Intelligent agent and system for intelligent doctor's order assistance

Abstract

The application relates to the technical field of natural language processing, and particularly discloses an intraoperative doctor order intelligent assistance intelligent agent and system, which comprises the following modules: a doctor order semantic processing unit acquires a doctor input natural language order and performs semantic analysis, generates a structured order execution item, a surgery context association unit combines surgery information and a surgery process node to call an order rule for verification, and establishes an association relationship between the structured order execution item and the surgery process node, an order execution control unit controls the order to enter an execution state according to a verification result, and a state feedback processing unit updates or maintains a surgery process node state according to the execution state. The application converts the natural language order into a data object for verification and process control, and integrates the order verification result and the execution feedback into the surgery process node state control, so that the order analysis, the execution feedback and the process promotion form a continuous data processing and control relationship.

Description

Technical Field

[0001] This invention relates to the field of natural language processing technology, and in particular to an intelligent assistant for intraoperative physician medical orders and its system. Background Technology

[0002] Natural Language Processing (NLP) technology primarily focuses on the acquisition, segmentation, semantic recognition, entity extraction, intent determination, structured conversion, rule matching, data storage, and terminal interaction of natural language text and speech-to-text. Typically, the terminal acquires text or speech data, which is then parsed by a processor that calls upon programs and dictionaries, rules, and business records in a database. The processing results are then transmitted to the corresponding business terminal. Traditional intraoperative physician order intelligent auxiliary systems refer to computer systems that process the reception, parsing, recording, and execution of physician orders during surgery. Physicians typically input medication, blood transfusion, consumable usage, and operational instructions via voice or touchscreen. The system converts the speech to text, identifies and structures the order based on the words, drug names, dosages, target, and operation type, and then matches and judges the order according to preset rules and patient, surgical, and drug data. The corresponding order content is then sent to the nurse's terminal, where the nurse confirms execution and reports the execution status. The relevant issuance time, reception record, execution record, and confirmation information are stored on a server or hospital information system.

[0003] Traditional surgical order processing relies primarily on doctors' voice or touchscreen input for text conversion and structured recording. The content of these orders lacks a direct link to surgical procedure milestones. Nurses' feedback is often recorded independently, making it difficult to use as a condition for process control. Personnel verification and order processing are separate processes, and there is a lack of linkage between surgical step status and order execution status. In actual surgery, the surgeon needs to continuously monitor the surgical field, patient status, and current surgical step, while simultaneously issuing orders for medication, blood transfusions, consumable preparation, positioning adjustments, equipment start / stop, or other procedures. The surgeon typically cannot continuously operate the terminal or check the execution status of each order after issuance; identity and professional authorization verification requires the joint participation of the surgeon, anesthesiologist, and circulating nurse. Furthermore, resources such as personnel, medications, consumables, and surgical equipment must be coordinated according to the current surgical step. When identity verification results, order content, personnel availability, equipment status, and nurse feedback are processed by different systems or operational links, the surgeon struggles to continuously confirm whether the above information meets the execution conditions of the current surgical step without interrupting the surgical procedure.

[0004] Existing medical order information systems typically focus on recording and transmitting medical orders. Even if they can perform voice recognition, structuring medical orders, or nurse signature verification, they often lack a mechanism to uniformly verify and access personnel identities, professional permissions, pending medical orders, and the personnel and equipment resources corresponding to the current surgical procedure. Especially when transitioning from one node to the next, existing systems struggle to determine whether to allow the procedure to proceed based on the tripartite verification status of the surgeon, anesthesiologist, and circulating nurse, the execution status of the medical orders associated with the current node, and the resource requirements of the current node.

[0005] Therefore, in the same surgery, the surgeon's orders, the verification results of the three parties, the surgical steps, personnel resources, equipment resources, and execution feedback are easily scattered across different processing links. Abnormal events and their handling results are also difficult to establish a continuous connection. This forces the surgeon to still rely on manual verbal confirmation or multi-terminal queries to complete intraoperative coordination, which not only increases their non-surgical workload but also easily leads to inconsistencies between the execution status of medical orders and the actual surgical procedure, affecting the consistency of intraoperative information processing and process traceability.

[0006] Furthermore, traditional medical order processing software typically operates passively by following preset interfaces or fixed business modules, with each module handling voice recognition, rule verification, information transmission, or status saving. The system itself usually does not continuously maintain a unified operational context that includes the current surgery, current process node, completed orders, pending orders, personnel status, and equipment status. It also lacks a mechanism to autonomously select the next processing action and invoke corresponding business resources based on this operational context. Therefore, how to construct an intelligent auxiliary agent capable of continuously perceiving the surgical context around the surgeon's intraoperative goals, verifying the actions of all three parties and related resources, parsing and judging natural language medical orders, selecting appropriate actions, and updating the surgical process status based on feedback, becomes a further problem that needs to be solved in intraoperative medical order information processing. Summary of the Invention

[0007] The purpose of this invention is to provide an intelligent auxiliary agent and system for intraoperative physician medical orders. It establishes a continuous technical processing chain around the formation of intraoperative medical orders, the structured processing of order content, the association of surgical procedure nodes, the feedback of execution status, and the confirmation of abnormal states. This enables the physician's intraoperative medical orders to form corresponding structured execution items based on the current surgical procedure status. The system also updates the execution status of the medical orders and the surgical procedure nodes based on the execution confirmation information or result feedback information during the execution process. When the execution information is inconsistent with the content of the medical orders or the completion conditions of the procedure, a pending confirmation status is retained. This solves the problems of insufficient association between intraoperative medical orders and the surgical procedure and actual execution status, as well as the difficulty in forming a closed loop for execution result feedback.

[0008] To achieve the above objectives, the present invention provides an intelligent assistant for intraoperative physician orders, including a physician order semantic processing unit, a surgical context association unit, a physician order execution control unit, and a status feedback processing unit. Each unit forms a data processing and status feedback relationship around the intraoperative physician orders, surgical process nodes, structured physician order execution items, and execution feedback information. The medical order semantic processing unit is used to receive medical order information issued by the surgeon during the operation, identify and structure the medical order object, medical order content and execution requirements in the medical order information, form a structured medical order execution item corresponding to a single medical order, and retain the correspondence between the structured medical order execution item and the original medical order information, so that subsequent process association and execution status processing are based on the same medical order object; The surgical context association unit is used to obtain the surgical process node corresponding to the current surgical process, associate the structured medical order execution item with the surgical process node, determine the association position of the medical order information in the current surgical process according to the execution conditions corresponding to the medical order content and the surgical process node, and form a structured medical order execution item based on the association result; the structured medical order execution item at least carries the association relationship between the medical order content, the corresponding surgical process node, and the execution status, so that the medical order content can enter the corresponding execution processing process according to the surgical process status; The surgical context association unit is used to determine the content of the medical order to be executed based on the structured medical order execution item, and compare the medical order content with the current state of the corresponding surgical process node; when the current state meets the execution conditions of the corresponding medical order content, the structured medical order execution item enters the execution state; when the current state does not meet the execution conditions of the corresponding medical order content, the structured medical order execution item remains in the pending execution state, so that the medical order execution order is constrained by the state of the surgical process node, rather than forming an execution result independently of the surgical process.

