Anesthesia quality control event identification method and system based on rule package and state machine

Abstract

This invention provides a method and system for identifying anesthesia quality control events based on rule packages and state machines, relating to the field of medical information processing technology. The invention first acquires multi-source anesthesia data and forms a unified event sequence. It then constructs a structured rule package containing event observation objects, time anchors, triggering conditions, duration conditions, termination conditions, exclusion conditions, evidence extraction rules, and alarm rules. Event judgment elements are extracted, an event state machine is constructed and state transitions are executed, quality control alarms are output, and quality control records are generated, thereby improving the consistency of anesthesia quality control event identification.

Description

Technical Field

[0001] This invention relates to the field of medical information processing technology, and in particular to a method and system for identifying anesthesia quality control events based on rule packages and state machines. Background Technology

[0002] Anesthesia information management systems have already been practically applied in perioperative anesthesia management. In their article, "The Role of Anesthesia Information Management Systems in Perioperative Quality Control," Zhang Wei, Chai Xiaoqing, and Xie Yanhu point out that anesthesia information management systems can automatically collect, organize, store, and statistically analyze perioperative information and anesthesia quality control indicators, and use them for process control, terminal control, and overall quality management. Therefore, existing technology already possesses the basic conditions for data collection, retention, and quality management during the anesthesia process.

[0003] In anesthesia quality management practice, in addition to routine monitoring data recording, the classification and reporting standards for anesthesia-related adverse events have already been researched. In their article "Exploring the Classification and Grading of Anesthesia-Related Adverse Events," Li Chao, Ke Jingdong, Wang Fang, and others, combining regulations, expert consensus, and clinical practice, provided a classification, grading, and definition of anesthesia-related adverse events, offering a reference for clinicians in completing and reporting such information. This indicates that existing technology has a certain foundation for classifying and recording anesthesia quality control events.

[0004] However, in actual quality control work, the same anesthesia quality control event is often not directly identifiable based on a single monitoring indicator. Instead, it usually requires a comprehensive judgment based on changes in monitoring parameters, anesthesia records, and treatment procedure records. While current technology can collect, record, and statistically analyze the aforementioned information separately, the relevant judgment criteria for a specific quality control event are still scattered across different record items. Quality control personnel typically need to manually compare these records during review to determine whether the event occurred and how the start and end points of the event should be defined. This can easily lead to inconsistent judgment methods when the same event is reviewed by different personnel. In other words, under current technological conditions, the organizational method for judging individual quality control events is still not direct enough. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, the purpose of this invention is to provide an anesthesia quality control event identification method and system based on rule packages and state machines. By using rule packages and state machines to uniformly determine anesthesia quality control events, the consistency of anesthesia quality control event identification can be improved, and it is convenient to form quality control records with clear judgment criteria.

[0006] To achieve the above objectives, the present invention provides the following solution: A method for identifying anesthesia quality control events based on rule packages and state machines includes: Acquire multi-source anesthesia data from the anesthesia information management system, monitoring equipment, and anesthesia treatment records, and correlate the multi-source anesthesia data based on a unified time reference to form a unified event sequence; A structured rule package is constructed for the target anesthesia quality control events. The structured rule package includes the event observation object, time anchor point, triggering condition, duration condition, termination condition, exclusion condition, evidence extraction rules, and alarm rules. Based on time anchors, the target event judgment interval is determined from the unified event sequence, and the event judgment elements corresponding to the event observation object are extracted from the target event judgment interval; An event state machine corresponding to the target anesthesia quality control event is constructed based on a structured rule package; the event state machine includes a pending judgment state, a triggered judgment state, a continuous judgment state, a confirmed state, and a terminated state; The event determination elements are input into the event state machine. The event state machine is driven to perform state transitions according to the triggering condition, the duration condition, the termination condition, and the exclusion condition in order to determine whether the target anesthesia quality control event is established, and to record the corresponding state transition path and determination timestamp. When a target anesthesia quality control event is established, the corresponding quality control alarm is output according to the alarm rules, and the evidence fragments corresponding to the target anesthesia quality control event are extracted according to the evidence extraction rules to generate a quality control record for the target anesthesia quality control event.

[0007] Preferably, multi-source anesthesia data are correlated based on a unified time reference to form a unified event sequence, including: Extract the original time stamps corresponding to each of the multi-source anesthesia data; Convert the original time signature to a standard time signature under a unified time base; Multi-source anesthesia data are sorted and merged according to standard time identifiers, data source identifiers, and event category identifiers to form a unified event sequence.

[0008] Preferably, a structured rule package is constructed for the target anesthesia quality control events, including: Define the event observation objects corresponding to the target anesthesia quality control events; Define corresponding time anchors around the event observation object; Define the corresponding triggering conditions, duration conditions, termination conditions, and exclusion conditions around the time anchor point; Based on the event observation object and time anchor point, define the corresponding evidence extraction rules and alarm rules to form a structured rule package.

[0009] Preferably, determining the target event judgment interval from a unified event sequence based on time anchor points includes: Locate the anchor event corresponding to the time anchor in the unified event sequence; Based on the anchor event, the target event judgment interval is extracted according to the pre-observation range and post-observation range defined by the structured rule package; Multi-source anesthesia data falling within the target event determination interval are identified as the determination data for the target anesthesia quality control event.

