Cross-department diagnosis and treatment path determination method and system, medium and electronic equipment

Abstract

The invention provides a cross-department diagnosis and treatment path determination method and system, a medium and electronic equipment, and relates to the field of data processing. The method comprises the following steps: responding to the condition information of a patient, and converting the condition information into a plurality of first condition tags; a decision tree is obtained, the decision tree comprises a plurality of first child nodes and a first leaf node, and the first child nodes represent a first diagnosis judgment condition based on the illness state information of the patient; the first leaf node represents diagnosis and treatment path decisions among different departments; and matching the plurality of first illness state labels with a first child node and a first leaf node in the decision tree to determine a first diagnosis path.

Description

Technical Field

[0001] This application relates to the field of data processing, and in particular to a method, system, medium, and electronic device for determining cross-departmental diagnostic and treatment pathways. Background Technology

[0002] In the clinical practice of large general hospitals, patients often need to undergo collaborative diagnosis and treatment from multiple departments during their first visit. However, because the information provided by patients is usually incomplete or vague, it is difficult for medical staff to quickly and accurately determine the appropriate cross-departmental diagnosis and treatment path. Summary of the Invention

[0003] In view of this, this application provides a method, system, medium and electronic device for determining cross-departmental diagnosis and treatment pathways.

[0004] One aspect of this application provides a method for determining cross-departmental treatment pathways, including:

[0005] The system receives the patient's condition information and converts it into multiple primary condition tags.

[0006] A decision tree is obtained, which includes multiple first child nodes and first leaf nodes. The first child nodes represent first diagnostic judgment conditions based on the patient's condition information; the first leaf nodes represent treatment path decisions between different departments.

[0007] The multiple first disease labels are matched with the first child node and the first leaf node in the decision tree to determine the first diagnostic path.

[0008] According to an embodiment of this application, the first child node includes a node condition function, which represents the judgment of the first disease label; the first leaf node includes a node decision function, which represents the diagnosis and treatment decision for the target department corresponding to the first leaf node.

[0009] According to an embodiment of this application, matching the plurality of first disease labels with the first child node and the first leaf node in the decision tree to determine the first diagnostic path includes:

[0010] For any of the first child nodes, the plurality of first disease labels are judged based on the node condition function of the first child node to obtain the Boolean judgment result of the plurality of first disease labels;

[0011] Based on the Boolean judgment result, calculate the first matching score between the first child node and the plurality of first disease labels;

[0012] Based on the first matching score of the first child node, the second matching score of each first leaf node is calculated by chain traversal of the decision tree.

[0013] The first leaf node with the highest second matching score is identified as the target leaf node;

[0014] Multiple target departments are determined by the node decision function of the target leaf node;

[0015] The first diagnostic pathway is determined based on the multiple target departments.

[0016] According to an embodiment of this application, the first leaf node further includes a resource load factor; the method further includes:

[0017] Obtain the resource load status of the target department corresponding to the first leaf node;

[0018] Calculate the resource load factor based on the resource load situation;

[0019] Calculate the comprehensive matching score of the first leaf node based on the first matching score of the first leaf node and the resource load factor;

[0020] The first leaf node with the highest overall matching score is identified as the target leaf node.

[0021] According to an embodiment of this application, the method further includes:

[0022] In response to changes in the patient's condition information, the changed condition information is converted into multiple secondary condition tags;

[0023] The multiple second disease labels are matched with the second child nodes and the second leaf nodes in the decision tree to update the first diagnostic path and obtain the second diagnostic path.

[0024] According to an embodiment of this application, the method further includes:

[0025] Acquire historical case data, which includes the disease tags of multiple historical patients and the corresponding treatment pathway results;

[0026] Based on the historical case data, calculate the error rate and complexity cost of each of the first sub-nodes;

[0027] Based on the error rate and complexity cost, determine the redundant first child node;

[0028] Remove the redundant first child node from the decision tree.

[0029] According to an embodiment of this application, the method further includes:

[0030] Calculate the conditional similarity between the first child nodes;

[0031] At least two first child nodes whose conditional similarity is greater than the threshold are identified as similar node pairs;

[0032] Merge the first child nodes in the similar node pairs to obtain the merged node;

[0033] The decision tree is updated based on the merged node.

[0034] Another aspect of this application provides a cross-departmental diagnostic and treatment pathway determination system, comprising:

[0035] The diagnosis and treatment information acquisition module is used to obtain the patient's first diagnosis and treatment information during the historical diagnosis and treatment process and the second diagnosis and treatment information during the current diagnosis and treatment process.

[0036] The disease information acquisition module is used to respond to the acquisition of the patient's disease information and convert the disease information into multiple first disease tags;

[0037] A decision tree acquisition module is used to acquire a decision tree, which includes multiple first child nodes and first leaf nodes. The first child nodes represent first diagnostic judgment conditions based on the patient's condition information; the first leaf nodes represent treatment path decisions between different departments.

[0038] The diagnostic path determination module is used to match the plurality of first disease labels with the first child node and the first leaf node in the decision tree to determine the first diagnostic path.

[0039] Another aspect of this application provides an electronic device comprising:

[0040] One or more processors;

[0041] Memory, used to store one or more programs.

[0042] When the one or more programs are executed by the one or more processors, the one or more processors implement the method described above.

[0043] Another aspect of this application provides a computer-readable storage medium storing computer-executable instructions that, when executed, are used to implement the method described above.

[0044] By employing the embodiments of this application, the problem of incomplete or ambiguous patient condition information is effectively solved by converting patient condition information into multiple first condition tags. This transforms previously difficult-to-process unstructured condition descriptions into standardized and quantifiable data. Based on a decision tree structure containing multiple first child nodes and first leaf nodes, the first diagnostic judgment conditions represented by the first child nodes can systematically handle various condition features, while the treatment path decisions represented by the first leaf nodes directly correspond to collaborative solutions between different departments, thus constructing a complete treatment decision support system. By matching multiple first condition tags with the first child nodes and first leaf nodes in the decision tree to determine the first diagnostic path, cross-departmental treatment paths can be quickly and accurately formulated during the patient's initial visit. This reduces the uncertainty of treatment paths caused by ambiguous condition information, avoids frequent path adjustments and repeated referrals, thereby improving the utilization efficiency of medical resources and enhancing the patient's medical experience. Attached Figure Description

[0045] The above and other objects, features and advantages of this application will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0046] Figure 1 The flowchart illustrating a cross-departmental diagnosis and treatment pathway determination method provided in this application is shown in the illustration.

[0047] Figure 2 The diagram illustrates the structural block of a cross-departmental diagnosis and treatment pathway determination system provided in this application.

[0048] Figure 3 A schematic block diagram of an electronic device provided in this application is shown. Detailed Implementation

[0049] The embodiments of this application will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of this application. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of this application for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0050] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0051] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0052] When using expressions such as "at least one of A, B, and C", they should generally be interpreted in accordance with the meaning that is commonly understood by a person skilled in the art (e.g., "a system having at least one of A, B, and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B, and C, etc.).

[0053] Figure 1 The flowchart illustrates a method for determining a cross-departmental treatment pathway provided in an embodiment of this application.

[0054] like Figure 1 As shown, the method for determining the cross-departmental diagnosis and treatment pathway may specifically include steps S101 to S104.

[0055] Step S101: Obtain the patient's condition information and convert the condition information into multiple primary condition tags;

[0056] Step S102: Obtain a decision tree. The decision tree includes multiple first child nodes and first leaf nodes. The first child nodes represent the first diagnostic judgment conditions based on the patient's condition information. The first leaf nodes represent the treatment path decisions between different departments.

[0057] Step S103: Match multiple first disease labels with the first child node and the first leaf node in the decision tree to determine the first diagnostic path.

[0058] In step S101, the medical condition information refers to the descriptive information provided by the patient during their visit to a medical institution regarding their health status, symptoms, medical history, medication, and other medical-related information. In this embodiment, it can be understood as a collection of subjective symptom descriptions, objective physical signs records, basic physiological indicator measurement results, and related background information presented by the patient to medical staff during their initial visit, which is used to provide basic data support for determining the subsequent diagnosis and treatment path.

