A method and system for intelligent recommendation of laboratory test items based on patient diagnosis and treatment information

Abstract

This invention relates to the field of medical intelligent technology and discloses a method and system for intelligent recommendation of laboratory tests based on patient diagnosis and treatment information. The method includes: extracting clinical events from time-series diagnosis and treatment information to obtain standardized clinical events; determining the medical logical relationships of the standardized clinical events and labeling causal, temporal, and co-occurrence relationships to obtain labeled clinical events; constructing a disease evolution network with labeled clinical events as nodes and medical logical relationships as edges; analyzing the topology of the disease evolution network to obtain critical paths and hub nodes, and resolving the hub nodes as core clinical diagnosis and treatment intentions; associating and matching the core clinical diagnosis and treatment intentions with a pre-constructed dynamic correlation graph of test-intentions to obtain test item combinations; and ranking the test item combinations according to the clinical path fit based on the current diagnosis and treatment stage and individual characteristics to obtain a personalized test item recommendation list. This invention can improve the efficiency of test item recommendation.

Description

Technical Field

[0001] This invention relates to the field of medical intelligent technology, and in particular to a method and system for intelligent recommendation of test items based on patient diagnosis and treatment information. Background Technology

[0002] Existing technologies for recommending laboratory tests lack the ability to systematically integrate time-series clinical data when processing patient medical information. They struggle to effectively combine structured medical information with unstructured free text information, leading to information bias, redundancy, or omissions during clinical event extraction. This makes it difficult to accurately capture key information in the progression of a patient's condition, resulting in laboratory test recommendations that cannot be based on the complete logic of disease evolution. Consequently, the recommendations fail to meet the actual medical needs of patients, and the overall accuracy and relevance of the recommendations are insufficient.

[0003] Existing technologies have failed to establish an effective mechanism for linking medical logical relationships between clinical events, lack the construction and analysis of disease evolution networks, and are unable to identify key pathways and core treatment intentions in disease development. In the recommendation process, they often ignore individual differences in patient characteristics and dynamic changes in the current stage of treatment, resulting in low alignment between test items and clinical pathways. This not only reduces the efficiency of the treatment process but may also lead to waste of medical resources or omission of necessary test items due to inappropriate recommendations, affecting the scientific and rational nature of treatment decisions. Therefore, how to improve the efficiency of test item recommendations has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a method and system for intelligent recommendation of laboratory test items based on patient diagnosis and treatment information, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a method for intelligent recommendation of laboratory test items based on patient diagnosis and treatment information, comprising: S1. Collect the medical records of the target patient generated within the time window into time-series medical information, and extract clinical events from the time-series medical information to obtain the standardized clinical events of the target patient. S2. Perform medical logic relationship determination on the standardized clinical events, and mark the causal, temporal and co-occurrence relationships between the standardized clinical events to obtain the labeled clinical events of the target patient; S3. Using the labeled clinical events as nodes and the medical logical relationships contained in the labeled clinical events as edges, construct the disease evolution network of the target patient; S4. Analyze the topology of the disease evolution network to obtain the critical path and hub node of the disease evolution network, and resolve the hub node as the core clinical diagnosis and treatment intention of the target patient; S5. Match the core clinical diagnosis and treatment intention with the pre-constructed dynamic correlation map of test-intention to obtain the test item combination for the target patient; S6. Based on the current treatment stage and individual characteristics of the target patient, the combination of test items is ranked according to clinical pathway fit to obtain a personalized test item recommendation list for the target patient.

[0006] In a preferred embodiment, the collection of medical records generated by the target patient within a time window is termed time-series medical information, and clinical events are extracted from the time-series medical information to obtain standardized clinical events for the target patient, including: Within the time window, receive the medical records of the target patient to obtain the time-series medical information of the target patient; The time-series medical information is divided according to its source and format to obtain a structured medical information stream and an unstructured free text information stream for the target patient. The unstructured free text information stream is subjected to deep text parsing based on medical natural language processing to identify text fragments containing symptom descriptions, changes in physical signs, medical procedures, and drug usage. The text fragments are encoded with standard clinical terminology, and the structured diagnosis and treatment information flow is simultaneously formatted to obtain the primary clinical events of the target patient. The primary clinical events are deduplicated, and the deduplicated clinical events are then conflict-resolved to obtain the standardized clinical events of the target patient.

[0007] In a preferred embodiment, the step of performing medical logical relationship determination on the standardized clinical events and labeling the causal, temporal, and co-occurrence relationships among the standardized clinical events to obtain the labeled clinical events of the target patient includes: The standardized clinical events are arranged chronologically to obtain the clinical timeline of the target patient; Based on the temporal proximity and continuity of the standardized clinical events, the temporal relationships between the standardized clinical events are marked; By analyzing the semantic content and type of events on the clinical timeline, the causal logical relationships of the standardized clinical events are derived. Traverse the clinical timeline and label the relationships between recurring events in the standardized clinical events as the co-occurrence relationships of the standardized clinical events; By integrating the temporal relationships, the causal logical relationships, and the co-occurrence relationships, a relationship description of the standardized clinical events is obtained. The relationship description and the standardized clinical events are associated and stored to obtain the labeled clinical events of the target patient.

[0008] In a preferred embodiment, constructing the disease evolution network of the target patient, using the labeled clinical events as nodes and the medical logical relationships contained in the labeled clinical events as edges, includes: Based on the occurrence time and unique identifier of the clinical events in the labeled clinical events, the clinical events are mapped to nodes, and the event type and corresponding timestamp of the clinical events are assigned as attributes to the nodes; The medical logical relationships contained in the labeled clinical events are used as edges, and the relationship type of the medical logical relationships is used as the weight of the edges; The nodes and edges are structurally integrated to obtain the initial disease evolution network of the target patient; The initial disease evolution network is topologically optimized to obtain the disease evolution network of the target patient.

[0009] In a preferred embodiment, the step of topology optimization of the initial disease evolution network to obtain the disease evolution network of the target patient includes: Analyze the connectivity of the initial disease evolution network, remove isolated nodes from the initial disease evolution network, and obtain the optimized node set of the initial disease evolution network; Redundant paths in the initial disease evolution network are examined and merged to obtain an optimized path set for the initial disease evolution network. Based on the optimized node set and the optimized path set, the initial disease evolution network is refined to obtain the disease evolution network of the target patient.