[0010] Preferably, the event determination elements corresponding to the event observation object are extracted from the target event determination interval, including: Extract the monitoring parameter records, anesthesia record content, and anesthesia treatment records corresponding to the event observation object from the target event determination interval; The monitoring parameter records, anesthesia record content, and anesthesia treatment records are converted into event-based data to form corresponding judgment items. The event decision elements are generated by sequentially combining the decision items according to the structured rule package.

[0011] Preferably, an event state machine corresponding to the target anesthesia quality control event is constructed based on a structured rule package, including: The process of determining whether the triggering condition is not met is defined as a pending determination state; The initial determination process after the triggering condition is met is defined as the triggering determination state; The process of continuing to determine the condition that is met is defined as the continuous determination state; The determination result that meets the requirements for the establishment of the target anesthesia quality control event is defined as the confirmed status; The determination result after the termination condition or exclusion condition is met is defined as the termination state.

[0012] Preferably, the event state machine is driven to perform state transitions according to triggering conditions, duration conditions, termination conditions, and exclusion conditions, including: In the pending judgment state, when the event judgment element meets the triggering condition, the event state machine is transitioned to the triggered judgment state; In the triggered judgment state, if the event judgment elements continue to meet the continuous conditions, the event state machine will be transitioned to the continuous judgment state or the confirmation state. In the triggered judgment state or the continuous judgment state, when the event judgment element meets the termination condition or the exclusion condition, the event state machine will be transitioned to the termination state. Record the preceding state, following state, and decision timestamp for each state transition to form a state transition path.

[0013] Preferably, when a target anesthesia quality control event is established, a corresponding quality control alarm is output according to the alarm rules, including: Read the alarm level determination conditions corresponding to the target anesthesia quality control event from the alarm rules; The alarm level of the quality control alarm is determined based on the judgment result corresponding to the confirmed status; The alarm level, the target anesthesia quality control event, and the judgment timestamp are associated and output as a quality control alarm.

[0014] Preferably, evidence fragments corresponding to the target anesthesia quality control event are extracted according to evidence extraction rules to generate a quality control record for the target anesthesia quality control event, including: Based on the evidence extraction rules, key record segments corresponding to the observed objects of the event are extracted from the target event determination interval to form evidence fragments; Associate evidence fragments with target anesthesia quality control events, time anchors, state transition paths, judgment timestamps, and quality control alarms; The output includes quality control records containing evidence fragments, target anesthesia quality control events, time anchors, state transition paths, judgment timestamps, and quality control alarms.

[0015] An anesthesia quality control event recognition system based on rule packages and state machines includes: The multi-source anesthesia data association module is used to acquire multi-source anesthesia data from the anesthesia information management system, monitoring equipment, and anesthesia treatment records, and associate the multi-source anesthesia data based on a unified time reference to form a unified event sequence. The structured rule package construction module is used to build structured rule packages for target anesthesia quality control events. The structured rule package includes event observation objects, time anchors, triggering conditions, duration conditions, termination conditions, exclusion conditions, evidence extraction rules, and alarm rules. The event determination element extraction module is used to determine the target event determination interval from a unified event sequence based on time anchor points, and extract the event determination elements corresponding to the event observation object from the target event determination interval; The event state machine construction module is used to construct an event state machine corresponding to the target anesthesia quality control event based on the structured rule package; the event state machine includes a pending judgment state, a triggered judgment state, a continuous judgment state, a confirmed state, and a termination state; The state transition determination module is used to input event determination elements into the event state machine, drive the event state machine to perform state transitions according to trigger conditions, duration conditions, termination conditions and exclusion conditions, so as to determine whether the target anesthesia quality control event is established, and record the corresponding state transition path and determination timestamp. The quality control record generation module is used to output the corresponding quality control alarm according to the alarm rules when the target anesthesia quality control event is established, and to extract the evidence fragments corresponding to the target anesthesia quality control event according to the evidence extraction rules, and generate the quality control record of the target anesthesia quality control event.

[0016] The present invention discloses the following beneficial effects: (1) This invention links multi-source anesthesia data from the anesthesia information management system, monitoring equipment, and anesthesia treatment records according to a unified time reference, forming a unified event sequence. This enables data originally scattered from different record sources to be included in the same time correspondence for judgment. For anesthesia quality control events that require joint identification by multiple types of records, this technical solution can reduce judgment bias caused by inconsistent recording times.

[0017] (2) This invention constructs a structured rule package for target anesthesia quality control events, and clarifies the event observation object, time anchor point, triggering condition, duration condition, termination condition, exclusion condition, evidence extraction rules, and alarm rules in the structured rule package, which can fix the judgment criteria for a single anesthesia quality control event. After adopting this technical solution, the judgment criteria for the same type of anesthesia quality control event in different cases are more easily unified.

[0018] (3) This invention determines the target event judgment interval from a unified event sequence based on time anchors, and extracts event judgment elements corresponding to the event observation object from the target event judgment interval, thus limiting event judgment to the time range related to the target anesthesia quality control event. After adopting this technical solution, the influence of irrelevant time period recordings on event judgment is reduced accordingly.