[0009] Furthermore, the status feedback processing unit is used to receive execution confirmation information or result feedback information corresponding to the structured medical order execution item, and match the execution confirmation information or result feedback information with the execution information in the structured medical order execution item, and then compare the matching result with the completion conditions of the corresponding surgical process node, thereby forming a feedback association between the medical order execution status and the surgical process node status. When the execution information matches and the completion conditions are met, the status feedback processing unit determines that the corresponding medical order execution status is valid, and updates the corresponding surgical process node according to the execution result, so that the status result after the completion of the medical order execution is returned to the surgical process processing chain; when the execution information does not match or the completion conditions are not met, the status feedback processing unit keeps the corresponding surgical process node in an incomplete state, generates an exception record containing the mismatched execution information and the corresponding surgical process node, and keeps the corresponding structured medical order execution item in a pending confirmation state, so as to avoid unconfirmed execution feedback directly changing the surgical process node.

[0010] Furthermore, for the structured medical order execution item in the pending confirmation state, the status feedback processing unit associates the structured medical order execution item, execution confirmation information, result feedback information, and surgical procedure node corresponding to the abnormal record, outputs the abnormal record, and receives the handling result for the abnormal record; when the handling result shows that the execution information is consistent with the structured medical order execution item and the completion condition of the surgical procedure node is met, the corresponding medical order execution status is updated to a valid state, and the corresponding surgical procedure node is updated to a completed state; when the handling result still does not meet the completion condition, the corresponding surgical procedure node remains in an incomplete state and the structured medical order execution item remains in the pending confirmation state, thereby ensuring that the abnormal confirmation result continues to be jointly constrained by the medical order execution information and the procedure completion condition; In the process of the intelligent assistant agent for intraoperative physician orders, after the medical order information is formed into structured medical order execution items, it is associated with the surgical process nodes. The association result further forms the structured medical order execution items. The structured medical order execution items enter the execution state or remain in the pending execution state according to the current state of the corresponding surgical process node. The execution confirmation information or result feedback information is then used to determine the medical order execution status and whether the surgical process node meets the update conditions. This constitutes a continuous processing relationship of medical order input, structured processing, process association, conditional execution, status feedback and anomaly confirmation.

[0011] This invention also provides an intelligent intraoperative physician order assistance system, which provides operational and execution support for the aforementioned intelligent intraoperative physician order assistance agent. The intelligent intraoperative physician order assistance system includes data processing resources for carrying medical order information and surgical procedure-related data, as well as processing components for data interaction with the intelligent intraoperative physician order assistance agent. The intelligent intraoperative physician order assistance agent performs structured processing of intraoperative physician orders and associates them with surgical procedure nodes. Based on execution confirmation information or result feedback information, it completes feedback processing of the medical order execution status and the surgical procedure node status, enabling the medical order processing logic formed by the agent to be carried and executed in the system.

[0012] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: This invention uses the continuous intraoperative operation scenario of the surgeon as the intelligent assistance object. A medical order semantic processing unit parses the natural language medical orders input by the surgeon into structured medical order execution items containing medical order type and execution information, transforming unstructured expressions into data objects that can be verified and controlled. A surgical context association unit combines surgical information and surgical process nodes to call medical order rules for verification, and establishes an association between the verified structured medical order execution items and the surgical process nodes, allowing the medical order processing results to enter the surgical process control chain. By comparing the identity data of the surgeon, anesthesiologist, and circulating nurse with personnel identity and professional authority data to form a three-party verification status, this three-party verification status becomes a prerequisite for the corresponding surgical process to enter the allowed-to-progress state, thus transforming personnel qualification verification from an independent information confirmation process into a surgical process control condition. Based on the current surgical process node, the corresponding personnel and equipment resources are determined, and based on the personnel arrival status, equipment availability status, and the verification results of the structured medical order execution items, it is determined whether the corresponding resources can be called, enabling the surgeon to... The surgeon can trigger resource scheduling corresponding to the current surgical step simply by issuing medical orders in natural language, without having to operate different terminals separately or independently confirm the status of multiple systems. The medical order execution control unit controls whether the medical order enters the execution state based on the verification results. The status feedback processing unit updates or maintains the status of surgical process nodes based on the execution status returned by the nurse, and synchronously incorporates the status of personnel resources, equipment resources, and abnormal handling results into the current surgical context. This allows the intelligent agent to determine the medical orders that can be executed and the surgical process nodes that can be advanced based on the updated context. This forms an intraoperative intelligent auxiliary processing chain with the surgeon's medical orders as input, third-party identity verification and the current surgical context as constraints, personnel and equipment resource calls as one of the execution objects, and nurse execution feedback and resource status feedback as closed-loop information. This enables the surgeon to complete the issuance of medical orders, resource coordination, and process status confirmation while maintaining the continuity of surgical operations, reducing the surgeon's need for manual operation and repeated confirmation on multiple business terminals, and allowing personnel, medical orders, resources, and surgical process status to be traced along the same data relationship. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A schematic diagram of the overall module interaction of the intelligent assistant for intraoperative physician orders; Figure 2A schematic diagram illustrating natural language medical order parsing and structured medical order execution item generation; Figure 3 A schematic diagram illustrating the verification of structured medical order execution items and the association of surgical process nodes; Figure 4 A schematic diagram of the structured medical order execution status and abnormal feedback closed loop; Figure 5 This is a schematic diagram illustrating the deployment and data interaction of an intelligent auxiliary system for intraoperative physician orders. Detailed Implementation

[0015] The technical solution described in this invention will be explained below with reference to the accompanying drawings. The intelligent assistant for intraoperative physician orders of this invention targets the continuous operational needs of the surgeon during the surgical process. During the operation of the intelligent assistant, it continuously maintains the operational context corresponding to the current surgery and selects corresponding processing actions based on the surgeon's natural language orders, personnel verification results, current surgical process nodes, resource status, and execution feedback.