[0010] In a preferred embodiment, analyzing the topology of the disease progression network to obtain the critical paths and hub nodes of the disease progression network, and resolving the hub nodes as the core clinical treatment intentions of the target patient, includes: The connection relationships of nodes in the disease evolution network are analyzed, and the association density and position adjustment of the nodes are evaluated to obtain the structural feature set of the nodes; By integrating the structural feature set and the medical attributes of the clinical events corresponding to the nodes, and comprehensively evaluating the nodes, the hub nodes of the disease evolution network are obtained. In the disease evolution network, the path connecting the hub node is taken as the critical path of the disease evolution network; Based on the semantics of the clinical events represented by the hub nodes, their roles in the critical paths, and the clinical context constituted by the hub nodes, the core clinical diagnostic and treatment intentions of the target patient are determined.

[0011] In a preferred embodiment, the step of associating and matching the core clinical diagnostic and treatment intent with a pre-constructed dynamic correlation map of test-intent to obtain the test item combination for the target patient includes: Logical relationships are identified among the core clinical diagnostic and treatment intentions to determine the types of logical relationships between them. Based on the logical relationship type, composite path exploration is performed in the pre-constructed dynamic association graph of test-intention to obtain the test recommendation path for the target patient; Extract the test items on the recommended test path, and assign path dependency weights and intent coverage attributes to the test items according to the core clinical diagnosis and treatment intent; By integrating the test items, the path dependency weights, and the intent coverage attributes, candidate test items for the target patient are obtained. The candidate test items are cross-validated with the critical path information in the disease evolution network to obtain the test item combination for the target patient.

[0012] In a preferred embodiment, the step of ranking the combination of laboratory tests based on clinical pathway fit according to the current stage of diagnosis and treatment and individual characteristics of the target patient to obtain a personalized list of recommended laboratory tests for the target patient includes: The overlap of the disease spectrum coverage between the key clinical nodes of the current diagnosis and treatment stage in the target patient and the combination of the test items is quantitatively evaluated to obtain the stage dynamic correction coefficient of the target patient. The physiological indicators of individual characteristics of the target patients and the physiological condition range of the combination of test items are compared in multiple dimensions, and the comparison results are normalized to obtain the individual characteristic fit coefficient of the target patients. Extract the connection paths between the pre-constructed test-intention dynamic association map and the core clinical diagnosis and treatment intention, and determine the path consistency factor of the test item combination on the connection path; The comprehensive fit evaluation value of the test item combination is calculated based on the stage dynamic correction coefficient, the individual feature adaptation coefficient, and the path consistency factor. Based on the comprehensive fit evaluation value, the test item combinations are sorted in descending order to obtain a personalized test item recommendation list for the target patient.

[0013] In a preferred embodiment, the formula for calculating the comprehensive compatibility evaluation value is as follows: ; in, This represents the overall fit evaluation value. This represents the preset basic value coefficient. This represents the dynamic correction coefficient for the aforementioned stage. This represents the individual characteristic fitting coefficient. This represents the path consistency factor. This indicates the time interval of the combination of test items relative to the current time. This represents the preset time decay factor. This represents the natural exponential function.

[0014] To address the above problems, the present invention also provides an intelligent recommendation system for laboratory tests based on patient medical information, the system comprising: The clinical event processing module is used to collect the diagnosis and treatment records of the target patient within a time window as time-series diagnosis and treatment information, and to extract clinical events from the time-series diagnosis and treatment information to obtain the standardized clinical events of the target patient. The logical relationship discrimination module is used to perform medical logical relationship discrimination on the standardized clinical events and to label the causal, temporal and co-occurrence relationships between the standardized clinical events to obtain the labeled clinical events of the target patient. The association network construction module is used to construct the disease evolution network of the target patient using the labeled clinical events as nodes and the medical logical relationships contained in the labeled clinical events as edges. The core intent analysis module is used to analyze the topology of the disease evolution network, obtain the critical path and hub node of the disease evolution network, and parse the hub node into the core clinical diagnosis and treatment intent of the target patient. The test item matching module is used to match the core clinical diagnosis and treatment intention with a pre-constructed dynamic correlation map of test-intention to obtain the test item combination for the target patient; The personalized list creation module is used to rank the combination of test items according to their clinical pathway fit based on the current stage of diagnosis and treatment and individual characteristics of the target patient, so as to obtain a personalized test item recommendation list for the target patient.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This technology aggregates the time-series diagnosis and treatment information of target patients within a time window, and constructs a disease evolution network with clinical events as nodes and logical relationships as edges through clinical event extraction, standardization processing, and medical logical relationship annotation. It can accurately identify key paths and hub nodes in the network, efficiently analyze the core clinical diagnosis and treatment intentions of target patients, ensure the relevance of test item recommendations from the source, and significantly improve the accuracy and overall efficiency of test item recommendations.

[0016] 2. Based on a pre-constructed dynamic correlation map of test-intention, this technology achieves precise matching between core clinical diagnostic and treatment intentions and test items. At the same time, combined with the patient's current treatment stage and individual characteristics, it ranks the clinical pathway fit through multi-dimensional coefficient calculation and generates a personalized recommendation list. This not only ensures a high degree of fit between test items and treatment needs, but also fully adapts to individual patient differences, further enhancing the scientific nature and practical value of test item recommendations. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating an intelligent recommendation method for laboratory tests based on patient medical information, provided in an embodiment of the present invention. Figure 2 This is a functional module diagram of an intelligent recommendation system for laboratory test items based on patient diagnosis and treatment information, provided in an embodiment of the present invention. The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides a method for intelligently recommending laboratory tests based on patient medical information. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method for intelligently recommending laboratory tests based on patient medical information can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0020] Reference Figure 1The diagram shown is a flowchart illustrating a method for intelligently recommending laboratory tests based on patient medical information, according to an embodiment of the present invention. In this embodiment, the method includes: S1. Collect the medical records of the target patient generated within the time window into time-series medical information, and extract clinical events from the time-series medical information to obtain the standardized clinical events of the target patient. In this embodiment of the invention, the collection of medical records generated by the target patient within a time window is time-series medical information, and clinical events are extracted from the time-series medical information to obtain standardized clinical events of the target patient, including: Within the time window, receive the medical records of the target patient to obtain the time-series medical information of the target patient; The time-series medical information is divided according to its source and format to obtain a structured medical information stream and an unstructured free text information stream for the target patient. The unstructured free text information stream is subjected to deep text parsing based on medical natural language processing to identify text fragments containing symptom descriptions, changes in physical signs, medical procedures, and drug usage. The text fragments are encoded with standard clinical terminology, and the structured diagnosis and treatment information flow is simultaneously formatted to obtain the primary clinical events of the target patient. The primary clinical events are deduplicated, and the deduplicated clinical events are then conflict-resolved to obtain the standardized clinical events of the target patient.