[0059] For example, medical information includes, but is not limited to, the patient's chief complaint of symptoms (such as chest pain, dizziness, fever, etc.), duration of symptoms, degree of pain, location of the symptom, past medical history, family medical history, current medication, vital signs data (blood pressure, heart rate, body temperature, etc.), and preliminary examination results.

[0060] Similarly, the first disease label refers to a standardized, structured data identifier that is converted from the patient's unstructured or semi-structured disease information according to a preset medical classification standard and data specification. In the embodiments of this application, it can be understood as a disease feature description unit with a unified format, clear semantics and quantifiable features, which is used to support the subsequent decision tree matching and path determination process.

[0061] For example, the first symptom label can be a standardized label with a key-value pair structure, such as "symptom type - chest pain", "duration - less than 1 hour", "pain level - moderate", "past medical history - hypertension", "age range - 50-65 years old", "gender - male", "blood pressure status - high".

[0062] In one feasible implementation, the response to obtain the patient's condition information includes the patient's subjective symptom description and objective examination results being entered into the electronic medical record system by medical staff. The system automatically receives and stores this raw condition information, while simultaneously triggering a condition information tagging and conversion process.

[0063] Specifically, the system first performs text preprocessing on the received medical information, including removing irrelevant characters, standardizing terminology, and correcting common input errors. Then, based on a pre-set medical dictionary and symptom classification system, it extracts key medical terms from the medical description and generates corresponding first medical condition tags according to standardized tag naming rules.

[0064] In another feasible implementation, the process of converting medical condition information into multiple primary medical condition tags can be carried out using a semantic mapping method based on a medical ontology library. The system pre-constructs a medical tag system that includes multiple dimensions such as disease classification, symptom classification, drug classification, and examination item classification. When receiving patient medical condition information, the system identifies medical entities in the medical condition description through natural language processing technology and matches the medical entities with the pre-set tag system to generate corresponding standardized tags.

[0065] For example, when a patient describes "chest pain started last night, like being pricked with needles, and lasted for about an hour," the system extracts key information such as "chest pain," "prickly pain," and "lasted for an hour," and converts it into primary condition labels such as "symptom location - chest," "pain nature - stabbing pain," and "duration - 1 hour."

[0066] In another feasible implementation, the process of generating the first condition label also incorporates the automatic labeling of the patient's basic information and examination data. The system can automatically generate corresponding classification labels based on the patient's basic information such as age, gender, and weight, as well as the measurement results of physiological indicators such as blood pressure, heart rate, and blood sugar.

[0067] Specifically, the system presets normal ranges and grading standards for various physiological indicators. When the patient's examination data is obtained, the system automatically converts the numerical data into categorical labels. For example, the blood pressure value "150 / 90 mmHg" is converted into the label "Blood Pressure Status - Mild Hypertension", and the age "58 years old" is converted into the label "Age Range - Middle-aged and Elderly". This forms a comprehensive set of primary disease labels covering multiple dimensions such as symptom characteristics, physiological indicators, and basic information.

[0068] It should be noted that the generation process of the first disease label fully considers the diversity and complexity of medical expressions. The system supports the generation of multiple related labels for the same disease information to ensure the integrity of the information. At the same time, a label quality inspection mechanism has been established to check the logical consistency and medical rationality of the generated labels to ensure the accuracy and reliability of the labeling results.

[0069] In step S102, a decision tree refers to a rule-based tree-structured data model that uses hierarchical condition judgment nodes and decision output nodes to realize complex logical reasoning. In this embodiment, it can be understood as a dedicated logical reasoning model for determining medical treatment pathways, which is used to determine the most suitable cross-departmental treatment pathway based on the patient's standardized condition information through hierarchical judgment.

[0070] The first child node refers to the intermediate node in the decision tree that is responsible for executing the condition judgment logic. Each node contains specific medical judgment conditions and corresponding logical branches. In this embodiment, it can be understood as a rule-based judgment unit that carries the diagnostic experience of medical experts. It is used to perform structured logical evaluation of the patient's condition labels and guide the traversal direction of the decision tree based on the evaluation results.

[0071] For example, the first sub-node could be a "chest pain duration judgment node" (to determine whether the chest pain lasts longer than 30 minutes), an "ECG abnormality detection node" (to determine whether there are ST segment changes on the ECG), or a "past medical history assessment node" (to determine whether the patient has a history of hypertension or diabetes), etc.

[0072] Similarly, the first leaf node refers to the terminal output node of the decision tree, which no longer contains child nodes or branches. It is responsible for providing the final diagnosis and treatment path decision result. In this embodiment, it can be understood as a decision output unit that carries the cross-departmental diagnosis and treatment plan. It is used to determine the specific department sequence that the patient should go to and the corresponding diagnosis and treatment strategy based on the final path of the decision tree traversal.

[0073] For example, the first leaf node can be a treatment path node such as "Emergency Department → Cardiology Department → Cardiothoracic Surgery", "Internal Medicine Department → Gastroenterology Department → Hepatobiliary Surgery", or "Emergency Department → Neurology Department".

[0074] In one feasible implementation, the decision tree is constructed based on a medical expert knowledge base and clinical practice guidelines. Specifically, firstly, clinical practice guidelines and expert consensus from various departments are collected, and key diagnostic conditions and decision logic are extracted. Then, these diagnostic conditions are transformed into standardized judgment rules, with each rule corresponding to a first child node in the decision tree. Next, based on path recommendations in the practice guidelines, departmental combination schemes for the first leaf node are designed. Finally, the decision tree is verified and optimized by a team of medical experts to ensure its medical rationality and clinical applicability.

[0075] For example, in a decision tree for the diagnosis and treatment of patients with chest pain, key diagnostic elements such as "nature of chest pain", "duration of pain", "electrocardiogram manifestations" and "level of myocardial markers" can be set as the judgment conditions of the first sub-node based on the diagnosis and treatment guidelines for acute coronary syndrome.

[0076] In another feasible implementation, the decision tree is constructed using a multi-departmental collaborative design approach. Specifically, medical experts from multiple departments, including emergency medicine, internal medicine, surgery, and radiology, are organized to jointly design cross-departmental diagnostic and treatment decision-making logic through multiple rounds of expert meetings and case discussions. During the design process, experts from each department propose diagnostic conditions and referral criteria from their own professional perspectives, forming the judgment rules for the first sub-node. At the same time, the departments negotiate to determine the optimal departmental transfer path for different conditions, forming the path decision for the first leaf node. In addition, a regular decision tree review mechanism is established to continuously optimize the decision tree structure based on clinical practice feedback and the latest medical evidence.

[0077] For example, in the diagnosis and treatment decision tree for patients with abdominal pain, internal medicine experts are responsible for designing judgment nodes related to internal medicine diseases, surgical experts are responsible for designing judgment nodes related to surgical indications, and radiology experts are responsible for designing judgment nodes related to imaging examinations, jointly constructing a complete cross-departmental diagnosis and treatment decision system.

[0078] It is important to note that the construction of a decision tree must fully consider the hospital's actual situation and resource allocation to ensure the feasibility of the designed treatment pathways in clinical practice. Simultaneously, the decision tree should possess good scalability, allowing for flexible adjustments and updates based on changes in hospital departmental setup, the introduction of new technologies, and the emergence of new diseases. Furthermore, the design of the decision tree should adhere to the principles of evidence-based medicine, ensuring that each judgment condition and decision path is supported by sufficient medical evidence, avoiding the impact of subjective assumptions or experience biases on the quality of diagnosis and treatment.

[0079] In step S103, the system first starts from the root node of the decision tree, obtains the judgment conditions of the first child node corresponding to the root node, and compares the patient's multiple first condition labels with the judgment conditions. When a certain first condition label or a group of first condition labels meets the judgment conditions of the current first child node, the system continues to traverse downwards along the branch that meets the conditions, enters the first child node of the next level for subsequent judgment, and repeats this process until a certain first leaf node is reached. The diagnosis and treatment path decision corresponding to the first leaf node is the determined first diagnosis path.

