Low / Medium =High, The contradictory data of =1) generates a strong negative reward signal, especially and The optimization algorithm significantly increases the weight of newly emerging pathogen-related disease threats in the knowledge base, as well as the contribution of related symptoms and test features to threat calculation, based on these high-reward samples (negative values ​​mean a large gradient needs to be changed). When encountering cases with similar characteristics later, the system can quickly provide a high-risk assessment, enabling rapid learning and response to emerging outbreaks, even before official guidelines are updated.

[0097] In this way, the system can gradually adapt to the hospital's unique diagnostic and treatment habits, resource conditions, and disease spectrum, achieving an intelligent evolution that becomes more accurate with use.

[0098] This invention represents a leap from static early warning to dynamic intelligent decision-making, encompassing both precise and dynamic risk assessment. Through a dynamic risk decision engine, it integrates real-time epidemic control conditions (such as resource scarcity) with individual case characteristics for comprehensive calculation, transforming the traditional static judgment model based on fixed rules or single keywords. This allows risk levels to accurately reflect the dynamic balance between disease threat and current control capabilities, resulting in more scientific and precise early warnings and effectively reducing false alarms and missed reports. Furthermore, it includes intelligent and user-friendly decision support, proactively providing clinicians with decision support information including disease type, quantified risk level, and clinical guidelines. This simplifies complex coding selection and reporting judgments into clear risk prompts and point-and-click operations. This significantly reduces the cognitive load and operational time of physicians, solving the problem of low reporting efficiency caused by traditional systems that only provide reminders without assistance.

[0099] This invention automates and closes the response process, automatically converting physician confirmation into a series of executable response instructions, driving the hospital system to execute differentiated prevention and control procedures. This completely changes the previous fragmented model of manual reminders, communication, and implementation, achieving second-level automatic response and significantly shortening the time window from risk detection to prevention and control initiation. It also enables tiered and precise prevention and control measures, automatically matching different levels of prevention and control instructions based on dynamic risk levels, such as from simple reporting to bed locking, infection control notification, and simultaneous disease control, avoiding waste or inadequacy of prevention and control resources. This ensures that limited infection control resources are prioritized and accurately allocated to the highest-risk cases and processes, improving prevention and control efficiency and safety.

[0100] This invention constructs a data-driven continuous optimization system, which includes intelligent management and traceability. The system can automatically trace back high-risk unresponsive cases and generate traceability reports, enabling management departments to shift from a crude mode of full-scale manual screening to a data-based precise verification mode, greatly improving management efficiency and targeting. It also includes the system's self-evolution capability. By collecting feedback data on decision-making results and prevention and control effects, the system can iteratively optimize its core assessment logic, enabling its risk judgment capability to continuously improve with actual operational data. It has a self-learning ability that becomes more accurate with use, fundamentally solving the problem of traditional system rules being rigid and difficult to adapt to new outbreaks or localized scenarios.

[0101] This invention strengthens the closed-loop linkage capability of medical and disease prevention collaboration, realizing the integration and linkage of internal and external data: by simultaneously receiving signals from the macro-level epidemic monitoring platform and in-hospital diagnosis and treatment data, hospitals are no longer information silos. Hospitals can act as intelligent nodes, quickly responding to higher-level early warnings and synchronizing internal risks externally, truly achieving vertical integration and horizontal collaboration of monitoring-early warning-decision-response within the region, thus improving the resilience and efficiency of overall infectious disease prevention and control in the region.

[0102] Example 2 A computer device 700, such as Figure 7 As shown, the system includes a memory 710, a processor 720, and a computer program 730 stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a method for intelligent hierarchical decision-making and automatic response within an infectious disease hospital. For a detailed description of the method, please refer to the corresponding description in the above method embodiments; it will not be repeated here.

[0103] Example 3 A computer-readable storage medium, such as Figure 8 As shown, a computer program is stored thereon. When executed by a processor, the computer program implements the steps of an intelligent hierarchical decision-making and automatic response method for infectious disease hospitals. For a detailed description of the method, please refer to the corresponding description in the above method embodiments, which will not be repeated here.

[0104] The number of devices and processing scale described herein are for the purpose of simplifying the description of the invention. Applications, modifications, and variations of the invention will be readily apparent to those skilled in the art.

[0105] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

[0106] The apparatus, computer device, and non-volatile computer storage medium and method provided in the embodiments of this specification are corresponding. Therefore, the apparatus, computer device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, computer device, and non-volatile computer storage medium will not be repeated here.