[0021] Within a preset timeframe—which is the entire period from the start of the target patient's current visit to the current treatment stage—the hospital information system comprehensively collects all types of medical records generated by the patient in all scenarios, including outpatient visits, hospitalizations, examinations, treatment procedures, and medications. This includes electronic medical records, laboratory test reports, imaging examination conclusions, medical order execution sheets, medication prescription sheets, nursing records, and all other relevant information. These records are then organized and archived sequentially according to the specific time sequence in which they were generated, forming a time-series medical information system arranged in an orderly manner along the time dimension.

[0022] Based on the source system identifiers and data presentation formats of time-series medical information, data is classified and categorized. Information sources include laboratory data management systems, medical order execution systems, electronic medical record writing systems, and drug prescription systems. Data formats are divided into two categories: those with fixed field structures and those without. Each piece of data in the time-series medical information is checked one by one for its source system and format characteristics. Information from laboratory data management systems, medical order execution systems, etc., that contains clearly preset fields such as examination item names, test result values, drug generic names, drug dosages, and drug frequencies, and whose data structures are uniform and fixed, is classified as structured medical information flow. Information from electronic medical record writing systems, such as medical records, symptom descriptions, and disease analysis opinions freely entered by doctors, that does not have fixed field divisions and is mainly expressed in natural language, is classified as unstructured free text information flow.

[0023] Based on medical natural language processing technology, the unstructured free text information flow is deeply analyzed. First, the text is segmented and identified word by word according to the medical professional vocabulary database, which includes all categories of medical terms such as symptoms, signs, medical procedures, and drugs. Then, through the preset medical grammar rules and semantic logic relation database, the collocation and semantic orientation between the segmented words are analyzed to accurately select continuous text content containing specific symptom descriptions such as persistent cough, limb numbness, fever, etc.; changes in signs such as increased heart rate, stable blood pressure, weight loss, etc.; medical procedures such as coronary angiography, appendectomy, intravenous puncture, etc.; and drug usage such as oral cefuroxime axetil, subcutaneous injection of low molecular weight heparin, etc. These screened and confirmed text contents are the required text fragments.

[0024] Using internationally recognized standard clinical terminology systems, such as the SNOMED CT terminology set, as the coding basis, the core medical concepts in each identified text fragment, such as acute appendicitis, routine blood tests, and ibuprofen, are precisely matched with their corresponding terms in the standard terminology system. Each text fragment is assigned a unique standard clinical term code to ensure the standardized expression of text information. At the same time, all data in the structured diagnosis and treatment information flow is standardized according to the unified data format specifications. Fields with the same meaning from different source systems are uniformly named, such as unifying blood glucose test results and serum glucose values ​​as "serum glucose test results". Data formats such as dates and numerical units are uniformly calibrated, with dates uniformly formatted as year-month-day hour:minute:second, length units uniformly formatted as centimeters, and weight units uniformly formatted as kilograms. The text fragments coded with standard clinical terms are integrated with the structured diagnosis and treatment information after the format is unified. Each independent integrated unit is the primary clinical event of the target patient.

[0025] All primary clinical events are comprehensively compared and deduplicated. The core information of each event, including event type, standard clinical terminology code, occurrence time, and key diagnostic parameters, is used as the basis for judgment. If multiple primary clinical events have completely identical core information, only the event with the earliest occurrence time is retained, and the remaining duplicate events are deleted. For clinical events with contradictory core information after deduplication, such as two different blood pressure readings recorded at the same time point, conflict resolution is performed according to a preset priority rule. The priority rule is based on the credibility of the information source. Test results from core laboratories of tertiary hospitals have higher priority than those from community medical institutions. Test results containing the model of the testing instrument, the qualifications of the operator, and calibration records have higher priority than test results without relevant auxiliary information. The event with the highest priority is selected as the valid event, and contradictory low-priority events are eliminated. The clinical events remaining after deduplication and conflict resolution are the standardized clinical events for the target patient.

[0026] The beneficial effects are that by comprehensively collecting patients' full-scene diagnosis and treatment records and sorting them by time, the structured and unstructured information is accurately separated. With the help of medical natural language processing technology, the text is deeply analyzed and standardized and encoded. The structured data format is unified in a synchronous manner. After deduplication and conflict resolution, the comprehensiveness, standardization and accuracy of standardized clinical events are ensured. This provides high-quality and reliable basic data support for subsequent medical logical relationship identification and construction of disease evolution network.

[0027] S2. Perform medical logic relationship determination on the standardized clinical events, and mark the causal, temporal and co-occurrence relationships between the standardized clinical events to obtain the labeled clinical events of the target patient; In this embodiment of the invention, the step of performing medical logical relationship discrimination on the standardized clinical events and labeling the causal, temporal, and co-occurrence relationships among the standardized clinical events to obtain the labeled clinical events of the target patient includes: The standardized clinical events are arranged chronologically to obtain the clinical timeline of the target patient; Based on the temporal proximity and continuity of the standardized clinical events, the temporal relationships between the standardized clinical events are marked; By analyzing the semantic content and type of events on the clinical timeline, the causal logical relationships of the standardized clinical events are derived. Traverse the clinical timeline and label the relationships between recurring events in the standardized clinical events as the co-occurrence relationships of the standardized clinical events; By integrating the temporal relationships, the causal logical relationships, and the co-occurrence relationships, a relationship description of the standardized clinical events is obtained. The relationship description and the standardized clinical events are associated and stored to obtain the labeled clinical events of the target patient.

[0028] All standardized clinical events were checked one by one, and the timestamps corresponding to each event were extracted. All standardized clinical events were arranged and integrated in chronological order from earliest to latest according to the timestamps, forming a coherent clinical timeline with a clear time dimension. The position of each standardized clinical event on this timeline is completely consistent with the actual order of occurrence.

[0029] Define clear criteria for determining the temporal proximity of events: the time interval between two standardized clinical events must not exceed a reasonable duration set in clinical practice. Simultaneously, confirm temporal continuity by ensuring events are sequentially connected on the clinical timeline without significant logical interruptions. For each pair of events on the clinical timeline, determine whether they meet the temporal proximity or continuity requirements. For events that meet the criteria, clearly indicate their chronological order, such as event M occurring before event N, or events P and Q occurring simultaneously. Completely record the temporal relationships between all standardized clinical events.