[0080] For example, for patients with primary disease labels such as "symptom type - chest pain", "duration - more than 30 minutes", and "pain nature - squeezing pain", the system first determines whether chest pain symptoms exist at the root node. If the condition is met, it enters the "chest pain analysis" branch. Then, it determines the duration of pain. If the condition of "more than 30 minutes" is met, it enters the "acute chest pain" branch. Finally, it determines the nature of pain. If the condition of "squeezing pain" is met, it reaches the first leaf node of "emergency department → cardiology department → cardiac catheterization laboratory", thereby determining the primary diagnostic path.

[0081] In another feasible implementation, the logical matching process employs a multi-path parallel evaluation approach. Specifically, the system simultaneously calculates the matching degree between multiple first disease labels of the patient and all possible paths in the decision tree, calculating a matching score for each complete path from the root node to the first leaf node. The matching score is calculated based on the condition satisfaction of each first child node on the path. When the patient's first disease label fully satisfies the judgment conditions of a certain first child node, that node receives full marks; if it partially satisfies them, it receives a corresponding proportion of marks; and if it does not satisfy them at all, it receives zero marks. The system summarizes the scores of all first child nodes on each path to obtain the overall matching score for that path, and then selects the first leaf node corresponding to the path with the highest matching score as the final diagnosis and treatment path decision. For example, a patient's first disease label may simultaneously match "cardiovascular disease path" and "digestive system disease path." The system calculates the matching scores for these two paths separately. If the matching score for the cardiovascular disease path is 85 points and the matching score for the digestive system disease path is 60 points, then the cardiovascular disease path with the higher matching score is selected as the first diagnostic path.

[0082] By employing the embodiments of this application, the problem of incomplete or ambiguous patient condition information is effectively solved by converting patient condition information into multiple first condition tags. This transforms previously difficult-to-process unstructured condition descriptions into standardized and quantifiable data. Based on a decision tree structure containing multiple first child nodes and first leaf nodes, the first diagnostic judgment conditions represented by the first child nodes can systematically handle various condition features, while the treatment path decisions represented by the first leaf nodes directly correspond to collaborative solutions between different departments, thus constructing a complete treatment decision support system. By matching multiple first condition tags with the first child nodes and first leaf nodes in the decision tree to determine the first diagnostic path, cross-departmental treatment paths can be quickly and accurately formulated during the patient's initial visit. This reduces the uncertainty of treatment paths caused by ambiguous condition information, avoids frequent path adjustments and repeated referrals, thereby improving the utilization efficiency of medical resources and enhancing the patient's medical experience.

[0083] In actual medical diagnosis and treatment, although decision trees provide a basic tree structure framework, their nodes often lack clear logical processing mechanisms. This makes it difficult for the system to make accurate condition judgments and departmental decisions when faced with complex and diverse patient condition labels. In particular, when the patient's first condition label involves multiple medical dimensions or has symptom descriptions with ambiguous boundaries, the traditional simple node structure is difficult to provide sufficient judgment accuracy and is prone to misjudgment or omission of the diagnosis and treatment path.

[0084] To address the aforementioned issues, based on the above embodiments, as an optional embodiment, the first child node includes a node condition function, which represents the judgment of the first disease label; the first leaf node includes a node decision function, which represents the diagnosis and treatment decision for the target department corresponding to the first leaf node.

[0085] A node condition function refers to an algorithm module specifically designed for medical judgment. It receives the patient's first condition label and performs Boolean logic operations according to preset medical condition rules, outputting a clear judgment result indicating whether the node condition is met. In this embodiment, it can be understood as the core logic processing unit of the first child node of the decision tree, carrying specific medical diagnostic conditions and corresponding judgment logic. It is used to systematically match and logically verify the standardized condition labels of the patient, ensuring that the decision tree can perform accurate path traversal based on the specific characteristics of the patient's condition.

[0086] For example, node conditional functions include, but are not limited to, a chest pain duration judgment function (conditional logic to check whether the patient's chest pain lasts more than 30 minutes), an electrocardiogram abnormality detection function (conditional logic to determine whether there is ST segment elevation or depression based on electrocardiogram related labels), a vital sign assessment function (conditional logic to determine whether the patient is in a critical state based on labels such as blood pressure, heart rate, and body temperature), a medical history screening function (conditional logic to check whether the patient has a history of specific diseases such as hypertension, diabetes, and coronary heart disease), a symptom severity rating function (conditional logic to determine the severity level of symptoms based on labels such as pain level and functional impairment), and a drug allergy history check function (conditional logic to verify whether the patient has an allergic reaction to a specific drug category), etc.

[0087] Similarly, the node decision function refers to a decision algorithm module specifically designed to generate cross-departmental treatment pathways. Based on the complete path information from the decision tree traversal to the first leaf node and the patient's comprehensive condition characteristics, it uses preset medical decision rules and departmental collaboration logic to output specific target department combinations and corresponding treatment strategy arrangements. In this embodiment, it can be understood as the core decision output unit of the first leaf node, integrating the diagnostic and treatment experience of medical experts and the departmental resource allocation of the hospital to formulate personalized cross-departmental treatment pathways for patients, ensuring that patients can obtain the department sequence and treatment plan most suitable for their condition characteristics.

[0088] For example, the node decision function could be an acute cardiovascular event management decision function (outputting the departmental path "Emergency Department → Cardiology Department → Cardiac Catheterization Laboratory → CCU" and the treatment strategy "Immediate ECG monitoring, establish intravenous access, prepare for emergency PCI"), a gastrointestinal bleeding comprehensive management decision function (outputting the departmental path "Emergency Department → Gastroenterology Department → Endoscopy Center → General Surgery" and the treatment strategy "Neglecting food and water, fluid resuscitation and hemostasis, endoscopic examination, and surgery if necessary"), or a multiple trauma treatment decision function (outputting the path "Emergency Department → Orthopedics Department → Neurosurgery Department → Thoracic Surgery Department → I..."). The system includes departmental pathways for "CU" and treatment strategies such as "stable vital signs, imaging examinations, specialized departmental treatment, and intensive care," chronic disease management decision functions (outputting departmental pathways such as "internal medicine → specialist outpatient clinic → laboratory department → pharmacy department" and treatment strategies such as "condition assessment, professional medication adjustment, regular follow-up, and medication guidance"), and complex case consultation decision functions (outputting departmental pathways such as "related specialties → multidisciplinary consultation center → laboratory and imaging department → treatment department" and treatment strategies such as "preliminary diagnosis, expert consultation, auxiliary examinations, and treatment plan development").

[0089] By employing the embodiments of this application, the node condition function ensures that the decision tree can make judgments when processing the patient's first condition label, avoiding path errors caused by ambiguous conditions, and improving the system's ability to handle complex conditions. The node decision function guarantees that the final treatment path decision conforms to medical standards and meets the patient's individual needs, avoiding a one-size-fits-all approach to departmental allocation and achieving a more scientific and rational cross-departmental collaborative arrangement.

[0090] Furthermore, medical experts can adjust the judgment logic of the node condition function and the decision-making strategy of the node decision function based on feedback from clinical practice and the latest medical evidence. This ensures that the system always keeps pace with best practices in clinical diagnosis and treatment, thereby providing patients with more accurate and efficient cross-departmental diagnosis and treatment services.

[0091] Based on the above embodiments, as an optional implementation, step S103 may further include the following steps:

[0092] Step S201: For any first child node, judge multiple first disease labels based on the node condition function of the first child node to obtain the Boolean judgment results of multiple first disease labels;

[0093] Step S202: Based on the Boolean judgment result, calculate the first matching score between the first child node and multiple first disease labels;

[0094] Step S203: Based on the first matching score of the first child node, calculate the second matching score of each first leaf node through chain traversal of the decision tree;

[0095] Step S204: Determine the first leaf node with the highest second matching score as the target leaf node;

[0096] Step S205: Determine multiple target departments by using the node decision function of the target leaf node;

[0097] Step S206: Determine the first diagnostic pathway based on multiple target departments.

[0098] In step S201, the system passes multiple first condition labels of the patient as input parameters to the node condition function of the current first child node, and the node condition function identifies the target label subset related to the current judgment condition according to the preset medical judgment logic of the first child node.