[0019] (4) This invention constructs an event state machine corresponding to the target anesthesia quality control event based on a structured rule package, and drives the event state machine to perform state transitions according to triggering conditions, duration conditions, termination conditions, and exclusion conditions, which can incorporate the occurrence process, duration process, and termination process of the target anesthesia quality control event into the same judgment process. For anesthesia quality control events that need to be identified in conjunction with process changes, this technical solution is beneficial to improving the consistency between the start time and the end time of the event.

[0020] (5) When a target anesthesia quality control event is established, this invention outputs a quality control alarm according to the alarm rules and extracts evidence fragments corresponding to the target anesthesia quality control event according to the evidence extraction rules, generates a quality control record for the target anesthesia quality control event, and records the state transition path and judgment timestamp at the same time, so as to retain the corresponding judgment basis while retaining the event conclusion. During subsequent review, the basis for the establishment of the event can be directly verified based on the quality control record. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the 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.

[0022] Figure 1A flowchart of a method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of state transitions provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the system structure provided in an embodiment of the present invention.

[0023] Explanation of reference numerals in the attached figures: 1. Event State Machine; 2. State to be determined; 3. State to be determined; 4. State to be determined; 5. Confirmed state; 6. Terminated state; 7. First state transition path; 8. Second state transition path; 9. Third state transition path; 10. Fourth state transition path; 11. Fifth state transition path; 12. Main state transition path. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] The purpose of this invention is to provide a method and system for identifying anesthesia quality control events based on rule packages and state machines. By using rule packages and state machines to identify anesthesia quality control events in a process-oriented manner, the standardization of anesthesia quality control event judgment can be enhanced, and it is convenient to form quality control records with clear time basis.

[0026] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] Figure 1 The method flowchart provided in the embodiments of the present invention is as follows: Figure 1 As shown, this invention provides a method for identifying anesthesia quality control events based on rule packages and state machines, including: Step 100: Obtain multi-source anesthesia data from the anesthesia information management system, monitoring equipment, and anesthesia treatment records, and correlate the multi-source anesthesia data based on a unified time reference to form a unified event sequence; Step 200: Construct a structured rule package for the target anesthesia quality control event; the structured rule package includes the event observation object, time anchor point, triggering condition, duration condition, termination condition, exclusion condition, evidence extraction rules, and alarm rules; Step 300: Determine the target event judgment interval from the unified event sequence based on the time anchor point, and extract the event judgment elements corresponding to the event observation object from the target event judgment interval; Step 400: Construct an event state machine corresponding to the target anesthesia quality control event based on the structured rule package; the event state machine includes a pending judgment state, a triggered judgment state, a continuous judgment state, a confirmed state, and a terminated state; Step 500: Input the event judgment elements into the event state machine, drive the event state machine to perform state transitions according to the triggering condition, duration condition, termination condition and exclusion condition, so as to determine whether the target anesthesia quality control event is established, and record the corresponding state transition path and judgment timestamp. Step 600: When the target anesthesia quality control event is established, output the corresponding quality control alarm according to the alarm rules, and extract the evidence fragments corresponding to the target anesthesia quality control event according to the evidence extraction rules to generate the quality control record of the target anesthesia quality control event.

[0028] In this embodiment, a unique anesthesia process identifier is first determined based on the same patient, the same surgery, and the same anesthesia procedure, and this unique identifier is used to limit the data collection scope. As shown in Table 1, the multi-source anesthesia data includes anesthesia records from the anesthesia information management system, monitoring parameter records output by monitoring equipment, and anesthesia treatment records. Each record has a corresponding original time identifier, which refers to the occurrence time or formation time recorded in the corresponding source. In this embodiment, Beijing time (24-hour format, second-level) is used as the unified time reference, and the original time identifiers of each record are converted to standard time identifiers under the unified time reference. For cases where the monitoring equipment time is inconsistent with the unified time reference, this embodiment corrects the time according to the most recent time synchronization result before the start of anesthesia. Records with a deviation of more than 3 seconds after correction are not included in subsequent association processing. After the above processing, records from different sources achieve consistent time positioning at the same time scale.

[0029] Table 1 Examples of multi-source anesthesia data packages After completing the standard time signature conversion, this embodiment further assigns a data source identifier and an event category identifier to each record. The data source identifier is used to distinguish between three sources: the anesthesia information management system, monitoring equipment, and anesthesia treatment records. The event category identifier is used to distinguish between record categories such as monitoring parameter updates, anesthesia operations, treatment execution, and anesthesia termination. Subsequently, under the premise that the unique anesthesia process identifier is consistent, this embodiment first sorts all records in ascending order according to the standard time signature, and then merges them in order according to the data source identifier. For multiple records with the same standard time signature and the same data source identifier, the order in which the records were written is determined to determine the order of records within the same second. The records are arranged in sequence according to the above sorting results to form a unified event sequence. The unified event sequence refers to a set of continuous events based on a unified time benchmark that can reflect the chronological relationship of the current anesthesia process records. For duplicate records that are missing a unique anesthesia process identifier, missing a standard time signature, missing record content, or have the same record source, standard time signature, event category identifier, and record content, this embodiment will discard them and not include them in the unified event sequence.