[0030] The semantic core content of each standardized clinical event on the clinical timeline is extracted one by one to clarify the specific medical meaning expressed by the event, such as the symptom manifestation of symptom events, the disease conclusion of diagnostic events, and the specific means of treatment events. At the same time, the type classification of each event is determined. Combining clinical practice guidelines and medical common sense, the inherent logical relationship between different types of events is analyzed. If the occurrence of one event directly leads to the occurrence or change of another event, such as the occurrence of an anti-infective treatment event after an infection symptom event, or the occurrence of a symptom relief event after a treatment event, then a causal logical relationship is determined between the two events. The causal logical relationship of all standardized clinical events is completely deduced and recorded.

[0031] Following the chronological order of the clinical timeline, each standardized clinical event is reviewed and verified one by one. The core characteristics of each event are recorded, including event type, standard clinical terminology code, and core medical information. When two or more events are found to have completely identical core characteristics, and these events are not consecutively adjacent on the clinical timeline, the relationship between these events is explicitly marked as a co-occurrence relationship. This ensures that all recurring standardized clinical events can be accurately identified and their co-occurrence relationships are marked.

[0032] The previously labeled temporal relationships, derived causal logical relationships, and identified co-occurrence relationships are comprehensively integrated. For each group of related standardized clinical events, a standardized and unified medical expression method is used to systematically describe the various relationships between them, forming a complete and clear description of the relationships between standardized clinical events, ensuring that each relationship is accurately represented without omission.

[0033] A structured, associated storage mechanism is established to bind each standardized clinical event to its corresponding relationship description, ensuring that the relevant relationship information of each event can be accurately linked. A data storage format that allows for rapid retrieval and parsing is used, making the association between standardized clinical events and relationship descriptions traceable and retrievable. The complete dataset formed after associated storage constitutes the labeled clinical events for the target patient.

[0034] The beneficial effects are that, through standardized temporal arrangement, accurate relationship derivation and annotation, systematic relationship integration and associated storage, the causal, temporal and co-occurrence relationships between standardized clinical events are comprehensively and accurately identified and clarified, forming a complete set of annotated clinical events. This provides a logically clear and data-reliable foundation for constructing a disease evolution network with events as nodes and relationships as edges, ensuring the accuracy of subsequent disease analysis and test item recommendations.

[0035] S3. Using the labeled clinical events as nodes and the medical logical relationships contained in the labeled clinical events as edges, construct the disease evolution network of the target patient; In this embodiment of the invention, constructing the disease evolution network of the target patient using the labeled clinical events as nodes and the medical logical relationships contained in the labeled clinical events as edges includes: Based on the occurrence time and unique identifier of the clinical events in the labeled clinical events, the clinical events are mapped to nodes, and the event type and corresponding timestamp of the clinical events are assigned as attributes to the nodes; The medical logical relationships contained in the labeled clinical events are used as edges, and the relationship type of the medical logical relationships is used as the weight of the edges; The nodes and edges are structurally integrated to obtain the initial disease evolution network of the target patient; The initial disease evolution network is topologically optimized to obtain the disease evolution network of the target patient.

[0036] The process of performing topology optimization on the initial disease evolution network to obtain the disease evolution network of the target patient includes: Analyze the connectivity of the initial disease evolution network, remove isolated nodes from the initial disease evolution network, and obtain the optimized node set of the initial disease evolution network; Redundant paths in the initial disease evolution network are examined and merged to obtain an optimized path set for the initial disease evolution network. Based on the optimized node set and the optimized path set, the initial disease evolution network is refined to obtain the disease evolution network of the target patient.

[0037] Extract the specific occurrence time and unique identifier assigned by the system from each labeled clinical event. Through a data mapping mechanism, each labeled clinical event is transformed into an independent node in the disease evolution network. At the same time, the event type corresponding to the clinical event, such as symptom event, diagnosis event, treatment event, etc., and the corresponding timestamp accurate to the minute and second, are bound and stored as inherent attributes of the node, so that each node has a clear identity and core information.

[0038] The three types of medical logical relationships—causality, temporality, and co-occurrence—that are clearly identified in the labeled clinical events are directly used as edges connecting the nodes. Fixed weights are assigned to different relationship types based on the importance of the medical diagnosis and treatment logic, with causal relationships corresponding to the highest weight, temporal relationships to the medium weight, and co-occurrence relationships to the basic weight. These weight values ​​are explicitly assigned to the corresponding edges, so that each edge has a distinguishable association strength attribute.

[0039] By adopting a network structured modeling approach, nodes with attributes are arranged in the network space according to their timestamp order. Then, the relevant nodes are precisely connected according to the edge relationships. At the same time, the relationship status between nodes and edges is presented through visualization technology, and the weight of the edges is intuitively reflected by the thickness of the lines or the depth of the colors, forming a complete and clear initial disease evolution network of the target patient.

[0040] A comprehensive connectivity test is performed on the initial disease evolution network. Each node is checked to see if it has at least one edge connecting it to other nodes. If a node is not effectively associated with any other node, it is determined to be an isolated node and is removed directly from the initial network. After filtering, an optimized node set of the initial disease evolution network containing all effectively associated nodes is obtained.

[0041] Traverse all paths in the initial disease evolution network, identify multiple paths with the same starting and ending points and identical node sequences and edge relationship types. These paths are redundant paths. Merge all redundant paths into one path, retain the attribute information of the edge with the highest weight, and delete the remaining duplicate paths to finally form the optimized path set of the initial disease evolution network.

[0042] Based on the optimized node set as the basic framework, the edges in the optimized path set are precisely deployed to the corresponding nodes according to the node association relationship. The network structure is adjusted and sorted out as a whole, and residual invalid connections and duplicate elements are eliminated. This makes the distribution of nodes in the network more reasonable and the association of edges clearer. Finally, a compact and logically rigorous target patient's disease evolution network is formed.

[0043] The beneficial effects are that by clarifying node attributes, assigning weights to edges, and performing structured integration, an initial disease evolution network with complete information is constructed. Then, through topology optimization operations such as connectivity analysis and redundant path merging, invalid elements are eliminated and the network structure is simplified, ensuring the accuracy, simplicity, and logic of the disease evolution network. This provides high-quality network model support for subsequent analysis of network topology and the discovery of critical paths and hub nodes.

[0044] S4. Analyze the topology of the disease evolution network to obtain the critical path and hub node of the disease evolution network, and resolve the hub node as the core clinical diagnosis and treatment intention of the target patient; In this embodiment of the invention, the step of analyzing the topology of the disease evolution network, obtaining the critical paths and hub nodes of the disease evolution network, and resolving the hub nodes as the core clinical treatment intentions of the target patient includes: The connection relationships of nodes in the disease evolution network are analyzed, and the association density and position adjustment of the nodes are evaluated to obtain the structural feature set of the nodes; By integrating the structural feature set and the medical attributes of the clinical events corresponding to the nodes, and comprehensively evaluating the nodes, the hub nodes of the disease evolution network are obtained. In the disease evolution network, the path connecting the hub node is taken as the critical path of the disease evolution network; Based on the semantics of the clinical events represented by the hub nodes, their roles in the critical paths, and the clinical context constituted by the hub nodes, the core clinical diagnostic and treatment intentions of the target patient are determined.