[0016] The intraoperative physician-assisted intelligent agent performs semantic processing on the natural language medical orders entered by the surgeon during the operation, and establishes a continuous data relationship between the semantic processing results and the current surgical information, surgical process nodes, and execution feedback. The agent does not simply call each processing module in a fixed order, but after obtaining the current surgical context, it determines the processing action to be executed based on the three-party verification status, surgical process node status, structured medical order execution item status, personnel resource status, and equipment resource status recorded in the current context, and then calls the data, personnel terminal, or equipment interface corresponding to that processing action.

[0017] The intraoperative physician order intelligent assistance system is used to carry and run the intraoperative physician order intelligent assistance intelligent agent. It forms a data input, processing, execution feedback and status update link through physician terminal, nurse terminal, identity recognition device, storage unit and hospital information system interface. It can also obtain the device identifier, availability status and execution feedback of surgical equipment through device interface to complete the device resource call corresponding to the current surgical process node.

[0018] The term "natural language medical order" as used in this article refers to medical order information generated by the surgeon or a physician authorized by the current surgical permissions through voice or touchscreen input, containing instructions for medication, blood transfusion, consumable use, or operation. "Structured medical order execution item" refers to a data object formed from a natural language medical order after semantic parsing, containing at least the medical order type and execution information, and capable of being associated with a surgical workflow node. "Surgical workflow node" refers to a workflow state object determined based on the sequence of steps corresponding to the current surgery. "Medical order execution status" refers to the state of a structured medical order execution item during issuance, receipt confirmation, execution confirmation, and result feedback. "Abnormal record" refers to a data record formed when the execution information does not match the structured medical order execution item or when the completion conditions of a surgical workflow node are not met. "Resource status" refers to the personnel or equipment resources corresponding to the current surgical workflow node being in at least one of the following states: available, invoked, executing, completed, or abnormal.

[0019] Before the intelligent agent runs, the intelligent agent assisting the surgeon's orders during the operation is initialized and built. The initialization and building of the intelligent agent includes at least the configuration of the running target, the configuration of context data, the loading of rules and knowledge data, resource registration, and action mapping.

[0020] The execution target configuration is used to limit the processing target of the current agent to: structuring medical orders, validating rules, associating processes, calling resources, providing execution feedback, and updating process status, all based on the current surgeon's intraoperative medical orders and in accordance with personnel permissions and surgical procedure conditions.

[0021] Context data configuration is used to establish the runtime context data object corresponding to the current surgery. The runtime context data object includes at least the surgery identifier, surgery type, current surgical workflow node, third-party verification status, current structured medical order execution item, execution status of each medical order, personnel resource status corresponding to the current workflow node, and equipment resource status. The agent reads the same runtime context data object when processing different inputs and updates the corresponding status fields after the processing action is completed.

[0022] The rules and knowledge data loading process is used to configure the intraoperative medical order dictionary, medical order rules, and multi-specialty surgical step time sequence data into data resources that the intelligent agent can access. Specifically, the intraoperative medical order dictionary is used for mapping standard terminology in natural language medical orders; medical order rules are used to determine whether structured medical order execution items meet the current patient and surgical conditions; and multi-specialty surgical step time sequence data is used to determine the current surgical procedure node and the completion conditions corresponding to each procedure node.

[0023] Resource registration is used to establish resource identification and invocation relationships between data sources and execution objects that can be invoked during the operation of the intelligent agent. Data sources include at least hospital information system interfaces, identification devices, and storage units; execution objects include at least nurse terminals and personnel and equipment resources corresponding to surgical process nodes. For equipment resources, equipment identifiers, equipment types, equipment interfaces, and equipment availability status are pre-recorded; for personnel resources, personnel identities, roles, professional permissions, and current arrival status are recorded.

[0024] Action mapping is used to establish the correspondence between the agent's state judgment results and processing actions. Processing actions include at least: generating structured medical order execution entries, performing medical order rule verification, performing third-party personnel verification, establishing surgical workflow node associations, sending structured medical order execution entries to the nurse's terminal, calling corresponding personnel or equipment resources, maintaining the medical order in a non-execution state, updating the surgical workflow node state, and generating exception records. The agent selects the corresponding processing action based on the conditions satisfied by the runtime context and updates the runtime context again after the processing action generates feedback.

[0025] Through the above construction process, the intelligent agent can form a continuous processing relationship during operation: "input perception - context reading - condition judgment - action selection - resource call - feedback reception - context update".

[0026] Example 1 Please see Figure 1-4 This embodiment provides an intelligent assistant for intraoperative physician orders. The intelligent assistant takes the surgeon as the main interaction object and receives the surgeon's natural language orders during the continuous operation of the surgery. Based on the current surgical context, it automatically completes the parsing of the orders, status judgment, and selection of subsequent resource call actions.

[0027] The intraoperative physician-assisted intelligent agent comprises a semantic processing unit for medical orders, a surgical context association unit, a medical order execution control unit, and a status feedback processing unit. The semantic processing unit receives natural language medical orders input by the physician, performs semantic parsing of the natural language orders, and forms structured medical order execution items. The surgical context association unit obtains the surgical information and surgical procedure nodes corresponding to the current surgery, invokes medical order rules to verify the structured medical order execution items, and establishes the association between the structured medical order execution items and the surgical procedure nodes when the verification is successful. The medical order execution control unit controls whether the corresponding medical order enters the execution state based on the verification result. The status feedback processing unit receives the execution state corresponding to the structured medical order execution item and updates or maintains the state of the surgical procedure node based on whether the execution state meets the completion conditions of the surgical procedure node.

[0028] In the aforementioned unit collaboration process, the agent uses the runtime context as the common state data among the processing units. When any unit generates a new verification result, association result, execution state, or process node state, it writes the corresponding result into the runtime context; subsequent units read the updated runtime context before determining the processing action, thereby enabling different processing units to continuously execute around the current surgical state.

[0029] Please see Figure 2 The input to the medical order semantic processing unit is natural language medical orders generated by doctors through voice input or touch screen input. For voice input, the natural language medical order is first converted into text content that can be semantically processed; for touch screen input, the text content entered by the doctor is directly used as the semantic processing object. The medical order semantic processing unit identifies the medical order type, execution object, and execution content based on the text content. The medical order type indicates the business category to which the current medical order belongs, and the execution content indicates the specific content that needs to be executed. After recognition, the medical order semantic processing unit writes the medical order type, execution object, and execution content into the corresponding business fields, and combines them according to the correspondence between the business fields to form structured medical order execution entries.