[0099] For example, when the system processes the first sub-node of "Chest Pain Severity Assessment", the system will call the node condition function of the node to filter out relevant labels such as "symptom type", "pain intensity", "pain nature" and "duration" from the patient's first set of condition labels. Then the system will perform logical judgment processing on the value of each label through the node condition function.

[0100] During the specific execution of the label judgment, the system performs Boolean logic operations using conditional rule matching through node conditional functions. For each relevant first symptom label, the system calls the node conditional function to compare the label value with the preset judgment conditions. When the label value meets the condition requirements, the system records the judgment result as True; otherwise, it records it as False.

[0101] For example, for the "pain level" label, if the judgment condition of the first child node is "pain level reaches moderate or above", and the patient's first condition label shows "pain level - severe", then the system obtains the Boolean judgment result of the label as True through the logical operation of the node condition function; if the patient's label shows "pain level - mild", then the system obtains the Boolean judgment result as True.

[0102] Similarly, for the "symptom type" label, if the judgment condition is "symptom type is chest pain related" and the patient label is "symptom type - chest pain", the system outputs True; if it is "symptom type - abdominal pain", the system outputs Fail.

[0103] When processing combined logical judgments of multiple first-condition labels, the system also performs multi-label joint judgment processing based on the pre-defined medical diagnostic complexity requirements in the node condition functions. Specifically, the judgment conditions of some first-child nodes may need to simultaneously satisfy specific combinations of values ​​of multiple labels. In this case, the system performs logical AND or logical OR operations on the relevant labels through the node condition functions.

[0104] For example, when the system processes the first sub-node of "Acute Myocardial Infarction Risk Assessment", the system calls the conditional function of the node to check whether the patient has the label combination of "symptom type - chest pain", "duration - more than 30 minutes" and "pain nature - squeezing". Only when the system verifies that these three conditions are met at the same time through the node conditional function will the system record the Boolean result of the combination judgment as True.

[0105] In other cases, the system may use a logical OR operation through node condition functions. For example, when processing the first sub-node of "Digestive System Disease Screening", the system can accept that any one of the following conditions is met: "Symptom Type - Abdominal Pain", "Symptom Type - Nausea and Vomiting", or "Symptom Type - Diarrhea".

[0106] In step S202, the first matching score refers to the Boolean judgment result of the node condition function of the first child node on the patient's first condition label. It is a numerical index that evaluates the degree of consistency between the node judgment condition and the patient's condition characteristics through quantitative calculation methods. In this embodiment, it can be understood as a comprehensive evaluation score that reflects the matching degree between the medical judgment logic of the first child node and the patient's actual condition label.

[0107] Specifically, the system calculates the initial matching degree of the first child node by counting the total number of first condition labels involved in the node condition function of the current first child node as the denominator, and the number of labels with a Boolean judgment result of True as the numerator. The system determines the initial matching degree of the first child node by calculating the proportion of labels that meet the conditions. For example, if a certain "acute chest pain assessment" first child node involves the judgment of 5 first condition labels, namely symptom type, pain intensity, duration, pain nature, and accompanying symptoms, and the Boolean judgment result of the first 4 labels is True and the last one is False, then the system calculates that the basic matching degree of the node is 80%, indicating that the patient's condition characteristics have a high degree of consistency with the judgment conditions of the node.

[0108] In one feasible implementation, the system also considers the differences in importance of different first disease labels in medical diagnosis when calculating the first matching score, and optimizes the accuracy of the matching score by introducing label weight coefficients.

[0109] In another feasible implementation, the system will also dynamically adjust the first matching score based on the confidence and completeness of the first disease label to address the label quality differences that may occur in clinical practice.

[0110] In step S203, the second matching score refers to the comprehensive evaluation index of the degree of matching between the complete treatment path and the patient's overall condition characteristics by comprehensively calculating and weighting the first matching scores of all first child nodes on the path from the root node to a specific first leaf node through a decision tree chain traversal algorithm. In this embodiment, it can be understood as a comprehensive quantitative score reflecting the degree of matching between the patient's multiple first condition labels and a complete cross-departmental treatment path.

[0111] Specifically, the system starts from the root node of the decision tree and performs a depth-first traversal along each possible branch direction, recording the identifier and position information of each first child node passed during the traversal, until a first leaf node is reached, forming a complete path sequence from the root node to the leaf node.

[0112] For example, for a diagnosis and treatment decision tree for patients with acute chest pain, the system may identify the path "root node → chest pain type judgment node → pain severity node → electrocardiogram abnormality detection node → acute myocardial infarction treatment leaf node". The system will collect the first matching score calculated from all the first child nodes on this path.

[0113] After collecting the first matching scores of all first child nodes on a given path, a hierarchical weighted algorithm can be used to calculate the second matching score of the corresponding first leaf node on that path. Specifically, the system assigns corresponding hierarchical weight coefficients to first child nodes at different levels based on the hierarchical structure of the decision tree and the logical importance of medical diagnosis. Generally, first child nodes closer to the leaf nodes are more important in the diagnosis and treatment decision, and therefore are given higher weight values. The system multiplies the first matching score of each first child node by its corresponding hierarchical weight, then sums all the weighted scores, and finally divides by the sum of the weights of all first child nodes on that path to obtain the standardized second matching score.

[0114] For example, if a path contains four first child nodes with first matching scores of 85%, 78%, 92%, and 88%, and corresponding hierarchical weights of 0.1, 0.2, 0.3, and 0.4, the system calculates the second matching score of the first leaf node as 87.1% by weighted average.

[0115] In step S204, the system sorts and compares the second matching scores of all possible first leaf nodes in the decision tree, and selects the node with the highest score as the final treatment path decision node. When scores are the same or close, the system further considers auxiliary factors such as the urgency of the patient's condition, the availability of hospital department resources, and the cost-effectiveness of treatment to ensure that the selected target leaf node is not only optimal in terms of matching degree, but also has the best operability and effectiveness in actual clinical application.

[0116] In step S205, the system calls the node decision function of the target leaf node to determine multiple target departments based on the diagnostic and treatment decision logic carried by the node and the specific characteristics of the patient's condition. Specifically, the node decision function comprehensively considers factors such as the severity of the patient's condition, the complexity of symptoms, and the urgency of treatment, combined with the hospital's departmental setup and professional expertise, to generate a combination of target departments that includes primary diagnostic and treatment departments and auxiliary support departments. For example, for a suspected patient with acute myocardial infarction, the node decision function may determine multiple target departments such as "Emergency Department," "Cardiology Department," "Cardiac Catheterization Laboratory," and "CCU Intensive Care Unit," each of which undertakes specific diagnostic and treatment functions and collaborative tasks.

[0117] In step S206, the system can determine the order of visits, length of stay, and focus of treatment for each target department based on the disease's diagnostic and treatment patterns and interdepartmental collaboration, thus forming a complete cross-departmental treatment process plan. For example, for the aforementioned suspected acute myocardial infarction patient, the system may formulate a first diagnostic path of "Emergency Department (stable vital signs, preliminary diagnosis) → Cardiology Department (professional assessment, treatment plan formulation) → Cardiac Catheterization Laboratory (interventional treatment) → CCU Intensive Care Unit (postoperative monitoring, rehabilitation guidance)," clarifying the treatment tasks and transfer criteria for each department.

[0118] In actual medical diagnosis and treatment, although the path determination method based on the second matching score can select the cross-departmental diagnosis and treatment path that best matches the patient's condition from the perspective of medical diagnosis, in clinical practice, there are often situations such as the selected target department having limited resources, busy doctors, occupied equipment, or insufficient beds. As a result, although the patient has obtained the medically optimal diagnosis and treatment path recommendation, they cannot receive the corresponding medical services in a timely manner due to resource constraints, which leads to problems such as prolonged patient waiting time, reduced diagnosis and treatment efficiency, and uneven distribution of medical resources.