[0045] A comprehensive analysis of the connections of all nodes in the disease progression network was conducted. The total number of connections established between each node and other nodes was counted one by one. The relationship type of the edge corresponding to each connection was clarified. At the same time, the association density of each node was calculated, which is the proportion of the number of different nodes connected to that node to the total number of nodes in the network. The positional attribute of each node in the network was determined, i.e., whether it is at the intersection of multiple different paths or only participates in a single path. The information such as the number of connections, association density, and positional attribute of each node was systematically organized to form a structural feature set of nodes containing the core structural information of all nodes.

[0046] The medical attributes of each node's corresponding clinical events are collected, including the specific type of clinical event (e.g., diagnostic event, symptom event, treatment event, follow-up event), the level of medical importance (e.g., core diagnostic event, auxiliary examination event), and the degree of correlation with disease diagnosis and treatment. These medical attributes are then integrated one-to-one with the node's structural feature set. According to preset evaluation rules, the structural features and medical attributes of each node are comprehensively considered, and nodes with a large number of connections, high correlation density, location at the intersection of paths, high level of medical importance, and close correlation with disease diagnosis and treatment are prioritized. These nodes are the hub nodes of the disease evolution network.

[0047] Traverse all paths in the disease evolution network, and check each path one by one whether it contains at least one hub node. Paths that start or end with a hub node or use a hub node as an intermediate connecting node are all included in the candidate path range. The candidate paths are deduplicated and sorted to ensure that each path can reflect the key context of disease development. The final candidate paths are the critical paths of the disease evolution network.

[0048] Each hub node is analyzed to reveal the specific semantics of the clinical event it represents, clarifying its core meaning in medical diagnosis and treatment. The role of each hub node in the critical path is also analyzed – whether it serves as the initial trigger, intermediate turning point, or final target node in the disease progression. By combining the distribution and interrelationships of all hub nodes, a holistic clinical scenario for the target patient is constructed. By integrating the semantic expression of the hub nodes, their functional roles in the critical path, and the resulting clinical scenario, content reflecting the core diagnostic and treatment needs of the target patient is extracted; this content represents the target patient's core clinical diagnostic and treatment intent.

[0049] The beneficial effects include the accurate screening of hub nodes by systematically analyzing network topology, integrating structural features and medical attributes, clarifying the critical paths connecting hub nodes, and deeply analyzing core clinical diagnostic and treatment intentions by combining multi-dimensional information. This ensures the accuracy and relevance of core clinical diagnostic and treatment intentions, provides clear and definite directional guidance for the precise matching of subsequent core clinical diagnostic and treatment intentions with test items, and guarantees the scientificity and rationality of test item recommendations.

[0050] S5. Match the core clinical diagnosis and treatment intention with the pre-constructed dynamic correlation map of test-intention to obtain the test item combination for the target patient; In this embodiment of the invention, the step of associating and matching the core clinical diagnosis and treatment intent with a pre-constructed dynamic correlation map of test-intent to obtain the test item combination for the target patient includes: Logical relationships are identified among the core clinical diagnostic and treatment intentions to determine the types of logical relationships between them. Based on the logical relationship type, composite path exploration is performed in the pre-constructed dynamic association graph of test-intention to obtain the test recommendation path for the target patient; Extract the test items on the recommended test path, and assign path dependency weights and intent coverage attributes to the test items according to the core clinical diagnosis and treatment intent; By integrating the test items, the path dependency weights, and the intent coverage attributes, candidate test items for the target patient are obtained. The candidate test items are cross-validated with the critical path information in the disease evolution network to obtain the test item combination for the target patient.

[0051] The specific connotations of each core clinical diagnostic and treatment intention are analyzed one by one, clarifying the diagnostic and treatment goals and core needs that each intention points to. By comparing the semantic associations and diagnostic and treatment logic of different core clinical diagnostic and treatment intentions, the types of logical relationships between them are identified, including but not limited to parallel relationships, i.e., multiple intentions serve the same diagnostic and treatment goal at the same time; progressive relationships, i.e., the later intention is a deepening and extension of the previous intention; and primary and secondary relationships, i.e., one core intention is primary and other intentions are auxiliary and supportive. The types of logical relationships between all core clinical diagnostic and treatment intentions are fully recorded.

[0052] The pre-constructed dynamic correlation graph of test results and their intended uses contains a vast amount of clinically validated test results, various clinical diagnostic intentions, and the relationships between them. These relationships are established based on medical guidelines, clinical practice data, and expert experience. Based on the logical relationships between the identified core clinical diagnostic intentions, starting from the graph node corresponding to each core clinical diagnostic intention, multiple paths are explored along the correlation edges in the graph. Paths that are logically related to multiple core clinical diagnostic intentions are integrated to form composite paths that simultaneously cover all core clinical diagnostic intentions. These composite paths constitute the recommended test paths for the target patient.

[0053] Each recommended testing path is traversed, and all testing items included along the path are extracted, ensuring no omissions or duplications. Based on the importance of the core clinical diagnostic intent and the critical level of the recommended testing path, a path dependency weight is assigned to each testing item; the more critical the path and the more directly related to the core intent, the higher the weight value. Simultaneously, the number of core clinical diagnostic intents that each testing item can cover is counted, thereby determining the intent coverage attribute of that testing item. The more intents covered, the higher the intent coverage. Finally, testing items with path dependency weights and intent coverage attributes are obtained.

[0054] A structured integration framework is established to bind each test item to its corresponding path dependency weight and intent coverage attribute, ensuring that the attribute information of each test item is complete and accurately associated. All the bound information is systematically organized to form a complete data set containing test items, path dependency weights, and intent coverage attributes. This set is the candidate test items for the target patient.

[0055] Clinical event information contained in the critical pathways of the disease progression network is extracted, including event type, core medical semantics, and time sequence relationships. Candidate test items are compared and verified one by one with this critical pathway information to check whether each candidate test item has a reasonable medical association with the clinical events on the critical pathway and whether it can provide effective diagnostic support or efficacy evaluation basis for the disease progression reflected by the critical pathway. Candidate test items that are not related to the critical pathway information or have unreasonable associations are eliminated. All test items that pass cross-validation are retained and integrated to finally form the test item combination for the target patient.