[0030] The medical order semantic processing unit invokes the intraoperative medical order dictionary during semantic recognition. The intraoperative medical order dictionary stores standard terms and corresponding semantic relationships according to the medical order type corresponding to medication, blood transfusion, consumable use, and operational instructions. Standard terms are used to map natural language expressions to terminology content that can be read by the same business field, and corresponding semantic relationships are used to define the correspondence between natural language words and medical order types, execution objects, and execution content. The medical order semantic processing unit first determines candidate medical order types based on target words in the natural language medical order, and then performs matching in the set of standard terms corresponding to the candidate medical order types. When a correspondence exists between a target word and a standard term, the corresponding standard term is written into the business field; when a target word cannot establish a matching relationship with a standard term in the intraoperative medical order dictionary, the natural language content corresponding to the target word is retained, and the corresponding business field remains in a pending verification state for further processing in subsequent verification processes.

[0031] When multiple words in a single natural language medical order correspond to the same business field, the medical order semantic processing unit does not directly write these multiple words into the same business field. Instead, it determines the field content based on the semantic relationship between the multiple words and the execution object. The semantic relationship is based on the semantic correspondence between the target words in the natural language medical order and the execution object. Words that can form a correspondence with the execution object are used as the field content of the current business field; words that cannot form a correspondence are not used as the determination result of the current business field. The structured medical order execution entry formed through the above processing includes at least the medical order type and execution information. The execution information consists of the execution object and execution content corresponding to the current medical order. The structured medical order execution entry is then sent to the surgical context association unit.

[0032] Please see Figure 2 and Figure 3 During field-level validation, the medical order semantic processing unit determines the corresponding validation fields based on the medical order type in the structured medical order execution entry. For structured medical order execution entries of medication type, the drug name, dosage, and execution content are used as the content to be validated, and this content is matched with the patient information and medical order rules corresponding to the current surgery. For structured medical order execution entries of blood transfusion, consumable use, or operation instructions, the execution object and execution content are used as the content to be validated, and these are matched with the corresponding medical order rules. When all fields to be validated satisfy the corresponding medical order rules, a validation pass result is generated; when any field to be validated does not satisfy the corresponding medical order rules, a validation fail result is generated, and the business field that does not satisfy the medical order rules is marked as a field to be confirmed. The field-level matching result and the structured medical order execution entry are passed to the surgical context association unit, so that subsequent association processing has definite medical order content and validation status.

[0033] Please see Figure 3 The surgical context association unit acquires the surgical information for the current surgery. This surgical information is used at least to determine the surgical type and the current surgical procedure status. Based on the surgical type, the surgical context association unit determines the sequence of steps corresponding to the current surgery from pre-stored multi-specialty surgical step timing data. The multi-specialty surgical step timing data stores the sequence of steps corresponding to different surgical types, and each surgical step has a data relationship that represents its order and completion status. The surgical context association unit reads the completed surgical steps from the step sequence and determines the current surgical procedure node based on the timing relationship between the completed surgical steps and the next surgical step to be executed.

[0034] After receiving a structured medical order execution entry, the surgical context association unit invokes the medical order rules based on the surgical information and combines the field-level validation results to form a validation conclusion for the current structured medical order execution entry. If the validation passes, the surgical context association unit matches the structured medical order execution entry to the current surgical workflow node according to the medical order type and execution information in the structured medical order execution entry. During matching, the correspondence between the execution conditions corresponding to the current surgical workflow node and the medical order type, execution object, and execution content of the structured medical order execution entry is used as the judgment basis. When the current surgical workflow node matches the execution conditions of the structured medical order execution entry, an association is established between the structured medical order execution entry and the current surgical workflow node; when the current surgical workflow node does not match the execution conditions, no execution association is established, and the structured medical order execution entry remains in a non-execution state. If the validation fails, the corresponding medical order is also not put into the execution state.

[0035] The surgical context association unit records the completion status of each surgical step according to the step sequence. For surgical steps corresponding to specimen verification, blood transfusion confirmation, and patient positioning, the corresponding surgical steps are identified as process nodes that require confirmation information. The confirmation information is part of the completion conditions of the corresponding process node. When the current process node obtains the corresponding confirmation information, and the structured medical order execution item associated with the current process node meets the execution conditions, the surgical context association unit switches the next process node to the allowed-to-progress state; when the current process node does not obtain the corresponding confirmation information, or the structured medical order execution item associated with the current process node does not meet the execution conditions, the next process node remains in the non-progressable state. Thus, the progression status of the step sequence is jointly determined by the completion status of the current process node and the medical order execution conditions associated with the current process node.

[0036] In addition to determining the current surgical procedure node, the surgical context association unit also determines the corresponding resource requirement information based on the current surgical procedure node. The resource requirement information indicates the personnel roles and equipment types required to complete the current procedure node. The surgical context association unit reads the corresponding personnel resource status and equipment resource status from the runtime context based on the resource requirement information.

[0037] For example, when the current surgical procedure node corresponds to anesthesia, the anesthesiologist is identified as one of the corresponding personnel resources, and the anesthesiologist's verification status, which has been verified by three parties, is read; when the current surgical procedure node requires the use of electrosurgical equipment, endoscopic equipment, infusion equipment, blood transfusion-related equipment, or other surgical equipment, the equipment to be called is determined according to the equipment type corresponding to the procedure node in the step sequence, and the availability status of the corresponding equipment is read.

[0038] When the corresponding personnel have the identity and professional authority matching the current surgery, are in place, and the corresponding equipment is available, the corresponding resource status is recorded as meeting the call conditions; if any necessary personnel or equipment resource does not meet the call conditions, the current process node is kept in an unprogressable state. The resource status, as part of the completion or execution conditions of the current surgical process node, is used together with the verification results of the structured medical order execution items for subsequent execution control.

[0039] The medical order execution control unit receives the structured medical order execution entries, verification conclusions, and surgical procedure node association results output by the surgical context association unit. When the verification conclusion is passed and the structured medical order execution entry has been associated with an authorized surgical procedure node, the medical order execution control unit outputs the corresponding structured medical order execution entry, putting the corresponding medical order into the execution state.

[0040] When the execution content corresponding to a structured medical order involves personnel coordination, equipment preparation, or equipment operation, the medical order execution control unit further generates a resource retrieval instruction based on the resource requirement information corresponding to the current surgical process node. For personnel resources, the resource retrieval instruction includes at least the corresponding personnel role and the current process node; for equipment resources, the resource retrieval instruction includes at least the equipment identifier or equipment type and the operation to be performed.