[0030] Further, in this embodiment, step 200 establishes a structured rule package in a single-event manner. First, the event number and event name of the target anesthesia quality control event are determined. Then, the event observation object, time anchor point, triggering condition, duration condition, termination condition, exclusion condition, evidence extraction rules, and alarm rules are written around the target anesthesia quality control event to form a corresponding structured rule package. The structured rule package is used to fix the judgment criteria for a single target anesthesia quality control event. The event observation object is used to limit the recorded content read during judgment, specifically including monitoring parameter records, anesthesia operation records, and anesthesia treatment records. The time anchor point is used to limit the judgment starting point of the target anesthesia quality control event. In this embodiment, the time anchor point is recorded as one of the following: anesthesia start, airway establishment, skin incision, extubation, and anesthesia end. Each target anesthesia quality control event corresponds to one main time anchor point; when the target anesthesia quality control event involves judgment of changes before and after treatment, an auxiliary time anchor point is written accordingly. After the above writing, the observation object and judgment starting point of a single target anesthesia quality control event are determined.

[0031] After writing the event observation object and time anchor, continue writing the triggering condition, duration condition, termination condition, and exclusion condition around the time anchor. The triggering condition defines the initial requirements for the target anesthesia quality control event to enter the judgment process; it must include at least the record name, comparison relationship, and judgment value. The duration condition defines the requirements for the target anesthesia quality control event to continue to be valid; it is recorded as the duration or the number of consecutive hits. The termination condition defines the judgment boundary for the end of the target anesthesia quality control event. The exclusion condition removes records that are not included in the judgment scope of this target anesthesia quality control event. Subsequently, write the evidence extraction rules and alarm rules based on the event observation object and time anchor. The evidence extraction rules determine the records that should be retained after the target anesthesia quality control event is established, fixing the extraction of trigger point records, duration stage records, and termination point records. The alarm rules determine the output requirements after the target anesthesia quality control event is established, including at least the alarm level and output timing. After completing the above writing, a structured rule package corresponding to a single target anesthesia quality control event is formed, and subsequent determination of the target event judgment range and construction of the event state mechanism are carried out accordingly.

[0032] This example uses the following intraoperative hypotension event as the target anesthesia quality control event for illustration: First, determine the event number of the target anesthesia quality control event as AE-01, and the event name as intraoperative hypotension event. The event observation objects are recorded as mean arterial pressure (MAP) records and vasopressor administration records. The MAP records are taken from the MAP parameters output by the monitoring equipment, and the vasopressor administration records are taken from the vasopressor administration records in the anesthesia administration record. The time anchor point is recorded as the start of anesthesia. The trigger condition is recorded as MAP below 65 mmHg; the duration condition is recorded as the trigger condition lasting for 300 seconds, or five consecutive MAP records being below 65 mmHg; the termination condition is recorded as MAP recovering to 65 mmHg or above, and two consecutive MAP records meeting this condition; the exclusion condition is recorded as the existence of a monitoring interruption record within the corresponding time period, or the absence of a MAP record within that time period. After writing the above information, the observation objects, the determination starting point, and the establishment boundary of the intraoperative hypotension event are fixed.

[0033] After writing the above conditions, the evidence extraction rules and alarm rules for the target anesthesia quality control event are then written. The evidence extraction rules specify the extraction of the mean arterial pressure (MAP) record corresponding to the trigger time, the MAP record during the duration, the MAP recovery record corresponding to the termination time, and the vasopressor treatment record between the trigger and termination times. Each extracted record retains a standard time identifier, record item name, and record value. The alarm rules specify that events lasting 300 seconds but less than 600 seconds are classified as general alarms; events lasting 600 seconds or more are classified as critical alarms; and the corresponding alarm is output when the confirmed state is established. After writing the above content, a structured rule package corresponding to the intraoperative hypotension event is formed. When determining the subsequent target event judgment interval, the MAP record and vasopressor treatment record within the corresponding time range are read using the start of anesthesia as the time anchor point; when constructing the subsequent event state mechanism, the trigger conditions, duration conditions, termination conditions, and exclusion conditions recorded in this structured rule package are used as the basis for state transition.

[0034] More specifically, in this embodiment, step 300 first locates the anchor event corresponding to the time anchor point in the unified event sequence based on the time anchor point recorded in the structured rule package. Each record in the unified event sequence has recorded a standard time identifier, a data source identifier, and an event category identifier. To avoid the situation where the anchor point is not unique when the same time anchor point corresponds to multiple candidate records, this embodiment calculates the anchor point matching index for the i-th candidate record, and the anchor point matching index is denoted as . ,according to Determined; among them, Indicates the first The standard time identifier for each candidate record This indicates the reference time corresponding to the target time anchor point recorded in the structured rule package. This indicates the upper limit of the allowable deviation in anchor point time. This indicates a category consistency item; a value of 1 is assigned when the event category of the candidate record matches the time anchor category, and a value of 0 is assigned when they do not match. This indicates the source priority; a value of 1 is assigned if the source of the candidate record matches the priority source recorded in the structured rule package, and a value of 0 is assigned if they do not match. , , These represent the time matching weight, category matching weight, and source priority weight, respectively. In this embodiment, Take 60 seconds. Take 0.40, Take 0.35, Take 0.25. When Greater than hour, Take 0. Among the candidate records, The record with the largest value is identified as the anchor event; When the values ​​are the same, the candidate record with the earlier standard time identifier is selected as the anchor event. The standard time identifier corresponding to the anchor event is determined as the anchor time.