[0119] To address the aforementioned issues, as an optional embodiment based on the above embodiments, the first leaf node further includes a resource load factor; the above method for determining cross-departmental treatment pathways may also include the following steps:

[0120] Step S301: Obtain the resource load status of the target department corresponding to the first leaf node;

[0121] Step S302: Calculate the resource load factor based on the resource load situation;

[0122] Step S303: Calculate the comprehensive matching score of the first leaf node based on the first matching score and resource load factor of the first leaf node;

[0123] Step S304: Determine the first leaf node with the highest comprehensive matching score as the target leaf node.

[0124] In step S301, the resource load status refers to the comprehensive reflection of the medical resource occupancy status and availability level of each target department in the hospital at the current time point. In this embodiment, it can be understood as a dynamic resource status indicator that includes multiple dimensions such as doctor workload, equipment utilization rate, bed occupancy rate, number of waiting patients, and department processing capacity, which is used to evaluate the ability and efficiency of a specific department to receive new patients.

[0125] Specifically, the system first identifies all target departments corresponding to the current first leaf node, and then queries the real-time resource status data of each department, including the ratio of the number of doctors currently on duty to the standard configuration, the usage status and availability of major medical equipment, the occupancy and vacancy of hospital beds or treatment beds, the length of the patient queue waiting for treatment in the department, the recent patient processing speed and average treatment time of the department, and other key indicators.

[0126] For example, for the cardiology department, the system will obtain specific data such as the number of doctors currently on duty in the department, the availability of electrocardiographs and ultrasound equipment, the bed occupancy rate of the cardiology ward, the number of patients waiting in the outpatient clinic, and the number of patients who have completed their diagnosis and treatment in the past hour.

[0127] In one feasible implementation, the system obtains resource load information by establishing standardized data interfaces with various information subsystems of the hospital. Specifically, the system interfaces with the hospital's human resource management system to obtain doctors' scheduling information and real-time on-duty status in each department; with the equipment management system to obtain the usage status and maintenance information of medical equipment; with the bed management system to obtain bed occupancy rates and appointment information; and with the queuing and calling system to obtain the number of patients waiting in each department and their waiting time, thereby achieving real-time monitoring and data updates of departmental resource load.

[0128] In another feasible implementation, the system also considers historical trends and predicted changes in departmental resource load to enhance the accuracy of resource load assessment. Specifically, the system analyzes the resource load patterns of specific departments at different times, such as a department typically being more heavily loaded in the morning and relatively less so in the afternoon, or experiencing resource shortages due to high disease incidence during specific seasons. The system combines these historical patterns to correct and predict the current resource load, in order to better guide the determination of treatment pathways.

[0129] In step S302, the resource load factor refers to a standardized numerical index obtained by quantitative calculation based on the resource load of the target department. It is used to measure the suitability of the department to receive new patients and its service provision capabilities. In this embodiment, it can be understood as converting the multi-dimensional resource status information of the department into a load assessment parameter with a unified evaluation standard. The higher the value, the more abundant the department's resources and the stronger its service capabilities. The lower the value, the heavier the department's load and the more limited its ability to receive new patients.

[0130] Specifically, the system first standardizes various resource indicators, converting indicators with different dimensions such as physician workload, equipment utilization, bed occupancy, and patient waiting time into standardized values ​​between zero and one. Then, the system assigns corresponding weight coefficients based on the importance of different resource types in departmental operations. For example, departments requiring specialized equipment may be given a higher weight for equipment availability, while departments primarily relying on manual diagnosis and treatment may place greater emphasis on physician workload. The system multiplies each standardized indicator by its corresponding weight and sums the results to obtain the department's overall load level. Finally, through numerical conversion, the load level is transformed into a resource load factor, ensuring that departments with more abundant resources receive higher resource load factor values.

[0131] In one feasible implementation, the system dynamically adjusts the resource load factor based on the urgency of the patient's condition. Specifically, for critically ill patients, the system prioritizes the availability of emergency equipment and emergency treatment capabilities within the department, correspondingly increasing the weight of emergency-related resource indicators; for patients with chronic diseases or those requiring routine care, the system focuses more on the department's daily service capabilities and patient waiting times, adjusting the weight allocation to better reflect the department's suitability for these patients.

[0132] In step S303, the comprehensive matching score refers to the weighted integration calculation of the second matching score of the first leaf node and the resource load factor to obtain a comprehensive evaluation index that considers both the medical diagnosis matching degree and the availability of departmental resources. In this embodiment, it can be understood as a comprehensive quantitative score that reflects both the medical suitability of the treatment pathway and the feasibility of implementation.

[0133] Specifically, the system sets weight parameters for the second matching score and the resource load factor. The weight of the second matching score reflects the importance of the accuracy of medical diagnosis, while the weight of the resource load factor reflects the importance of optimal resource allocation. The sum of the two weight parameters equals one to ensure the standardization of the comprehensive score. The system multiplies the second matching score by its corresponding weight and the resource load factor by its corresponding weight, and then adds the two weighted results to obtain the final comprehensive matching score.

[0134] In one feasible implementation, the system dynamically adjusts the weighting of the second matching score and the resource load factor based on the hospital's management strategy and clinical needs. Specifically, during normal periods when medical resources are sufficient, the system may set a higher medical matching weight to prioritize the quality of diagnosis and treatment; during peak periods when medical resources are strained or in case of emergencies, the system may appropriately increase the resource load weight to better balance resource allocation and ensure the efficient operation of overall medical services.

[0135] In step S304, the system sorts and compares the comprehensive matching scores of all candidate first leaf nodes, and selects the node with the highest score as the final target leaf node.

[0136] Compared to the previous method of selection based solely on the second matching score, the selection mechanism based on the comprehensive matching score can ensure the accuracy of medical diagnosis while rationally allocating patients to departments with relatively light resource loads, avoiding situations where some departments are overcrowded while others are relatively idle.

[0137] By incorporating a resource load factor into the first leaf node and determining the target leaf node based on a comprehensive matching score, this method effectively addresses the problem of traditional treatment pathway determination methods that only consider medical matching degree while neglecting resource availability. This achieves a balance between medical accuracy and optimal resource allocation. The method can dynamically adjust patient allocation strategies based on the real-time resource status of each department, avoiding uneven utilization of medical resources, reducing patient waiting time due to excessive departmental load, and improving the overall efficiency of medical services and patient satisfaction.

[0138] In actual medical diagnosis and treatment, a patient's condition is often a dynamic process. Although the initial diagnostic path determined at the first visit is based on the information about the patient's condition at that time and a corresponding interdisciplinary treatment plan is formulated, as the diagnosis and treatment process progresses, the patient may develop new symptoms, obtain new test results after undergoing various examinations, experience drug reactions or complications during treatment, or have their original symptoms worsen or improve. If the system still strictly follows the treatment plan based on the initially determined initial diagnostic path, it may cause the treatment direction to deviate from the patient's actual needs, affecting the patient's treatment effect and safety.

[0139] To address the aforementioned issues, as an optional embodiment based on the above embodiments, the method for determining cross-departmental treatment pathways may further include the following steps:

[0140] Step S401: In response to changes in the patient's condition information, the changed condition information is converted into multiple secondary condition tags;

[0141] Step S402: Match multiple second disease labels with the second child nodes and second leaf nodes in the decision tree to update the first diagnostic path and obtain the second diagnostic path.

[0142] In step S401, the system establishes a real-time monitoring mechanism for patient condition information. By connecting with information systems across various hospital departments, it continuously tracks new patient condition information generated during cross-departmental treatment. Specifically, the system connects with the laboratory information system of the laboratory department to obtain patient test results, with the medical imaging information system of the radiology department to obtain CT, MRI, and ultrasound imaging reports, with the electronic medical record systems of various departments to obtain doctors' treatment records and updated condition assessments, with the nursing information system to obtain changes in patient vital signs and clinical observation records, and with the pharmacy management system to obtain patient medication information and drug response records, thereby achieving comprehensive real-time monitoring of changes in patient condition information.

[0143] In step S402, the system first analyzes the degree of difference and trend of change between the second disease label and the original first disease label, assesses the nature and severity of the disease change, and then decides whether to perform a complete rematch or to make local adjustments and optimizations based on the degree of change.