[0056] The beneficial effects are that by accurately identifying the logical relationships of core clinical diagnosis and treatment intentions, exploring composite paths in a pre-constructed association graph, screening candidate test items by combining weight and coverage attributes, and then cross-validating with key path information, the combination of test items is ensured to be highly consistent with the core diagnosis and treatment intentions and the evolution of the disease. This provides a scientific and reliable basis for subsequent personalized ranking of test items and ensures the pertinence and effectiveness of test item recommendations.

[0057] S6. Based on the current treatment stage and individual characteristics of the target patient, the combination of test items is ranked according to clinical pathway fit to obtain a personalized test item recommendation list for the target patient.

[0058] In this embodiment of the invention, the step of ranking the combination of test items according to clinical pathway fit based on the current treatment stage and individual characteristics of the target patient to obtain a personalized test item recommendation list for the target patient includes: The overlap of the disease spectrum coverage between the key clinical nodes of the current diagnosis and treatment stage in the target patient and the combination of the test items is quantitatively evaluated to obtain the stage dynamic correction coefficient of the target patient. The physiological indicators of individual characteristics of the target patients and the physiological condition range of the combination of test items are compared in multiple dimensions, and the comparison results are normalized to obtain the individual characteristic fit coefficient of the target patients. Extract the connection paths between the pre-constructed test-intention dynamic association map and the core clinical diagnosis and treatment intention, and determine the path consistency factor of the test item combination on the connection path; The comprehensive fit evaluation value of the test item combination is calculated based on the stage dynamic correction coefficient, the individual feature adaptation coefficient, and the path consistency factor. Based on the comprehensive fit evaluation value, the test item combinations are sorted in descending order to obtain a personalized test item recommendation list for the target patient.

[0059] The formula for calculating the comprehensive compatibility evaluation value is as follows: ; in, This represents the overall fit evaluation value. This represents the preset basic value coefficient. This represents the dynamic correction coefficient for the aforementioned stage. This represents the individual characteristic fitting coefficient. This represents the path consistency factor. This indicates the time interval of the combination of test items relative to the current time. This represents the preset time decay factor. This represents the natural exponential function.

[0060] Identify the current stage of diagnosis and treatment for the target patient, such as the initial diagnosis stage, treatment implementation stage, efficacy evaluation stage, and rehabilitation follow-up stage. Extract the key clinical nodes that are decisive within this stage, such as the etiology screening node in the initial diagnosis stage, the medication adjustment node in the treatment implementation stage, and the indicator monitoring node in the efficacy evaluation stage. Analyze the disease spectrum coverage corresponding to the combination of laboratory tests, clarifying the types of diseases and treatment stages that each test can assist in diagnosing or monitoring. By statistically analyzing the number of overlapping nodes between the key clinical nodes and the disease spectrum coverage of the test combination, calculate the proportion of overlapping nodes to the total number of key clinical nodes to complete the quantitative assessment. This proportion is the dynamic adjustment coefficient for the target patient's stage.

[0061] Collect physiological indicators related to the individual characteristics of the target patient, including age, gender, height, weight, baseline blood pressure, baseline blood glucose, liver and kidney function indicators, cardiopulmonary function indicators, and past medical history. Define the preset physiological condition range for each test item in the test combination, i.e., the applicable physiological indicator interval standard for each test item. Compare each physiological indicator of the target patient with the corresponding physiological condition range of the test item one by one, count the number of physiological indicator dimensions that meet the physiological condition range for each test item, and then normalize the statistical results according to a fixed numerical range standard, mapping the results to a value between 0 and 1. This normalized value is the individual characteristic fit coefficient of the target patient.

[0062] From the pre-constructed dynamic correlation graph of test-intention, all connected paths directly related to the core clinical diagnostic and treatment intentions are precisely extracted. These paths are correspondence chains between test items and diagnostic and treatment intentions formed based on medical diagnostic and treatment logic and clinical practice verification. Each test item in the test item combination is checked one by one to confirm whether it is a valid component of the extracted connected path. The number of test items in the test item combination that are on the connected path is counted, and the proportion of this number to the total number of test item combinations is calculated. This proportion is the path consistency factor of the test item combination on the connected path.

[0063] A preset baseline value coefficient is determined, which is a fixed value set based on the clinically general importance of the test items. First, the baseline value coefficient is multiplied by the stage dynamic correction coefficient, and then the result is multiplied by the individual characteristic fit coefficient to form an intermediate product. This intermediate product is then raised to the power corresponding to the path consistency factor to obtain the exponentiation result. The time interval between the test item combination and the current time is calculated, i.e., the time difference between the time when the test item was recommended and the optimal implementation time of the test item. This time difference is divided by a preset time decay factor to obtain the time correlation ratio. The negative value of this ratio is then used for natural exponentiation to obtain the time decay adjustment result. The exponentiation result is multiplied by the time decay adjustment result, and the final value is the comprehensive fit evaluation value of the test item combination.

[0064] The preset base value coefficient is a fixed coefficient set in advance. The stage dynamic correction coefficient is generated by quantitatively assessing the overlap of the disease spectrum coverage of the key clinical nodes and test item combinations corresponding to the current diagnosis and treatment stage of the target patient. The individual characteristic adaptation coefficient is obtained by comparing the physiological indicators of the target patient's individual characteristics and the physiological condition range of the test item combinations in multiple dimensions, and then normalizing the comparison results. The path consistency factor is obtained by extracting the connecting paths with the core clinical diagnosis and treatment intention in the pre-constructed dynamic association map of test-intention, and then determining the corresponding factor of the test item combination on the connecting path. The time interval of the test item combination relative to the current time is obtained by directly collecting the interval data between the corresponding time of the test item combination and the current time. The preset time decay factor is a fixed decay coefficient set in advance.

[0065] This calculation process comprehensively considers the preset basic value, the suitability of the target patient's current treatment stage with the test combination, the degree of matching between the patient's individual characteristics and the test combination, the degree of fit between the test combination and the relevant pathway of the core clinical treatment intention, and the impact of the time interval between the test combination and the current time. It comprehensively quantifies the degree of fit between the test combination and the target patient's clinical pathway. The final result can accurately reflect the suitability of the test combination for the target patient and provide a reliable basis for the subsequent descending order of test combinations.