[0041] Resource retrieval instructions are only output when the tripartite verification status meets the process conditions and the corresponding personnel or equipment resources are in an authorized retrieval state. When the equipment to be retrieved is unavailable, the corresponding personnel's professional permissions do not meet the current surgical requirements, or the necessary personnel are not in place, the medical order execution control unit will not output the corresponding resource retrieval instructions and will keep the structured medical order execution entries in a non-execution state or a pending confirmation state.

[0042] When the verification result is "failed," the medical order execution control unit prevents the corresponding medical order from entering the execution state; when the verification result is "passed" but the corresponding surgical procedure node has not yet met the execution conditions, the medical order execution control unit keeps the structured medical order execution item in a non-execution state. Therefore, the medical order execution control unit uses the tripartite verification status, the medical order verification result, the surgical procedure node status, and the necessary resource status as continuous judgment conditions, rather than solely relying on the parsing results of the natural language medical order to output the execution content.

[0043] Please see Figure 4The status feedback processing unit receives reception confirmation information, execution confirmation information, and result feedback information corresponding to the structured medical order execution item, and determines the medical order execution status in the order of order issuance, reception confirmation, execution confirmation, and result feedback. Order issuance indicates that the structured medical order execution item has been output by the medical order execution control unit; reception confirmation indicates that the corresponding nurse terminal has received the structured medical order execution item; execution confirmation indicates that the nurse terminal has returned execution confirmation information for the corresponding medical order content; result feedback indicates that the nurse terminal has returned the execution result of the corresponding medical order. These statuses form a status chain for the same structured medical order execution item in a predetermined order, enabling subsequent processing to identify the execution stage that the current medical order has reached.

[0044] After receiving execution confirmation or result feedback information, the status feedback processing unit matches this information with the execution information in the corresponding structured medical order execution entry. The matching object must include at least information indicating whether the execution object and execution content correspond. After matching the execution information, the status feedback processing unit further compares the matching result with the completion conditions of the corresponding surgical procedure node. When the execution information matches and both the execution confirmation and result feedback information meet the completion conditions of the corresponding surgical procedure node, the status feedback processing unit determines that the corresponding medical order execution status is valid and updates the corresponding surgical procedure node to a completed status. The surgical context association unit reads the updated procedure node status and determines whether subsequent procedure nodes meet the conditions for advancement based on the step sequence.

[0045] For structured medical order execution entries involving the use of personnel or equipment resources, the status feedback processing unit also receives resource feedback information after the corresponding resource use. Personnel resource feedback information indicates that the corresponding personnel have responded, arrived, or completed the corresponding operation; equipment resource feedback information indicates that the corresponding equipment has been prepared, started, completed execution, or encountered an error.

[0046] The status feedback processing unit associates resource feedback information with the current structured medical order execution item and the corresponding surgical procedure node. When the medical order execution information matches, the necessary personnel resource feedback meets the requirements of the current procedure node, the necessary equipment resource feedback meets the requirements of the current procedure node, and all other completion conditions are met, the corresponding surgical procedure node is updated to a completed state. If any necessary resource feedback does not meet the conditions, the corresponding surgical procedure node remains in an incomplete state, and an exception record is generated based on the resource that did not meet the conditions.

[0047] When the execution confirmation information does not match the execution information in the structured medical order execution entry, or the result feedback information does not match the execution information in the structured medical order execution entry, or the completion conditions of the corresponding surgical procedure node are not met, the corresponding surgical procedure node remains in an incomplete state, while the corresponding structured medical order execution entry remains in a pending confirmation state. The status feedback processing unit generates an exception record based on the mismatch. The exception record is associated with at least the mismatched execution information and the corresponding surgical procedure node, enabling the exception record to be traced back to the structured medical order execution entry and specific procedure location where the inconsistency occurred.

[0048] For structured medical order execution entries that generate anomaly records, the status feedback processing unit associates the structured medical order execution entry, execution confirmation information, result feedback information, and the corresponding surgical procedure node, and outputs the anomaly record. After the anomaly record is processed, the status feedback processing unit receives the processing result for the anomaly record and makes another judgment based on the consistency of the execution information and the completion conditions of the surgical procedure node. When the processing result indicates that the execution information is consistent with the structured medical order execution entry and the completion conditions of the corresponding surgical procedure node are met, the status feedback processing unit updates the execution status of the corresponding medical order to a valid status and updates the corresponding surgical procedure node to a completed status. When the processing result still does not meet the completion conditions of the corresponding surgical procedure node, the status feedback processing unit continues to keep the surgical procedure node in an incomplete state and keeps the structured medical order execution entry in a pending confirmation state. The processing result only changes the status of the corresponding anomaly record and does not change the status of other structured medical order execution entries.

[0049] During the above operation, the natural language medical orders entered by the surgeon are used as input. The medical order semantic processing unit forms structured medical order execution items. The structured medical order execution items are then verified by the surgical context association unit to complete the medical order rule verification and surgical process node association. Finally, the medical order execution control unit determines whether to output based on the three-party verification status, verification conclusion, surgical process node status, and corresponding resource status.

[0050] When a structured medical order execution item involves specific personnel cooperation or equipment operation, the intelligent agent determines the corresponding personnel or equipment resources based on the current process node and generates a corresponding resource call action. The reception confirmation information, execution confirmation information, and result feedback information returned by the nurse terminal, as well as the resource status information returned by the personnel or equipment, are used by the status feedback processing unit to form the medical order execution status and resource execution status, and further affect the status of the surgical process node.

[0051] After an exception record is generated for results that do not match execution information, resource call anomalies, or fail to meet completion conditions, the result of the handling will re-enter the status judgment chain.

[0052] Each time the agent completes the above judgment or action, it writes the processing result back to the current surgical execution context and determines the next allowed medical order, resource to be called, or process node to be advanced based on the updated execution context. This forms a closed-loop agent processing relationship from verification by three parties, natural language medical order input by the surgeon, structured processing, process association, resource call, execution feedback to the return flow of abnormal handling results.

[0053] In this embodiment, whether a structured medical order execution item forms a correct processing result can be identified by whether the fields of medical order type, execution object, and execution content in the structured medical order execution item form definite content; the process association result can be identified by whether the corresponding structured medical order execution item has a definite surgical process node association status; the execution control result can be identified by whether the structured medical order execution item is in an execution state, a non-execution state, or a pending confirmation state; and the feedback result can be identified by whether the surgical process node is in a completed state or an incomplete state and whether an abnormal record is generated.