[0035] After determining the anchor time, this embodiment uses the anchor time as a reference to extract the target event determination interval. The target event determination interval is denoted as... ,according to Determined; among them, Indicates the anchor time. This indicates the scope of prior observations recorded in the structured rule package. This indicates the scope of subsequent observations recorded in the structured rule package. and All rules are pre-written into the structured rule package in seconds. The pre-observation range limits the time length before the anchor point time that is included in the judgment, and the post-observation range limits the time length after the anchor point time that is included in the judgment. For intraoperative hypotension events, this embodiment will... Recorded as 300 seconds, Recorded as 1800 seconds. In the unified event sequence, the standard time marker falls within the target event determination interval. The data is used to determine the target anesthesia quality control events; records with standard time markers located at the interval boundaries are also included in the determination data.

[0036] After the judgment data is determined, this embodiment extracts the monitoring parameter records, anesthesia record content, and anesthesia treatment records corresponding to the event observation object from the target event judgment interval. When the event observation object is recorded as a monitoring parameter object, the corresponding monitoring parameter records are extracted; when the event observation object is recorded as an anesthesia operation object, the corresponding anesthesia record content is extracted; when the event observation object is recorded as an anesthesia treatment object, the corresponding anesthesia treatment record is extracted. To ensure that the recording basis for subsequent event-based conversion is clear, this embodiment only retains data with complete standard time identifiers, complete record content, and not identified as duplicate records, and forms the retained records into a judgment data set. Each record in the judgment data set retains a standard time identifier, record item name, and record value. After the above processing, the valid records related to the event observation object within the target event judgment interval are limited to the same judgment range.

[0037] After completing the record extraction, this embodiment performs event-based transformation on the monitoring parameter records, anesthesia record content, and anesthesia treatment records. For the monitoring parameter records, the anomaly hit value of a single record is first calculated based on the comparison relationships and judgment values ​​recorded in the structured rule package. The anomaly hit value is denoted as... When the recorded value meets the triggering condition, If the record value is 1 and does not meet the trigger condition, Set the value to 0. Then, calculate the continuous correlation value for two adjacent monitoring parameter records; this continuous correlation value is denoted as... The time interval between two adjacent monitoring parameter records shall not exceed the preset merging interval. hour, Set to 1, otherwise set to 0. In this embodiment, Take 30 seconds. Only when two adjacent monitoring parameter records simultaneously meet the condition... "and Only when the time is right will the corresponding consecutive record segments be merged into the same parameter decision item. For anesthesia record content and anesthesia treatment record, each record is converted into one operation decision item or one treatment decision item, and consecutive merging is not performed. After the above conversion, the monitoring parameter record forms a parameter decision item, the anesthesia record content forms an operation decision item, and the anesthesia treatment record forms a treatment decision item. Each decision item records the start time, end time, corresponding event observation object, and decision result.

[0038] After forming each decision item, this embodiment combines the decision items sequentially according to a structured rule package to generate event decision elements. To ensure that the combination result simultaneously reflects the proximity between the decision item and the anchor point time, the sequential relationship between the decision items, and the connection between the parameter decision items and the handling decision items, this embodiment calculates an element combination index for the candidate decision item set. This element combination index is denoted as... ,according to Determined; among them, Indicates the first The time offset between the start time and the anchor time of each decision item This represents the total number of decision items in the candidate decision item set. Indicates the allowable range of anchor point offset. This represents the number of decision items that satisfy the preset order of the structured rule package. This represents the total number of decision items that participate in the order comparison. This represents the number of association criteria that fall within the associated time interval. This represents the total number of correlation criteria that require interval checks. , , These represent the anchor point proximity weight, the order consistency weight, and the handling association weight, respectively. In this embodiment, Take 900 seconds. Take 0.35, Take 0.30, Set the value to 0.35, and the associated time interval to 120 seconds. When When the value is not lower than 0.75, the corresponding set of candidate decision items is determined as the event decision element; when When the value is below 0.75, only the individual judgment results are retained, and no event judgment element is formed. After the above time sequence combination, the parameter judgment items, operation judgment items, and disposal judgment items that exist scattered within the target event judgment interval are organized into event judgment elements corresponding to the event observation object.

[0039] Further, in this embodiment, step 400 constructs an event state machine corresponding to the target anesthesia quality control event based on the structured rule package. First, the triggering condition, duration condition, termination condition, and exclusion condition corresponding to the target anesthesia quality control event are read, and then a state set is established around the target anesthesia quality control event. Specifically, the judgment process that does not meet the triggering condition is recorded as a pending judgment state, the initial judgment process after meeting the triggering condition is recorded as a triggered judgment state, the continued judgment process that meets the duration condition is recorded as a continuous judgment state, the judgment result that meets the requirements for the establishment of the target anesthesia quality control event is recorded as a confirmed state, and the judgment result after meeting the termination condition or exclusion condition is recorded as a terminated state. Each of the above states is written with a state code, an entry condition, and an exit condition, where the entry condition and exit condition are both taken from the corresponding rule items recorded in the structured rule package.