[0066] When the degree of adaptation of the diagnosis and treatment stage corresponding to the stage dynamic correction coefficient increases, the comprehensive fit evaluation value will increase accordingly. When the degree of matching of individual characteristics corresponding to the individual characteristic adaptation coefficient increases, the comprehensive fit evaluation value will increase accordingly. When the degree of path fit corresponding to the path consistency factor increases, the comprehensive fit evaluation value will increase synchronously. When the time interval between the test item combination and the current time increases, the calculation result of the natural exponential function part will decrease, resulting in a decrease in the comprehensive fit evaluation value. The preset basic value coefficient and the preset time decay factor are fixed values ​​and only participate in the calculation as fixed parameters during the calculation process. They will not affect the trend of the comprehensive fit evaluation value.

[0067] The comprehensive fit evaluation value of each test item in the test item combination is systematically organized. All test items are arranged in descending order of evaluation value, with the test item with the highest evaluation value at the top, and the subsequent test items are sorted in descending order of evaluation value. This ensures that the sorting result can intuitively reflect the degree of fit between each test item and the patient's diagnosis and treatment needs. The complete test item sequence formed after sorting is the personalized test item recommendation list for the target patient.

[0068] The beneficial effects are as follows: by quantitatively assessing the overlap between the treatment stage and the disease spectrum, comparing the physiological conditions of individual physiological indicators and test items from multiple dimensions, determining the consistency between test items and the path of the correlation map, comprehensively considering multiple key factors to calculate the comprehensive fit evaluation value and sorting them in descending order, the final personalized test item recommendation list not only fits the patient's current treatment progress, but also adapts to individual physiological characteristics and conforms to the logic of medical diagnosis and treatment, significantly improving the personalization, accuracy and clinical applicability of test item recommendations, and providing a scientific and reliable reference for clinical diagnosis and treatment decisions.

[0069] like Figure 2 The diagram shown is a functional module diagram of an intelligent recommendation system for laboratory test items based on patient diagnosis and treatment information provided in an embodiment of the present invention.

[0070] The intelligent recommendation system 100 for laboratory tests based on patient medical information described in this invention can be installed in an electronic device. Depending on the functions implemented, the intelligent recommendation system 100 may include a clinical event processing module 101, a logical relationship discrimination module 102, an association network construction module 103, a core intent analysis module 104, a laboratory test matching module 105, and a personalized list creation module 106. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.

[0071] In this embodiment, the functions of each module / unit are as follows: The clinical event processing module 101 is used to collect the diagnosis and treatment records of the target patient within the time window as time-series diagnosis and treatment information, and to extract clinical events from the time-series diagnosis and treatment information to obtain the standardized clinical events of the target patient. The logical relationship discrimination module 102 is used to perform medical logical relationship discrimination on the standardized clinical events and to mark the causal, temporal and co-occurrence relationships between the standardized clinical events to obtain the labeled clinical events of the target patient. The association network construction module 103 is used to construct the disease evolution network of the target patient using the labeled clinical events as nodes and the medical logical relationships contained in the labeled clinical events as edges. The core intent analysis module 104 is used to analyze the topology of the disease evolution network, obtain the critical path and hub node of the disease evolution network, and parse the hub node into the core clinical diagnosis and treatment intent of the target patient. The test item matching module 105 is used to match the core clinical diagnosis and treatment intention with the pre-constructed dynamic correlation map of test-intention to obtain the test item combination for the target patient. The personalized list generation module 106 is used to rank the combination of test items according to the clinical pathway fit based on the current treatment stage and individual characteristics of the target patient, so as to obtain a personalized test item recommendation list for the target patient.

[0072] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0073] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0074] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0075] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0076] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0077] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for intelligent recommendation of test items based on patient diagnosis information, characterized in that, The method comprises: S1, collecting the diagnosis and treatment records of the target patient generated within a time window as time-series diagnosis and treatment information, and performing clinical event extraction on the time-series diagnosis and treatment information to obtain standardized clinical events of the target patient; S2, performing medical logical relationship discrimination on the standardized clinical events, and labeling the cause-and-effect, time sequence and co-occurrence relationships between the standardized clinical events to obtain labeled clinical events of the target patient; S3, constructing a disease evolution network of the target patient by taking the labeled clinical events as nodes and the medical logical relationships contained in the labeled clinical events as edges; S4, analyzing the topological structure of the disease evolution network to obtain a key path and a pivot node of the disease evolution network, and parsing the pivot node into a core clinical diagnosis and treatment intention of the target patient; S5, associating and matching the core clinical diagnosis and treatment intention with a pre-constructed test-intention dynamic association graph to obtain a test item combination of the target patient; S6, performing clinical pathway fit degree sorting on the test item combination according to the current diagnosis and treatment stage and individual characteristics of the target patient to obtain a personalized test item recommendation list of the target patient.

2. The method of claim 1, wherein the patient diagnosis information is obtained from at least one of a patient diagnosis database, a patient diagnosis information database, and a patient diagnosis information server. The method comprises: S1, collecting the diagnosis and treatment records of the target patient generated within a time window as time-series diagnosis and treatment information, and performing clinical event extraction on the time-series diagnosis and treatment information to obtain standardized clinical events of the target patient; S2, performing medical logical relationship discrimination on the standardized clinical events, and labeling the cause-and-effect, time sequence and co-occurrence relationships between the standardized clinical events to obtain labeled clinical events of the target patient; S3, constructing a disease evolution network of the target patient by taking the labeled clinical events as nodes and the medical logical relationships contained in the labeled clinical events as edges; S4, analyzing the topological structure of the disease evolution network to obtain a key path and a pivot node of the disease evolution network, and parsing the pivot node into a core clinical diagnosis and treatment intention of the target patient; S5, associating and matching the core clinical diagnosis and treatment intention with a pre-constructed test-intention dynamic association graph to obtain a test item combination of the target patient; 3. The method of claim 1, wherein the patient diagnosis information is obtained from at least one of a patient diagnosis database, a patient diagnosis information database, and a patient diagnosis information server. S6, performing clinical pathway fit degree sorting on the test item combination according to the current diagnosis and treatment stage and individual characteristics of the target patient to obtain a personalized test item recommendation list of the target patient. The method comprises: S1, collecting the diagnosis and treatment records of the target patient generated within a time window as time-series diagnosis and treatment information, and performing clinical event extraction on the time-series diagnosis and treatment information to obtain standardized clinical events of the target patient; S2, performing medical logical relationship discrimination on the standardized clinical events, and labeling the cause-and-effect, time sequence and co-occurrence relationships between the standardized clinical events to obtain labeled clinical events of the target patient; S3, constructing a disease evolution network of the target patient by taking the labeled clinical events as nodes and the medical logical relationships contained in the labeled clinical events as edges; S4, analyzing the topological structure of the disease evolution network to obtain a key path and a pivot node of the disease evolution network, and parsing the pivot node into a core clinical diagnosis and treatment intention of the target patient; S5, associating and matching the core clinical diagnosis and treatment intention with a pre-constructed test-intention dynamic association graph to obtain a test item combination of the target patient; S6, performing clinical pathway fit degree sorting on the test item combination according to the current diagnosis and treatment stage and individual characteristics of the target patient to obtain a personalized test item recommendation list of the target patient. The relationship description and the standardized clinical events are associated and stored to obtain the labeled clinical events of the target patient.