[0054] Example 2 Please see Figure 5 This embodiment provides an intelligent intraoperative physician order assistance system to support and run the aforementioned intelligent intraoperative physician order assistance agent. Compared with Embodiment 1, the difference in this embodiment lies in deploying the aforementioned order processing, process association, execution control, and status feedback logic into a data interaction environment composed of a physician terminal, an agent running device, a nurse terminal, an identity recognition device, a storage unit, and a hospital information system interface. Furthermore, it incorporates personnel and equipment resources that can be called upon by the agent during the operation into the operating environment, enabling the agent to determine, call upon, and provide status feedback based on the surgeon's orders and the current surgical process node.

[0055] In this embodiment, the surgeon interacts primarily with the intelligent agent through a doctor's terminal; the operator, anesthesiologist, and circulating nurse participate in tripartite verification through an identity recognition device; the nurse's terminal serves as the receiver and feedback end for structured medical order execution items; the hospital information system interface provides current surgical and personnel permission data; the storage unit saves the rules, step sequence data, and operating status required for the intelligent agent's operation; and devices related to the current surgical procedure provide device status information or receive resource call instructions to the intelligent agent's operating devices through corresponding device interfaces.

[0056] The doctor's terminal acts as the input end for natural language medical orders, sending the natural language medical orders generated by the doctor through voice or touchscreen input to the intelligent agent running device. The intelligent agent running device operates the intraoperative doctor's medical order intelligent assistant and inputs the received natural language medical orders into the medical order semantic processing unit. After the medical order semantic processing unit forms a structured medical order execution item, the structured medical order execution item is transmitted to the surgical context association unit within the intelligent agent running device. In this embodiment, the doctor's terminal is responsible for inputting medical orders and sending medical order information, but does not undertake the process status judgment of the structured medical order execution item.

[0057] The hospital information system interface is used to establish a data exchange path between the intelligent agent operating device and the hospital information system. The intelligent agent operating device obtains surgical information corresponding to the current surgery and personnel identity and professional authorization data through the hospital information system interface. The surgical information of the current surgery is used by the surgical context association unit to determine the surgical type and invoke the corresponding step sequence; the personnel identity and professional authorization data are used to verify the identity recognition data obtained by the identity recognition device. The hospital information system interface transmits the obtained corresponding data to the intelligent agent operating device, which then provides the data to the surgical context association unit and the personnel verification process according to the different uses of the data.

[0058] The identification device is used to obtain the identification data of the surgeon, anesthesiologist, and circulating nurse. The device uses facial recognition, card swiping, or QR code scanning to obtain the identification data of the corresponding personnel. After receiving the identification data, the intelligent agent operating device compares the identification data of the surgeon, anesthesiologist, and circulating nurse with the personnel identity and professional authorization data obtained through the hospital information system interface, and establishes a three-way verification status according to the verification order of surgeon, anesthesiologist, and circulating nurse.

[0059] When the lead surgeon acts as the operator, the surgeon's identity verification result is used to determine both their current role in the surgery and whether their input natural language medical orders have the authority to enter the subsequent intelligent agent processing chain. The verification results of the anesthesiologist and circulating nurse are used to determine whether they can participate in subsequent process nodes as the corresponding personnel resources for the current surgery.

[0060] Once the identification data of each person is matched with the corresponding person's identity and professional authority data, the verification status of the corresponding person is recorded as passed; if the verification of any person is not completed or the verification result is not passed, the tripartite verification status does not meet the process conditions.

[0061] The tripartite verification status serves as a prerequisite for the intelligent agent's operating equipment to enter the execution state of subsequent surgical procedures. The intelligent agent's operating equipment writes the tripartite verification status into the current surgical operation context and reads this status each time it determines whether a structured medical order execution item is allowed to be executed and whether personnel resources are allowed to be accessed.

[0062] After the tripartite verification status meets the process conditions, the intelligent agent operating device continues to perform medical order rule verification and surgical process node association on the structured medical order execution items according to the processing method in Implementation Example 1. When the structured medical order execution item passes the verification and the corresponding surgical process node has the execution conditions, the intelligent agent operating device sends the corresponding structured medical order execution item to the nurse terminal. When the tripartite verification status does not meet the process conditions, the intelligent agent operating device does not allow the subsequent surgical process to enter the allowed progress state, and the corresponding structured medical order execution item remains in the non-execution state.

[0063] When the reason for not meeting the process conditions is that the corresponding personnel have not completed identity verification, the corresponding professional authority does not meet the current surgical requirements, or the necessary personnel are not in place, the intelligent agent operation device will record the status of the corresponding personnel as the unmet condition of the current process node, and re-determine the conditions after obtaining new identity verification results or personnel arrival information.

[0064] The intelligent agent operating equipment also determines the corresponding personnel and equipment resources based on the current surgical procedure node. In addition to storing the sequential relationship between each surgical step, the multi-specialty surgical procedure timing data can also store the resource requirement relationship corresponding to each surgical step. The resource requirement relationship is used to determine at least the personnel roles and equipment types required to perform the corresponding surgical step.

[0065] Once the intelligent agent operating device determines the current surgical procedure node, it forms a set of resources to be invoked based on the resource demand relationship corresponding to that node. For personnel resources, the intelligent agent operating device determines whether the corresponding personnel meet the invocation conditions based on the third-party verification status, personnel identity and professional authority data, and personnel arrival status. For equipment resources, the intelligent agent operating device determines whether the corresponding equipment meets the invocation conditions based on the equipment identifier, equipment type, and current equipment status.

[0066] For example, when the current surgical procedure requires the participation of an anesthesiologist, the intelligent agent operating device will identify the anesthesiologist who has been verified for identity and professional authority as the personnel resource to be called upon; when the current surgical procedure requires the use of a specific surgical device, the device that corresponds to that device type and is in an available state will be identified as the device resource to be called upon.

[0067] When the content to be executed in the surgeon's natural language medical order corresponds to the resource to be invoked, the medical order execution control unit generates a corresponding resource invocation instruction. Personnel resource invocation can be achieved by sending structured execution entries through the nurse's terminal or the corresponding personnel's terminal; equipment resource invocation can be achieved by sending equipment preparation, status query, or operation instructions corresponding to the current structured medical order execution entry through the equipment interface.

[0068] For resource calls involving actual device actions, the device interface only executes the corresponding interface operation based on the structured operation fields that have been judged by the agent and conform to the preset device operation rules; the agent does not directly control the device with unstructured and unverified natural language text.

[0069] After the device executes, it returns device status information; after the personnel execute, it returns personnel response or execution status information. The intelligent agent running the device writes the above resource status information into the current surgical operation context, and the status feedback processing unit jointly determines whether the current process node meets the completion conditions based on the resource status and the structured medical order execution items.