[0144] When the system detects a fundamental change in a patient's condition, such as a change from a general illness to an acute or critical illness, or from an internal medicine condition to a condition requiring surgery, the system will initiate a complete rematching mode. Using a second set of condition labels, the system will start a complete rematching process from the root node of the decision tree. Logical judgments will be made through the conditional functions of the second child nodes, and the matching scores of each second leaf node will be calculated to determine a new treatment path that best matches the current condition, forming a second diagnostic path that is completely different from the original path.

[0145] When the system assesses that a patient's condition changes as either a gradual adjustment or a progression from the original disease, it employs a pathway optimization adjustment mode. While maintaining the core framework of the original treatment pathway, this mode refines elements such as specific departmental arrangements, treatment priorities, and resource allocation based on the characteristics of the second disease label, creating an optimized and improved second diagnostic pathway. This adjustment mode responds to changes in the patient's condition while maintaining the continuity and stability of the treatment process, avoiding the inconvenience caused to patients and medical staff by frequent pathway changes.

[0146] Once the second diagnostic pathway is determined, the system automatically generates a pathway change notification and execution instructions, pushing pathway adjustment information to medical staff in relevant departments. This includes detailed information such as new treatment requirements, changed departmental arrangements, and adjustment priority levels, ensuring that the new treatment pathway is implemented promptly and accurately. Simultaneously, the system records the detailed process and basis for the pathway change, providing crucial feedback data for subsequent medical quality assessments and decision tree optimization, forming a closed-loop management mechanism for continuous improvement.

[0147] By employing the embodiments of this application, a dynamic path adjustment mechanism that responds to changes in patient condition information is established, effectively solving the problem that static treatment paths cannot adapt to dynamic changes in patient conditions. This enables real-time optimization and adjustment of cross-departmental treatment paths. The method can promptly capture various changes in patient condition information, quickly generate a second condition label reflecting the current condition status, and redetermine the most suitable treatment path based on the updated condition characteristics, ensuring that patients always receive cross-departmental treatment services that best match their actual condition.

[0148] In practical applications, as system uptime increases and the number of cases processed grows, some first child nodes in the decision tree may exhibit insufficient accuracy or excessive logical complexity. These nodes often produce incorrect judgments when faced with a large number of real cases, or although their judgments are relatively accurate, their complex condition settings lead to low computational efficiency. Additionally, some first child nodes may have highly redundant judgment logic with other nodes, resulting in redundancy in the decision tree structure. This not only affects the overall path determination speed and accuracy but also increases the complexity of system maintenance and the workload of medical experts in updating rules.

[0149] To address the aforementioned issues, as an optional embodiment based on the above embodiments, the method for determining cross-departmental treatment pathways may further include the following steps:

[0150] Step S501: Obtain historical case data, which includes the disease labels of multiple historical patients and the corresponding treatment path results;

[0151] Step S502: Based on historical case data, calculate the error rate and complexity cost of each first child node;

[0152] Step S503: Determine the first redundant child node based on the error rate and complexity cost;

[0153] Step S504: Remove the redundant first child node from the decision tree.

[0154] In step S501, historical case data refers to a complete set of records of all patient cases processed by the hospital using the cross-departmental diagnosis and treatment path determination method of this application over a period of time. It includes key information such as the disease label generated for each historical patient during system processing, the diagnosis and treatment path recommended by the system, the actual diagnosis and treatment process performed, and the final diagnosis and treatment result. In this embodiment of the application, it can be understood as a basic data resource used to evaluate and optimize the performance of the decision tree.

[0155] Specifically, the system extracts all relevant case records within a specified time range from the hospital's case database. Each case record contains detailed information such as the patient's complete set of disease labels, the recommended treatment path determined by the system through a decision tree, the treatment path actually used by medical staff, the patient's treatment effect evaluation, and path adjustment records during the treatment process.

[0156] For example, for a patient's history of acute chest pain, the system will record the patient's condition tags such as "symptom type - chest pain", "duration - 45 minutes", and "pain nature - squeezing pain", the system's recommended treatment path of "emergency department → cardiology department → cardiac catheterization laboratory", the actual treatment process performed, and the final diagnosis of acute myocardial infarction and successful completion of interventional treatment.

[0157] In step S502, the error rate of the first child node refers to the proportion of times when the condition function judgment result of a specific first child node does not match the actual diagnosis and treatment needs of the patient when processing historical cases. The accuracy level of the node's judgment is quantified by statistically analyzing the ratio between the number of cases in which the node makes an incorrect judgment and the total number of cases in which the node participates in the judgment. In this embodiment, it can be understood as a quantitative indicator reflecting the degree of consistency between the medical judgment logic of the first child node and clinical practice.

[0158] Similarly, complexity cost refers to a comprehensive quantitative indicator of the computational resources consumed and the complexity of logical processing required by the first child node in the process of executing condition judgment. It includes a complexity assessment of multiple dimensions such as the number of judgment conditions contained in the node's condition function, the nesting level of logical operations, the diversity of disease label types involved, and the computational time cost of condition matching. In this embodiment, it can be understood as a comprehensive evaluation parameter to measure the degree of impact of the first child node on the overall system performance, used to identify those nodes that may have a certain degree of judgment accuracy but affect system efficiency due to excessive logic complexity.

[0159] Specifically, when calculating the error rate of the first child node, the system first establishes a correspondence between the node judgment results and the actual diagnosis and treatment needs for verification. For each historical case, the system extracts the judgment outputs of all the first child nodes and their condition functions that the case traversed during the decision tree traversal, and then compares and analyzes these judgment outputs with the patient's actual diagnosis and treatment process and the final diagnosis and treatment results.

[0160] When the conditional function of a first child node outputs a true value, it means that the node believes the patient should continue to be treated along the corresponding treatment branch. The system will check whether the patient's actual treatment path does indeed require the treatment strategy represented by the branch. If the patient's final treatment result shows that the branch judgment is correct, it is recorded as a correct judgment. If the patient's actual treatment shows that the branch judgment is inappropriate or causes a delay in treatment, it is recorded as an incorrect judgment.

[0161] For example, a first sub-node in the "Acute Myocardial Infarction Risk Assessment" function might classify a patient with chest pain as high-risk and requiring immediate cardiology intervention. However, if the patient is ultimately diagnosed with gastroesophageal reflux causing chest pain and only requires gastroenterology intervention, this judgment would be recorded as incorrect. The system calculates the error rate of each first sub-node by statistically analyzing the number of correct and incorrect judgments across all relevant historical cases, and then determining the proportion of incorrect judgments to the total number of judgments.

[0162] When calculating complexity costs, the system counts the number of basic judgment conditions contained in the condition function of each first child node. A higher number of conditions indicates more complex judgment logic and a greater diversity of disease label types to be processed, thus increasing the computational burden of condition matching. Secondly, the system analyzes the nested structure of logical operations in the condition functions, including the combination levels of logical AND, OR, and NOT operations. A deeper nesting level indicates more complex judgment logic and a greater number of recursive calculations required. Thirdly, the system assesses the diversity of disease label types involved in the node. When a node needs to process multiple different types of labels simultaneously, such as symptom labels, examination result labels, and medical history labels, its complexity increases accordingly.

[0163] In addition, the system measures the actual execution time of the conditional function of each first child node. By running a large amount of case data in a standard test environment, the average judgment time of the node is calculated. The longer the execution time, the higher the computational complexity of the node. After standardizing the above complexity indicators, the system performs a weighted summation according to a preset weight allocation to obtain a complexity cost value that reflects the overall complexity of the first child node.

[0164] In step S503, the system identifies redundant nodes based on the error rate and complexity cost of each first child node. The system first sets error rate thresholds and complexity cost thresholds as criteria for redundancy determination. These thresholds can be adjusted according to the hospital's quality requirements and system performance needs. Typically, the error rate threshold is set to the upper limit of acceptable diagnostic errors for the hospital, and the complexity cost threshold is set to the target value for system performance optimization.