4. The method of claim 1, wherein the patient diagnosis information is obtained from a patient diagnosis information database. The step of constructing the disease evolution network of the target patient, using the labeled clinical events as nodes and the medical logical relationships contained in the labeled clinical events as edges, includes: ​ Based on the occurrence time and unique identifier of the clinical events in the labeled clinical events, the clinical events are mapped to nodes, and the event type and corresponding timestamp of the clinical events are assigned as attributes to the nodes; The medical logical relationships contained in the labeled clinical events are used as edges, and the relationship type of the medical logical relationships is used as the weight of the edges; The nodes and edges are structurally integrated to obtain the initial disease evolution network of the target patient; The initial disease evolution network is topologically optimized to obtain the disease evolution network of the target patient.

5. The method of claim 4, wherein the patient diagnosis information is obtained from a patient diagnosis information database. The process of performing topology optimization on the initial disease evolution network to obtain the disease evolution network of the target patient includes: Analyze the connectivity of the initial disease evolution network, remove isolated nodes from the initial disease evolution network, and obtain the optimized node set of the initial disease evolution network; Redundant paths in the initial disease evolution network are examined and merged to obtain an optimized path set for the initial disease evolution network. Based on the optimized node set and the optimized path set, the initial disease evolution network is refined to obtain the disease evolution network of the target patient.

6. The method of claim 1, wherein the patient diagnosis information is obtained from a patient diagnosis information database. The analysis of the disease progression network's topology yields the critical paths and hub nodes, and the hub nodes are parsed as the core clinical treatment intentions of the target patient, including: The connection relationships of nodes in the disease evolution network are analyzed, and the association density and position adjustment of the nodes are evaluated to obtain the structural feature set of the nodes; By integrating the structural feature set and the medical attributes of the clinical events corresponding to the nodes, and comprehensively evaluating the nodes, the hub nodes of the disease evolution network are obtained. In the disease evolution network, the path connecting the hub node is taken as the critical path of the disease evolution network; Based on the semantics of the clinical events represented by the hub nodes, their roles in the critical paths, and the clinical context constituted by the hub nodes, the core clinical diagnostic and treatment intentions of the target patient are determined.

7. The intelligent recommendation method for laboratory tests based on patient diagnosis and treatment information as described in claim 1, characterized in that, The process of associating and matching the core clinical diagnostic and treatment intent with a pre-constructed dynamic correlation map of test-intent to obtain the test item combination for the target patient includes: Logical relationships are identified among the core clinical diagnostic and treatment intentions to determine the types of logical relationships between them. Based on the logical relationship type, composite path exploration is performed in the pre-constructed dynamic association graph of test-intention to obtain the test recommendation path for the target patient; Extract the test items on the recommended test path, and assign path dependency weights and intent coverage attributes to the test items according to the core clinical diagnosis and treatment intent; By integrating the test items, the path dependency weights, and the intent coverage attributes, candidate test items for the target patient are obtained. The candidate test items are cross-validated with the critical path information in the disease evolution network to obtain the test item combination for the target patient.

8. The method of claim 1, wherein the patient diagnosis information is obtained from a patient diagnosis information database. The process involves ranking the combination of laboratory tests based on clinical pathway fit according to the current stage of diagnosis and treatment and individual characteristics of the target patient, resulting in a personalized list of recommended laboratory tests for the target patient, including: ​ The overlap of the disease spectrum coverage between the key clinical nodes of the current diagnosis and treatment stage in the target patient and the combination of the test items is quantitatively evaluated to obtain the stage dynamic correction coefficient of the target patient. The physiological indicators of individual characteristics of the target patients and the physiological condition range of the combination of test items are compared in multiple dimensions, and the comparison results are normalized to obtain the individual characteristic fit coefficient of the target patients. Extract the connection paths between the pre-constructed test-intention dynamic association map and the core clinical diagnosis and treatment intention, and determine the path consistency factor of the test item combination on the connection path; The comprehensive fit evaluation value of the test item combination is calculated based on the stage dynamic correction coefficient, the individual feature adaptation coefficient, and the path consistency factor. Based on the comprehensive fit evaluation value, the test item combinations are sorted in descending order to obtain a personalized test item recommendation list for the target patient.

9. The method of claim 8, wherein the patient diagnosis information is obtained from a patient diagnosis information database. The formula for calculating the comprehensive compatibility evaluation value is as follows: ​ ； wherein, represents the comprehensive fitness evaluation value, represents a preset base value coefficient, represents the stage dynamic correction coefficient, represents the individual characteristic adaptation coefficient, represents the path consistency factor, represents a time interval of the test item combination relative to the current time, represents a preset time decay factor, represents a natural exponential function. 10.A test item intelligent recommendation system based on patient diagnosis information, characterized in that, The system for implementing the intelligent recommendation method for laboratory tests based on patient diagnosis and treatment information as described in claim 1 includes: The clinical event processing module is used to collect the diagnosis and treatment records of the target patient within a time window as time-series diagnosis and treatment information, and to extract clinical events from the time-series diagnosis and treatment information to obtain the standardized clinical events of the target patient. The logical relationship discrimination module is used to perform medical logical relationship discrimination on the standardized clinical events and to label the causal, temporal and co-occurrence relationships between the standardized clinical events to obtain the labeled clinical events of the target patient. The association network construction module is used to construct the disease evolution network of the target patient using the labeled clinical events as nodes and the medical logical relationships contained in the labeled clinical events as edges. The core intent analysis module is used to analyze the topology of the disease evolution network, obtain the critical path and hub node of the disease evolution network, and parse the hub node into the core clinical diagnosis and treatment intent of the target patient. The test item matching module is used to match the core clinical diagnosis and treatment intention with a pre-constructed dynamic correlation map of test-intention to obtain the test item combination for the target patient; The personalized list creation module is used to rank the combination of test items according to their clinical pathway fit based on the current stage of diagnosis and treatment and individual characteristics of the target patient, so as to obtain a personalized test item recommendation list for the target patient.

Images