[0070] The nurse terminal receives structured medical order execution entries sent by the intelligent agent operating device and returns reception confirmation information, execution confirmation information, and result feedback information to the intelligent agent operating device based on the actual execution process. The intelligent agent operating device inputs the information returned by the nurse terminal into the status feedback processing unit and forms the medical order execution status according to the status determination method in Embodiment 1. When the execution information matches and the completion conditions of the surgical procedure node are met, the intelligent agent operating device updates the status of the surgical procedure node associated with the corresponding structured medical order execution entry; when the execution information does not match or the completion conditions are not met, the intelligent agent operating device keeps the corresponding surgical procedure node in an incomplete state and generates an exception record.

[0071] The storage unit is used to store the surgical order dictionary, medical order rules, multi-specialty surgical step sequence data, as well as the structured medical order execution entries, medical order execution status, surgical process node status and abnormal records generated during the processing. It also stores personnel resource status, equipment resource information, equipment resource status and the current surgical operation context.

[0072] The surgical order dictionary is invoked by the medical order semantic processing unit during the natural language medical order semantic parsing process. The medical order rules are invoked by the medical order semantic processing unit and the surgical context association unit during the field matching and verification process. The multi-specialty surgical step time sequence data is invoked by the surgical context association unit according to the surgical type.

[0073] The time sequence data of multi-specialty surgical procedures can also establish the relationship between surgical process nodes and corresponding resource requirements, which can be used to determine the personnel roles and equipment types required for each process node.

[0074] Structured medical order execution items, medical order execution status, surgical process node status and abnormal records, personnel resource status and equipment resource status are associated according to the corresponding surgical process, so that data objects formed within the same surgical process can be read by subsequent processing along the same data relationship.

[0075] The current surgical execution context is used to store the set of states on which the agent made its current judgment, and is updated based on the new state information after the medical order processing, resource call or feedback processing is completed.

[0076] In the data transmission chain of this embodiment, the surgeon outputs natural language medical orders through the doctor's terminal, and the intelligent agent running device generates structured medical order execution entries. The hospital information system interface provides surgical information, personnel identity and professional permission data to the intelligent agent running device, and the identity recognition device provides the intelligent agent running device with the identity recognition data of the surgeon, anesthesiologist, and circulating nurse. The intelligent agent running device establishes a three-way verification status. The intelligent agent running device determines the current surgical process node based on the surgical information and determines the corresponding personnel and equipment resources based on the current surgical process node. When the three-way verification status meets the process conditions, the structured medical order execution entries pass the verification, and the corresponding surgical process node meets the execution conditions, the structured medical order execution entries are sent to the nurse's terminal, or a resource call instruction is generated for the corresponding personnel or equipment resources. After the nurse's terminal returns the execution status information, and the corresponding personnel or equipment returns the resource status information, the intelligent agent running device updates the corresponding medical order execution status, resource status, and surgical process node status. The aforementioned three-party verification status, structured medical order execution items, current surgical workflow nodes, resource status, and execution feedback are all written into the current surgical operation context. The intelligent agent running device continues to determine the next processing action to be executed based on the updated operation context. All of the above inputs, intermediate processing states, and outputs correspond to the intelligent agent processing chain in Example 1.

[0077] The system operation results in this embodiment can be identified through the following states: After the doctor's terminal generates and sends a natural language medical order, the intelligent agent operating device has a structured medical order execution item associated with the corresponding natural language medical order; after the identity recognition device completes the identity recognition of the three parties and compares the personnel identity and professional permission data, the intelligent agent operating device generates a corresponding three-party verification state; a corresponding personnel resource and equipment resource state is generated according to the current surgical process node; after the structured medical order execution item is sent to the nurse's terminal, the nurse's terminal returns the corresponding execution state; when resource calls are involved, the corresponding personnel or equipment generates a resource feedback state; after the execution state and necessary resource state meet the completion conditions of the associated surgical process node, the corresponding surgical process node is updated to the completed state; when the execution state or necessary resource state does not meet the completion conditions, the corresponding surgical process node remains in the incomplete state. The correspondence between the above states is stored in the operating context. From these observable states, it can be determined that the input, processing, intermediate states, resource calls, outputs, and feedback in the system have formed a continuous relationship.

[0078] In the foregoing embodiments, the intraoperative medical order dictionary, medical order rules, and multi-specialty surgical step sequence data are all invoked according to the data usage defined by the technical solution described in this invention. For content without a fixed quantity, fixed threshold, fixed field length, or fixed data size, the intraoperative medical order dictionary is defined as capable of establishing a correspondence between natural language words and standard terminology; the medical order rules are defined as capable of forming a satisfaction or non-satisfaction judgment for the fields to be verified for the corresponding medical order type; and the multi-specialty surgical step sequence data is defined as capable of determining the step sequence based on the surgical type and determining the surgical process nodes based on the step completion status. This does not change the role of the aforementioned data objects in the processing chain.

[0079] The specific data fields, processing order, status names, unit collaboration relationships, terminal interaction relationships, and system carrying relationships in the foregoing embodiments are used to explain the possible implementations of the present invention and are not intended to limit the present invention to the specific implementation forms listed. Without departing from the technical solutions described in this invention, any substitutions, modifications, or combinations made by those skilled in the art regarding the medical order input carrier, equivalent data field names, equivalent execution entities, equivalent data transmission forms, equivalent unit correspondences, and equivalent carrier forms, as long as the technical relationships of the surgeon's natural language medical orders forming structured medical order execution items, the verification status of third-party personnel forming a pre-constraint on medical order execution or process advancement, the establishment of association between structured medical order execution items and surgical process nodes, the determination and invocation of corresponding personnel or equipment resources based on surgical process nodes, the impact of execution status and resource feedback on the status of surgical process nodes, and the return of abnormal records to the treatment judgment chain, all fall within the scope of protection of this patent.

Claims

1. An intelligent assistant for intraoperative physician orders, characterized in that: It includes a medical order semantic processing unit, a surgical context association unit, a medical order execution control unit, and a status feedback processing unit; The medical order semantic processing unit obtains the natural language medical orders input by the doctor, performs semantic parsing on the natural language medical orders, and generates structured medical order execution entries containing medical order type and execution information; The surgical context association unit obtains the surgical information and surgical process nodes of the current surgery, calls the medical order rules to verify the structured medical order execution items according to the surgical information, and establishes an association between the verified structured medical order execution items and the surgical process nodes; The medical order execution control unit outputs the structured medical order execution item when the verification passes, and prevents the corresponding medical order from entering the execution state when the verification fails. The status feedback processing unit obtains the execution status of the structured medical order execution item. When the execution status meets the completion conditions of the surgical procedure node, the surgical procedure node is updated. When the execution status does not meet the completion conditions, the surgical procedure node is kept in an incomplete state.