[0040] After establishing the state set, this embodiment further writes the connection relationships between the states. The state to be judged serves as the initial state, the triggered judgment state serves as the entry state after triggering, the continuous judgment state serves as the continuation state after continuously satisfying the conditions, the confirmation state serves as the result state after the event is established, and the termination state serves as the exit state after the event ends. For the same target anesthesia quality control event, only one event state machine is established. After completing the above writing, an event state machine corresponding to the target anesthesia quality control event is formed.

[0041] In this exemplary embodiment, the event state machine does not statically mark event results, but rather uses it to segment and define the judgment process of the target anesthesia quality control event. Specifically, event judgment elements that do not meet the triggering conditions are first placed in the pending judgment state; when an event judgment element first meets the triggering conditions, the corresponding judgment process is placed in the triggered judgment state; if the continuous conditions are met after triggering, the corresponding judgment process is placed in the continuous judgment state; when the event judgment element meets the requirements for the establishment of the target anesthesia quality control event, the corresponding judgment result is placed in the confirmed state; when the event judgment element meets the termination or exclusion conditions, the corresponding judgment result is placed in the terminated state. Through the above processing, the start, continuous, establishment, and end stages of the target anesthesia quality control event fall into different states, thus providing clear boundaries for subsequent state transition judgments.

[0042] Furthermore, in this embodiment, step 500 drives the event state machine to perform state transitions according to triggering conditions, duration conditions, termination conditions, and exclusion conditions. First, the event determination elements formed in step 300 are input into the event state machine in standard time identifier order, and then a trigger transition value is calculated for the currently input event determination elements. The trigger transition value is denoted as... ,according to Determined; among them, This indicates that the trigger condition is met. The value is 1 when the event judgment element meets the trigger condition, and 0 when the trigger condition is not met. This indicates that the exclusion condition is met. The value is 1 if the event element meets the exclusion condition, and 0 if the exclusion condition is not met. When in a pending judgment state... When P_tr is set to 1, the event state machine transitions from the pending judgment state to the triggered judgment state, and the start time corresponding to the event judgment element that triggered this transition is recorded as the first trigger time; when P_tr is set to 0, the event state machine remains in the pending judgment state. Through the above processing, there is a clear and unique transition threshold for whether the target anesthesia quality control event enters the judgment process.

[0043] After the event state machine enters the trigger determination state, this embodiment continues to determine the subsequent migration direction based on the persistence condition and the establishment requirements of the target anesthesia quality control event. For this purpose, the persistence cumulative amount is calculated and denoted as... ,according to Determined; among them, This indicates the termination time corresponding to the current event determination element. This indicates the initial trigger time. The structured rule package pre-defines the persistence threshold and the confirmation threshold. When the cumulative persistence reaches the persistence threshold but not the confirmation threshold, the event state machine transitions from the trigger state to the persistence state. When the cumulative persistence reaches the confirmation threshold, or when the requirements for establishing the target anesthesia quality control event are continuously met, the event state machine directly transitions from the trigger state to the confirmation state. If the event state machine is already in the persistence state, and the current event judgment element continues to meet the persistence condition but has not met the termination or exclusion condition, the persistence state is maintained; if the current event judgment element meets the requirements for establishing the target anesthesia quality control event, it transitions to the confirmation state. Through the above processing, the post-trigger persistence judgment process and the event establishment judgment process are handled separately.

[0044] When the event state machine is in the triggered decision state or the continuous decision state, this embodiment continues to check the termination condition and exclusion condition, and completes the exit transition accordingly. The exit transition value is denoted as... ,according to Determined; among them, The flag indicates that the termination condition is met. The flag is 1 when the event determination element meets the termination condition and 0 when the termination condition is not met. This still indicates that the exclusion condition is met. If the value is 1, the event state machine transitions to the terminated state; If the value is 0, the event state machine does not perform an exit transition. For each state transition, this embodiment records the corresponding preceding state, succeeding state, and judgment timestamp. The judgment timestamp is the start or end time of the event judgment element that triggered the transition. All transition records are arranged in chronological order to form a state transition path. Through the above processing, the entire process of the target anesthesia quality control event, from entry judgment, continuous judgment, establishment judgment to exit judgment, is included in the same state transition path.

[0045] Exemplary, such as Figure 2 As shown, the event state machine 1 includes a pending state 2, a triggered state 3, a continuous state 4, a confirmed state 5, and a terminated state 6. The pending state 2 transitions to the triggered state 3 via the first state transition path 7. The triggered state 3 transitions to the continuous state 4 via the second state transition path 8. The triggered state 3 transitions to the confirmed state 5 via the third state transition path 9. The continuous state 4 transitions to the confirmed state 5 via the fourth state transition path 10. The triggered state 3 and the continuous state 4 transition to the terminated state 6 via the fifth state transition path 11. The main state transition path 12 is used to illustrate the main transition process between the pending state 2, the triggered state 3, the continuous state 4, and the confirmed state 5. The first state transition path 7 corresponds to the transition process after the event judgment element meets the triggering condition. The second state transition path 8 corresponds to the transition process after the event judgment element continues to meet the continuous condition. The third state transition path 9 and the fourth state transition path 10 correspond to the transition process after the target anesthesia quality control event meets the establishment requirements. The fifth state transition path 11 corresponds to the exit transition process after the event judgment element meets the termination condition or the exclusion condition.