[0107] Those skilled in the art will also know that, besides implementing the controller in the form of purely computer-readable program code, the same functions can be achieved by logically programming the method steps, making the controller take the form of logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers (PLCs), and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the devices included within it for implementing various functions can also be considered structures within that hardware component. Alternatively, the devices for implementing various functions can be considered as both software units implementing the method and structures within a hardware component.

[0108] The systems, apparatuses, or units described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. For ease of description, the above apparatuses are described separately as various units based on their functions. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0109] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0110] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1A device that provides the functions specified in one or more boxes.

[0111] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0112] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0113] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0114] This specification may be described in the general context of computer-executable instructions, such as program units, that are executed by a computer. Generally, program units include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification may also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program units may reside in local and remote computer storage media, including storage devices.

[0115] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0116] The above description is merely an embodiment of this specification and is not intended to limit the scope of one or more embodiments of this specification. Various modifications and variations can be made to one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of one or more embodiments of this specification.

Claims

1. A method for intelligent hierarchical decision and automatic response in an infectious hospital, characterized in that, The method comprises the following steps: obtaining a clinical feature data set of a target case, the data set being derived from a macro epidemic monitoring platform and / or an in-hospital real-time diagnosis and treatment business system; inputting the clinical feature data set into a dynamic risk decision engine, and outputting at least one suspected infectious disease and a dynamic risk level associated therewith for the case; wherein the dynamic risk level is dynamically calculated and generated by the engine based on real-time prevention and control situation data and the clinical feature data; generating and pushing decision assistance information comprising the disease, its dynamic risk level and clinical guidelines to a clinical terminal; in response to a confirmation operation on the decision assistance information, automatically generating and triggering a response operation instruction matched with the risk level according to the selected disease and its risk level, to drive the execution of a corresponding differentiated prevention and control process.

2. A method for intelligent hierarchical decision and automatic response in an infectious hospital according to claim 1, characterized in that, The dynamic risk decision engine calculates the dynamic risk level by the following way: based on real-time prevention and control resource data, calculating a prevention and control pressure coefficient representing the current in-hospital prevention and control urgency and resource carrying capacity, the prevention and control resource data including the isolation ward bed vacancy rate of the infectious department, the inventory level of specific protective materials and the dispatchable state of the infection control personnel; based on the clinical feature data and the pre-stored infectious disease knowledge graph, calculating a disease threat coefficient representing the potential transmission and harm influence of the case, the infectious disease knowledge graph being associatedly stored with the transmission route, typical symptoms and abnormal test indicators of different infectious diseases; according to a pre-set fusion algorithm, comprehensively calculating the prevention and control pressure coefficient and the disease threat coefficient, outputting a quantitative risk value, and mapping it to a pre-defined risk level label.

3. A method for intelligent hierarchical decision and automatic response in an infectious hospital according to claim 2, characterized in that, The prevention and control pressure coefficient is calculated by the following steps: from the prevention and control resource data, respectively extracting the isolation ward bed vacancy rate of the infectious department, the inventory level of specific protective materials and the dispatchable state of the infection control personnel as corresponding resource indicators; normalizing each resource indicator Rx to obtain a tension sub-score Sx of the indicator; wherein an ideal state threshold To, a warning threshold Tw and a severe shortage threshold Tc are pre-set for each resource indicator Rx, and the actual value of Rx is mapped to a tension sub-score Sx in the interval [0, 1] through a segmented S-shaped function; assigning a dynamic weight to each resource index ; wherein, , a static weight of the resource index preset, a current hospital special diagnosis and treatment load coefficient calculated based on the number of patients of a specific type of outpatient service, a load sensitivity factor, 0≤ ≤0.3; The stress sub-score Sx of each resource index is multiplied by its dynamic weight The preliminary weighted sum is obtained by weighted aggregation And through the convex function transformation Nonlinear amplification is performed to obtain the prevention and control pressure coefficient; wherein, τ is a pressure amplification index, 0.5 < τ < 0.

8.