2. The intelligent assistant for intraoperative physician orders according to claim 1, characterized in that, The semantic processing unit for medical orders includes the following steps for semantic parsing of natural language medical orders: textualizing the natural language medical orders generated by doctors through voice or touchscreen input; identifying the order type, execution object, and execution content based on a pre-defined intraoperative medical order dictionary and business field mapping relationship; writing the identified order type, execution object, and execution content into corresponding business fields, and combining them according to the mapping relationship between fields to form the structured medical order execution entries; when the target words in the natural language medical order can establish a matching relationship with the standard terms in the intraoperative medical order dictionary, using the corresponding standard terms to form the business field content; when a matching relationship cannot be established, retaining the corresponding natural language content for subsequent verification.

3. The intelligent assistant for intraoperative physician orders according to claim 2, characterized in that, The intraoperative medical order dictionary stores standard terms and corresponding semantic relationships according to the medical order types corresponding to medication, blood transfusion, consumable use, and operation instructions. The medical order semantic processing unit determines the business field to be extracted according to the medical order type and matches the words in the natural language medical order with the standard terms under the corresponding medical order type. When multiple words that can correspond to the same business field are identified in the same natural language medical order, the field content corresponding to the business field is determined according to the semantic relationship of the words in the natural language medical order, and the determined field content is written into the structured medical order execution entry.

4. The intelligent assistant for intraoperative physician orders according to claim 3, characterized in that, The surgical context association unit determines the corresponding verification fields based on the medical order type in the structured medical order execution entries. It matches the drug name, dosage, and execution content in structured medical order execution entries of medication type with the patient information and medical order rules corresponding to the surgical information. It also matches the execution object and execution content in structured medical order execution entries of blood transfusion, consumable use, or operation instructions with the corresponding medical order rules. When all fields to be verified satisfy the corresponding medical order rules, the structured medical order execution entry is determined to have passed verification. When any field to be verified does not satisfy the corresponding medical order rules, the structured medical order execution entry is determined to have failed verification, and the field that does not satisfy the medical order rules is designated as a field to be confirmed.

5. The intelligent assistant for intraoperative physician orders according to claim 1, characterized in that, The surgical context association unit determines the sequence of steps corresponding to the current surgery from pre-stored multi-specialty surgical step time data based on the surgical type in the surgical information. It determines the surgical process node based on the surgical steps that have been completed in the step sequence, and matches the verified structured medical order execution entries to the surgical process node according to the medical order type and execution information. When the surgical process node matches the execution conditions of the structured medical order execution entry, an association relationship is established. When the surgical process node does not match the execution conditions, the structured medical order execution entry remains in a non-execution state.

6. The intelligent assistant for intraoperative physician orders according to claim 5, characterized in that, The surgical context association unit records the corresponding completion status according to the temporal relationship of each surgical step in the step sequence, and determines the surgical steps corresponding to specimen verification, blood transfusion confirmation, and patient positioning as process nodes that need to obtain confirmation information; when the current process node obtains the corresponding confirmation information and the structured medical order execution item associated with the current process node meets the execution conditions, the next process node is switched to the allowed progress state; when the current process node does not obtain the corresponding confirmation information or the structured medical order execution item associated with the current process node does not meet the execution conditions, the next process node is kept in the unprogressable state.

7. The intelligent assistant for intraoperative physician orders according to claim 1, characterized in that, The status feedback processing unit receives the reception confirmation information, execution confirmation information, and result feedback information corresponding to the structured medical order execution item, and determines the medical order execution status in the order of medical order issuance, reception confirmation, execution confirmation, and result feedback. When both the execution confirmation information and the result feedback information meet the completion conditions of the surgical procedure node associated with the structured medical order execution item, the surgical procedure node is updated to the completed state; when the execution confirmation information or the result feedback information does not meet the completion conditions, the surgical procedure node remains in the incomplete state, and the corresponding structured medical order execution item remains in the pending feedback state.

8. The intelligent assistant for intraoperative physician orders according to claim 7, characterized in that, When the status feedback processing unit receives execution confirmation information or result feedback information, it matches the execution confirmation information or result feedback information with the execution information in the corresponding structured medical order execution item, and compares the matching result with the completion conditions of the corresponding surgical process node; when the execution information matches and the completion conditions are met, it determines that the corresponding medical order execution status is valid and updates the surgical process node. When the execution information does not match or the completion conditions are not met, the corresponding surgical procedure node remains in an incomplete state, an exception record containing the mismatched execution information and the corresponding surgical procedure node is generated, and the corresponding structured medical order execution item remains in a pending confirmation state.

9. The intelligent assistant for intraoperative physician orders according to claim 8, characterized in that, The status feedback processing unit associates the structured medical order execution items, execution confirmation information, result feedback information and surgical process nodes corresponding to the abnormal records, outputs the abnormal records and receives the handling results for the abnormal records; When the handling result indicates that the execution information is consistent with the structured medical order execution item and the completion conditions of the surgical procedure node are met, the corresponding medical order execution status is updated to valid and the surgical procedure node is updated to completed. When the handling result still does not meet the completion conditions, the surgical procedure node remains in an incomplete state and the structured medical order execution item remains in a pending confirmation state.

10. An intelligent assist system for intraoperative physician orders, used to carry and run the intelligent assist system for intraoperative physician orders as described in any one of claims 1 to 9, characterized in that, It includes doctor terminals, intelligent agent operation devices, nurse terminals, identity recognition devices, storage units, and interfaces for hospital information systems; The doctor's terminal sends natural language medical orders to the intelligent agent running device. The intelligent agent running device runs the intelligent assistant intelligent agent that assists in the operation to generate structured medical order execution items, and obtains surgical information and personnel identity and practice permission data through the hospital information system interface; The identity recognition device acquires the identity recognition data of the surgeon, anesthesiologist, and circulating nurse. The intelligent agent operation device determines the three-party verification status based on the identity recognition data and the personnel's identity and professional authority data. When the three-party verification status meets the process conditions and the structured medical order execution item passes the verification, the device sends the structured medical order execution item to the nurse terminal. The nurse terminal returns to the execution status, and the intelligent agent running device updates the status of the surgical process nodes associated with the structured medical order execution items based on the execution status.

Images