[0046] Furthermore, in this embodiment, step 600 first determines the corresponding quality control alarm level based on alarm rules when the target anesthesia quality control event is established. Specifically, the alarm level determination conditions corresponding to the target anesthesia quality control event are first read from the structured rule package, and then, when the event state machine transitions to the confirmation state, the element combination index formed in step 300 is called. and the continuous cumulative amount formed in step 500 Calculate the alarm determination value. The alarm determination value is denoted as... ,according to Determined; among them, This represents the factor combination index formed in step 300. This represents the cumulative amount formed in step 500. This indicates the baseline duration recorded in the structured rules package. This is used to define the duration corresponding to the baseline risk level of the target anesthesia quality control event. In this embodiment, The structured rule package is written in seconds; for intraoperative hypotension events, The recording time is 300 seconds. When When the value is less than 1, it is determined to be a general quality control alarm; when When the value is not less than 1 and less than 2, it is determined as a key quality control alarm; when When the value is not less than 2, it is determined to be an immediate quality control alarm. After the above processing, the judgment result corresponding to the confirmed status is converted into a quality control alarm with a clear level boundary.

[0047] After the alarm level is determined, this embodiment correlates and outputs the alarm level, the target anesthesia quality control event, and the judgment timestamp. The judgment timestamp is the standard time identifier corresponding to the transition of the event state machine to the confirmation state, and is recorded as follows. To ensure that alarm outputs for the same target anesthesia quality control event have a fixed structure, this embodiment writes the event number, alarm level, and judgment timestamp into the same alarm result. The event number is taken from the structured rule package, and the alarm level is taken from the aforementioned alarm judgment value. The corresponding grading results are determined by taking the timestamp from the standard time identifier corresponding to the confirmed status. After completing the above association, a quality control alarm is generated that corresponds one-to-one with the target anesthesia quality control event. Subsequent evidence extraction and quality control record generation are all based on this quality control alarm.

[0048] After completing the quality control alarm output, this embodiment extracts key record segments corresponding to the event observation object from the target event judgment interval according to the evidence extraction rules, forming evidence fragments. Specifically, the initial trigger time formed in step 500 is used first... The timestamp corresponding to the confirmation status Using the boundary as a reference, and combining it with the target event determination interval formed in step 300. The evidence coverage area is determined. This evidence coverage area is denoted as... ,according to Determined; among them, Indicates the forward fetching time. This indicates the backward fetching time; both are pre-written by the structured rules package. Used to retain necessary prior records before the first trigger time. This is used to retain necessary follow-up records after the confirmed status is established. In this embodiment, Take 60 seconds. Take 120 seconds. The standard time marker falls within the evidence coverage area. Records of monitoring parameters, anesthesia, and anesthesia procedures that are consistent with the events observed were identified as candidate evidence records. These candidate evidence records were then arranged in standard time-stamped order and categorized by trigger point record, duration record, and confirmation point record to form evidence fragments.

[0049] After forming the evidence fragment, this embodiment associates the evidence fragment with the target anesthesia quality control event, time anchor point, state transition path, judgment timestamp, and quality control alarm to generate a quality control record for the target anesthesia quality control event. To ensure the completeness of the output results, this embodiment calculates the record completeness of the quality control record. The record completeness is denoted as... ,according to Determined; among them, This indicates the number of necessary elements that have been written into the quality control record. This indicates the total number of pre-defined essential elements in the structured rule package. The essential elements are fixed and include six items: evidence fragments, target anesthesia quality control events, time anchors, state transition paths, decision timestamps, and quality control alarms. Fixed at 6. When it equals 6, If the value is 1, the output will be a quality control record containing evidence fragments, target anesthesia quality control events, time anchors, state transition paths, judgment timestamps, and quality control alarms. When it is less than 6, If the result is less than 1, it will not be output as a complete quality control record. After the above processing, the judgment results, judgment basis, and migration process corresponding to the target anesthesia quality control event are uniformly solidified into the same quality control record, as shown in Table 2.

[0050] Table 2. Examples of quality control records corresponding to target anesthesia quality control events. Table 2 shows examples of quality control records corresponding to the target anesthesia quality control events in this embodiment. The target anesthesia quality control events selected in Table 2 are intraoperative hypotension, intraoperative hypoxia, and hypothermia. The evidence fragments for intraoperative hypotension revolve around changes in mean arterial pressure and records of vasopressor administration. Vasopressor administration records use norepinephrine administration records to reflect more common circulatory support situations during anesthesia. The evidence fragments for intraoperative hypoxia revolve around changes in blood oxygen saturation and records of oxygen concentration adjustments to demonstrate the identification process of abnormal oxygenation after airway establishment. The evidence fragments for hypothermia revolve around changes in body temperature monitoring values ​​and records of warming measures such as warming blankets and intravenous fluid warming to demonstrate the judgment process for temperature management-related quality control events. Through the above setup, each example record in Table 2 is organized around the event type, key monitoring indicators, and corresponding treatment records of anesthesia quality control concern, thereby making the correspondence between target anesthesia quality control events, state transition paths, quality control alarms, and evidence fragments clearer.