4. A method for intelligent hierarchical decision and automatic response in an infectious hospital according to claim 2, characterized in that, The disease threat coefficient is calculated by the following steps: matching the clinical feature data with the infectious disease knowledge graph to form an evidence set composed of etiological confirmation evidence, clinical high-priority evidence, clinical general-priority evidence and epidemiological association evidence; The collective confidence of each evidence set is calculated by using the confidence attenuation aggregation algorithm ; wherein, for an evidence set containing multiple evidence items, the collective confidence , of the evidence set is wherein, is the confidence of the ith evidence item in the evidence set, λ is the cooperation gain coefficient, and Π is the continuous multiplication symbol. According to the preset basic contribution weight, the collective confidence of each evidence set is weighted and synthesized to obtain a basic threat degree ; wherein, , , , , are respectively preset basic contribution weights for etiology confirmation evidence, clinical high suggestion evidence, clinical general suggestion evidence and epidemiological correlation evidence, and satisfy , , , , are respectively collective confidences of etiology confirmation evidence, clinical high suggestion evidence, clinical general suggestion evidence and epidemiological correlation evidence, is an inhibition factor based on negative or contradictory evidence; Based on the current urgency coefficient U of the case and the risk perception coefficient of the source department , the base threat degree is gain-corrected to obtain the corrected threat degree ; wherein, , and are the corrected intensity coefficients of urgency and department respectively; The modified threat degree is subjected to a saturation function nonlinear calibration is performed, and a final disease threat coefficient β is output; wherein, , and ξ is a saturation rate parameter greater than zero.

5. A method for intelligent hierarchical decision and automatic response in an infectious hospital according to claim 2, wherein, The fusion algorithm is configured to: when the prevention and control stress coefficient increases, the risk value corresponding to the same disease threat coefficient is amplified; specifically, the calculation formula of the risk value R is: wherein, a is the prevention and control stress coefficient, β is the disease threat coefficient, and k is a tension amplification coefficient greater than zero.

6. A method for intelligent hierarchical decision and automatic response in an infectious hospital according to claim 2, wherein, The calculation of the dynamic risk level also introduces a department risk correction factor based on the case source department; wherein the method further comprises the following steps: after calculating the risk value R, querying a pre-stored department risk profile table to obtain the historical infectious disease misreporting rate of the target case source department; the department risk profile table records the historical infectious disease misreporting rates of various departments in the hospital; determining whether the obtained historical infectious disease misreporting rate is greater than a pre-set misreporting rate threshold; if yes, assigning a department correction factor γ greater than 1 to the case, otherwise setting the department correction factor γ = 1; multiplying the risk value R by the department correction factor γ to obtain a corrected risk value; The final risk level label is mapped according to the revised risk value.

7. A method for intelligent hierarchical decision and automatic response in an infectious hospital according to claim 1, characterized in that, The decision assistance information is presented on the clinical terminal in a visual interface, wherein different risk levels of suspected infectious disease categories are highlighted in differentiated colors, icons or sorting methods.

8. A method for intelligent hierarchical decision and automatic response in an infectious hospital according to claim 1, characterized in that, The differentiated prevention and control process includes at least three response levels from low to high, and the response operation instructions are automatically matched to the corresponding response level according to the selected risk level, wherein: The instruction combination of the lowest response level at least includes generating a standard report document; The instruction combination of the intermediate response level adds sending a collaborative treatment reminder to a preset internal collaborative role on the basis of the lowest response level; The instruction combination of the highest response level further adds the instruction of synchronizing information to a preset external collaborative data interface and at least one clinical process intervention on the case on the basis of the intermediate response level.

9. A method for intelligent hierarchical decision making and automatic response in an infectious hospital as claimed in claim 8, wherein, The clinical process intervention includes: through interface calling, applying a process locking instruction to the target case to limit the advancement of part or all of its non-urgent clinical business processes.

10. The intelligent hierarchical decision and automatic response method in an infectious hospital according to claim 1, characterized in that, It also includes a backtracking analysis step: Periodically call the historical output records of the dynamic risk decision engine; Identify abnormal case records that are evaluated as high risk level but have not associated with effective response operation instructions within a preset time; Integrate the abnormal case records to generate a high-risk false alarm source report and push it to the management terminal.

11. A method for intelligent hierarchical decision and automatic response in an infectious hospital according to claim 1, characterized in that, It also includes an engine optimization step: Collect the historical decision records of the dynamic risk decision engine, the confirmation operation results of the clinical terminal and the execution effect data of the triggered prevention and control process to form an optimization feedback set; Based on the optimization feedback set, the dynamic risk decision engine is iteratively optimized in evaluation logic.

12. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to realize the steps of the method of any one of claims 1-11.

13. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the method of any one of claims 1-11.

Images