[0165] When the error rate of a first child node exceeds a set threshold, it indicates that the node's judgment accuracy is insufficient and cannot provide reliable decision support for determining the treatment path; when the complexity cost of a first child node exceeds a threshold, it indicates that the node's logic is too complex and affects the overall efficiency of the system; when a first child node simultaneously meets the conditions of excessive error rate and / or excessive complexity cost, the system identifies it as a redundant node and removes it from the decision tree.

[0166] In step S504, the system first analyzes the position and connection relationship of each first child node marked as redundant in the decision tree, determines the impact of removing the node on the overall tree structure, and then redesigns the connection logic of the affected branches to ensure that the decision tree still maintains complete path coverage from the root node to the leaf node after removing the redundant node.

[0167] During the removal process, the system will directly connect the parent node that was originally connected to the redundant node to the optimal child node of the redundant node, or redesign the connection relationship according to medical logic to avoid path breakage or logical loss due to node removal.

[0168] After the removal is complete, the system will perform an integrity verification on the updated decision tree to ensure that all possible combinations of disease labels can find the corresponding processing path in the new decision tree.

[0169] By employing the embodiments of this application, the decision tree optimization based on historical case data effectively solves the problems of performance degradation and structural redundancy that may occur during long-term use of the decision tree, and realizes the self-optimization and continuous improvement of the system.

[0170] In practical applications of medical decision trees, the construction of decision trees often involves the joint participation of medical experts from multiple departments. Based on their respective professional backgrounds and clinical experience, different experts may design first sub-nodes with similar functions or overlapping logic. Although these nodes may differ in their surface condition descriptions, their core medical judgment logic and the characteristics of the disease they handle are essentially highly similar. For example, one node judges "chest pain lasting more than 30 minutes and accompanied by sweating," while another node judges "typical angina symptoms lasting more than half an hour and accompanied by vegetative symptoms." These two nodes are actually identifying the same risk characteristics of acute coronary syndrome, but they are set as different nodes due to differences in description methods and professional terminology. This functional duplication not only causes redundancy and complexity in the decision tree structure.

[0171] To address the aforementioned issues, as an optional embodiment based on the above embodiments, the method for determining cross-departmental treatment pathways may further include the following steps:

[0172] Step S601: Calculate the conditional similarity between the first child nodes;

[0173] Step S602: Determine at least two first child nodes with a conditional similarity greater than a threshold as similar node pairs;

[0174] Step S603: Merge the first child nodes in the similar node pairs to obtain the merged node;

[0175] Step S604: Update the decision tree based on the merged nodes.

[0176] In step S601, conditional similarity refers to the similarity index that quantifies the degree of functional duplication between two or more nodes by comparing the consistency of the conditional functions of different first child nodes in terms of medical judgment logic, handling of disease label types, and judgment condition settings. In this embodiment, it can be understood as a quantitative evaluation parameter that reflects whether there is a possibility of merging between the first child nodes.

[0177] Specifically, the system first analyzes the types and number of disease labels involved in the conditional functions of each first child node, and statistically analyzes the overlap of label types processed by different nodes. When there is a high degree of overlap in the disease label types processed by two nodes, it indicates that they may have similar medical judgment objectives. Then, the system further analyzes the specific judgment logic of the node conditional functions, including the judgment threshold settings for the same disease label, the combination of logical operations, and the requirements for condition satisfaction, etc. The similarity of node functions is evaluated by comparing the consistency of these logical elements.

[0178] For example, if the system finds that two first child nodes both process chest pain-related disease labels and both set similar pain duration thresholds and pain nature judgment conditions, then it considers that the two nodes have a high degree of conditional similarity.

[0179] In the process of calculating conditional similarity, the system adopts a multi-dimensional comprehensive evaluation method to ensure the accuracy and comprehensiveness of the similarity calculation. The system first constructs a feature vector for each first child node. This vector contains key information such as all disease label types involved in the node, the judgment conditions for each label, logical operation relationships, and conditional threshold settings. Then, the system uses a vector similarity calculation method to quantify the degree of similarity between different nodes.

[0180] In step S602, the system identifies and pairs similar nodes based on the calculated conditional similarity results. The system first sets a conditional similarity threshold as a standard for judging whether nodes have merging value. The setting of this threshold needs to comprehensively consider the accuracy requirements of medical judgment and the effect goals of system optimization. Generally, a higher threshold can ensure that only nodes with truly duplicated functions will be identified as similar, while a lower threshold may identify more nodes with merging potential but requires more careful manual review.

[0181] When the conditional similarity of two or more first child nodes exceeds a set threshold, the system organizes them into similar node pairs or similar node groups. All nodes in each node pair or group are considered to have sufficient functional similarity and are suitable for merging. During the identification of similar node pairs, the system also considers the nodes' positional relationships and hierarchical structure within the decision tree, prioritizing the identification of similar nodes at the same or similar levels. This is because merging these nodes will not significantly impact the overall structure of the decision tree. Simultaneously, the system assesses the impact of node merging on downstream paths to ensure that the merging operation does not disrupt the medical logical integrity of the decision tree.

[0182] In step S603, merging a node refers to generating a new comprehensive node to replace the original multiple similar nodes by integrating the condition functions and judgment logic of each first child node in the similar node pair. In this embodiment, it can be understood as an optimized node with the comprehensive functions of the original multiple nodes, which maintains the capabilities of the original nodes and simplifies the structural complexity of the decision tree.

[0183] Specifically, the system first analyzes the condition function of each node in a pair of similar nodes, extracting the common judgment logic and differentiated conditions. Then, it designs a new condition function for merging nodes, which can cover the core judgment capabilities of the original nodes while eliminating redundant judgment conditions. During the merging process, the system uses logic optimization techniques to merge multiple similar judgment conditions into a simpler but more comprehensive condition with the same coverage.

[0184] For example, "chest pain lasting more than 30 minutes" and "typical angina symptoms lasting more than half an hour" can be merged into a single condition of "chest pain symptoms lasting more than 30 minutes". At the same time, the system will also integrate the performance data of each original node in historical cases to calculate a comprehensive performance index for the merged node, ensuring that the merged node inherits the advantages of the original node while avoiding its shortcomings.

[0185] By adopting the embodiments of this application, the structural redundancy and inefficiency caused by functionally repetitive nodes in the decision tree are effectively solved by establishing a conditional similarity calculation and similar node merging mechanism, thus realizing intelligent structural optimization of the decision tree.

[0186] Figure 2 This schematic diagram illustrates a structural block diagram of a cross-departmental diagnosis and treatment pathway determination system provided in this application, which may include:

[0187] The diagnosis and treatment information acquisition module is used to obtain the patient's first diagnosis and treatment information during the historical diagnosis and treatment process and the second diagnosis and treatment information during the current diagnosis and treatment process.

[0188] The disease information acquisition module is used to respond to the acquisition of the patient's disease information and convert the disease information into multiple first disease tags;

[0189] A decision tree acquisition module is used to acquire a decision tree, which includes multiple first child nodes and first leaf nodes. The first child nodes represent first diagnostic judgment conditions based on the patient's condition information; the first leaf nodes represent treatment path decisions between different departments.

[0190] The diagnostic path determination module is used to match the plurality of first disease labels with the first child node and the first leaf node in the decision tree to determine the first diagnostic path.

[0191] Based on the above embodiments, as an optional embodiment, the diagnostic path determination module is further configured to, for any first child node, judge the plurality of first disease labels based on the node condition function of the first child node to obtain the Boolean judgment result of the plurality of first disease labels; calculate the first matching score between the first child node and the plurality of first disease labels according to the Boolean judgment result; calculate the second matching score of each first leaf node based on the first matching score of the first child node through chain traversal of the decision tree; determine the first leaf node with the highest second matching score as the target leaf node; determine the corresponding plurality of target departments through the node decision function of the target leaf node; and determine the first diagnostic path based on the plurality of target departments.

[0192] Based on the above embodiments, as an optional embodiment, the diagnostic path determination module is further configured to obtain the resource load status of the target department corresponding to the first leaf node; calculate the resource load factor based on the resource load status; calculate the comprehensive matching score of the first leaf node based on the first matching score of the first leaf node and the resource load factor; and determine the first leaf node with the highest comprehensive matching score as the target leaf node.

[0193] Based on the above embodiments, as an optional embodiment, the diagnostic path determination module is further configured to respond to changes in the patient's condition information, convert the changed condition information into multiple second condition labels; and match the multiple second condition labels with the second child nodes and second leaf nodes in the decision tree to update the first diagnostic path to obtain a second diagnostic path.

[0194] Based on the above embodiments, as an optional embodiment, the decision tree acquisition module is further used to acquire historical case data, which includes multiple historical patients' disease labels and corresponding treatment path results; based on the historical case data, calculate the error rate and complexity cost of each first sub-node; determine redundant first sub-nodes according to the error rate and complexity cost; and remove redundant first sub-nodes from the decision tree.

[0195] Based on the above embodiments, as an optional embodiment, the decision tree acquisition module is further configured to calculate the conditional similarity between the first child nodes; determine at least two first child nodes whose conditional similarity is greater than a threshold as similar node pairs; merge the first child nodes in the similar node pairs to obtain merged nodes; and update the decision tree based on the merged nodes.

[0196] It should be noted that the cross-departmental diagnosis and treatment path determination system part in the embodiments of this application corresponds to the cross-departmental diagnosis and treatment path determination method part in the embodiments of this application. For a detailed description of the cross-departmental diagnosis and treatment path determination system part, please refer to the cross-departmental diagnosis and treatment path determination method part, which will not be repeated here.

[0197] Figure 3 The diagram illustrates a structural block diagram of an electronic device suitable for implementing the methods described above, according to an embodiment of this application. Figure 3 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0198] like Figure 3 As shown, an electronic device 300 according to an embodiment of this application includes a processor 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage portion 308 into a random access memory (RAM) 303. The processor 301 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 301 may also include onboard memory for caching purposes. The processor 301 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of this application.

[0199] RAM 303 stores various programs and data required for the operation of electronic device 300. Processor 301, ROM 302, and RAM 303 are interconnected via bus 304. Processor 301 executes various operations of the method flow according to embodiments of this application by executing programs in ROM 302 and / or RAM 303. It should be noted that the programs may also be stored in one or more memories other than ROM 302 and RAM 303. Processor 301 may also execute various operations of the method flow according to embodiments of this application by executing programs stored in said one or more memories.

[0200] According to embodiments of this application, the electronic device 300 may further include an input / output (I / O) interface 305, which is also connected to a bus 304. The system 300 may also include one or more of the following components connected to the input / output (I / O) interface 305: an input section 306 including a keyboard, mouse, etc.; an output section 307 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN card, modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the input / output (I / O) interface 305 as needed. A removable medium 311, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 310 as needed so that computer programs read from it can be installed into the storage section 308 as needed.

[0201] According to embodiments of this application, the method flow according to embodiments of this application can be implemented as a computer software program. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable storage medium, the computer program containing program code for performing the methods shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network via communication section 309, and / or installed from removable medium 311. When the computer program is executed by processor 301, it performs the functions defined in the system of embodiments of this application. According to embodiments of this application, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0202] This application also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of this application.

[0203] According to embodiments of this application, the computer-readable storage medium can be a non-volatile computer-readable storage medium. Examples include, but are not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0204] For example, according to embodiments of this application, a computer-readable storage medium may include the ROM 302 and / or RAM 303 described above and / or one or more memories other than ROM 302 and RAM 303.

[0205] Embodiments of this application also include a computer program product comprising a computer program containing program code for performing the methods provided in the embodiments of this application. When the computer program product is run on an electronic device, the program code is used to enable the electronic device to implement the methods provided in the embodiments of this application.

[0206] When the computer program is executed by the processor 301, it performs the functions defined in the system / apparatus of this application embodiment. According to the embodiments of this application, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0207] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and may be downloaded and installed via communication section 309, and / or installed from removable medium 311. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0208] According to embodiments of this application, program code for executing the computer programs provided in the embodiments of this application can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0209] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions. Those skilled in the art will understand that the features recited in the various embodiments and / or claims of this application can be combined and / or combined in various ways, even if such combinations or combinations are not expressly stated in this application. In particular, the various embodiments and / or features described in the claims of this application may be combined and / or combined in various ways without departing from the spirit and teachings of this application. All such combinations and / or combinations fall within the scope of this application.

[0210] The embodiments of this application have been described above. However, these embodiments are merely illustrative and not intended to limit the scope of this application. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. The scope of this application is defined by the appended claims and their equivalents. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this application, and all such substitutions and modifications should fall within the scope of this application.

[0211] The embodiments of this application have been described above. However, these embodiments are merely illustrative and not intended to limit the scope of this application. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. The scope of this application is defined by the appended claims and their equivalents. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this application, and all such substitutions and modifications should fall within the scope of this application.

Claims

1. A method for determining cross-departmental diagnostic and treatment pathways, comprising: The system receives the patient's condition information and converts it into multiple primary condition tags. A decision tree is obtained, which includes multiple first child nodes and first leaf nodes. The first child nodes represent first diagnostic judgment conditions based on the patient's condition information; the first leaf nodes represent treatment path decisions between different departments. The multiple first disease labels are matched with the first child node and the first leaf node in the decision tree to determine the first diagnostic path.

2. The method according to claim 1, wherein the first sub-node includes a node condition function, the node condition function representing the judgment of the first disease label; the first leaf node includes a node decision function; the node decision function representing the diagnosis and treatment decision for the target department corresponding to the first leaf node.

3. The method according to claim 2, wherein matching the plurality of first disease labels with the first child node and the first leaf node in the decision tree to determine the first diagnostic path includes: For any of the first child nodes, the plurality of first disease labels are judged based on the node condition function of the first child node to obtain the Boolean judgment result of the plurality of first disease labels; Based on the Boolean judgment result, calculate the first matching score between the first child node and the plurality of first disease labels; Based on the first matching score of the first child node, the second matching score of each first leaf node is calculated by chain traversal of the decision tree. The first leaf node with the highest second matching score is identified as the target leaf node; Multiple target departments are determined by the node decision function of the target leaf node; The first diagnostic pathway is determined based on the multiple target departments.

4. The method according to claim 3, wherein the first leaf node further includes a resource load factor; the method further includes: Obtain the resource load status of the target department corresponding to the first leaf node; Calculate the resource load factor based on the resource load situation; Calculate the comprehensive matching score of the first leaf node based on the first matching score of the first leaf node and the resource load factor; The first leaf node with the highest overall matching score is identified as the target leaf node.

5. The method according to claim 1, further comprising: In response to changes in the patient's condition information, the changed condition information is converted into multiple secondary condition tags; The multiple second disease labels are matched with the second child nodes and the second leaf nodes in the decision tree to update the first diagnostic path and obtain the second diagnostic path.

6. The method according to claim 2, further comprising: Acquire historical case data, which includes the disease tags of multiple historical patients and the corresponding treatment pathway results; Based on the historical case data, calculate the error rate and complexity cost of each of the first sub-nodes; Based on the error rate and complexity cost, determine the redundant first child node; Remove the redundant first child node from the decision tree.

7. The method according to claim 6, further comprising: Calculate the conditional similarity between the first child nodes; At least two first child nodes whose conditional similarity is greater than the threshold are identified as similar node pairs; The first child node in the similar node pair is merged to obtain the merged node; The decision tree is updated based on the merged node.

8. A cross-departmental diagnostic and treatment pathway determination system, comprising: The disease information acquisition module is used to respond to the acquisition of the patient's disease information and convert the disease information into multiple first disease tags; A decision tree acquisition module is used to acquire a decision tree, which includes multiple first child nodes and first leaf nodes. The first child nodes represent first diagnostic judgment conditions based on the patient's condition information; the first leaf nodes represent treatment path decisions between different departments. The diagnostic path determination module is used to match the plurality of first disease labels with the first child node and the first leaf node in the decision tree to determine the first diagnostic path.

9. An electronic device, comprising: One or more processors; Memory, used to store one or more programs. Wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of claims 1 to 7.

10. A computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, cause the processor to perform the method of any one of claims 1 to 7.

Images