[0051] Corresponding to the above methods, such as Figure 3 As shown, this embodiment also provides an anesthesia quality control event recognition system based on rule packages and state machines, including: The multi-source anesthesia data association module is used to acquire multi-source anesthesia data from the anesthesia information management system, monitoring equipment, and anesthesia treatment records, and associate the multi-source anesthesia data based on a unified time reference to form a unified event sequence. The structured rule package construction module is used to build structured rule packages for target anesthesia quality control events. The structured rule package includes event observation objects, time anchors, triggering conditions, duration conditions, termination conditions, exclusion conditions, evidence extraction rules, and alarm rules. The event determination element extraction module is used to determine the target event determination interval from a unified event sequence based on time anchor points, and extract the event determination elements corresponding to the event observation object from the target event determination interval; The event state machine construction module is used to construct an event state machine corresponding to the target anesthesia quality control event based on the structured rule package; the event state machine includes a pending judgment state, a triggered judgment state, a continuous judgment state, a confirmed state, and a termination state; The state transition determination module is used to input event determination elements into the event state machine, drive the event state machine to perform state transitions according to trigger conditions, duration conditions, termination conditions and exclusion conditions, so as to determine whether the target anesthesia quality control event is established, and record the corresponding state transition path and determination timestamp. The quality control record generation module is used to output the corresponding quality control alarm according to the alarm rules when the target anesthesia quality control event is established, and to extract the evidence fragments corresponding to the target anesthesia quality control event according to the evidence extraction rules, and generate the quality control record of the target anesthesia quality control event.

[0052] The beneficial effects of this invention are as follows: (1) This invention first associates multi-source anesthesia data from the anesthesia information management system, monitoring equipment, and anesthesia treatment records according to a unified time benchmark to form a unified event sequence. Then, it establishes a structured rule package around the target anesthesia quality control event, thereby incorporating monitoring parameters, anesthesia operations, and treatment records that were originally scattered across different record sources into the same judgment framework. For anesthesia quality control events that require simultaneous identification based on parameter changes, the time of operation, and the treatment process, the above technical solution can reduce the judgment bias caused by inconsistencies in time correspondence between different record sources, and ensure that the observation objects, judgment starting points, judgment boundaries, and exclusion scenarios for the same type of event all have fixed bases. As a result, the target anesthesia quality control event no longer relies on manual item-by-item comparison of records for judgment, and the event judgment criteria are more likely to remain consistent.

[0053] (2) Based on a unified event sequence, this invention determines the target event judgment interval according to time anchor points, extracts event judgment elements from the target event judgment interval, and then uses an event state machine to segment and judge the pending judgment state, triggered judgment state, continuous judgment state, confirmed state, and termination state. The above technical solution processes whether the event has started, whether it continues, whether it is established, and when it ends by placing them into corresponding states, which is applicable to anesthesia quality control events that require joint identification of the occurrence process and the continuous process. Compared with the processing method that makes judgments based on only a single abnormal record, this invention can distinguish the event start time, continuous stage, and termination boundary, which helps to reduce the interference of instantaneous abnormalities, isolated records, or non-continuous records on the judgment of event establishment, thereby improving the stability of the identification results of complex quality control events.

[0054] (3) When a target anesthesia quality control event is established, this invention outputs a corresponding quality control alarm according to alarm rules and extracts evidence fragments according to evidence extraction rules, ultimately forming a quality control record that includes the target anesthesia quality control event, time anchor point, state transition path, judgment timestamp, quality control alarm, and evidence fragments. After this processing, the output result no longer only stays at the single conclusion of whether the event is established, but also retains the key record segment and state transition process corresponding to the formation of the conclusion. During subsequent review, reviewers can directly trace back the triggering basis, continuity basis, and confirmation basis of the event based on the quality control record, reducing the workload of repeatedly retrieving the original record and reorganizing the temporal relationship, and also helping to improve the integrity of anesthesia quality control record keeping.

[0055] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0056] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A rule package and state machine based narcotic quality control event identification method, characterized by, include: Acquire multi-source anesthesia data from the anesthesia information management system, monitoring equipment, and anesthesia treatment records, and correlate the multi-source anesthesia data based on a unified time reference to form a unified event sequence; Develop a structured rule package for target anesthesia quality control events; The structured rule package includes event observation objects, time anchors, triggering conditions, duration conditions, termination conditions, exclusion conditions, evidence extraction rules, and alarm rules; Based on the time anchor point, a target event determination interval is determined from the unified event sequence, and event determination elements corresponding to the event observation object are extracted from the target event determination interval; An event state machine corresponding to the target anesthesia quality control event is constructed based on the structured rule package; the event state machine includes a pending judgment state, a triggered judgment state, a continuous judgment state, a confirmed state, and a termination state; The event determination elements are input into the event state machine, and the event state machine is driven to perform state transitions according to the triggering conditions, duration conditions, termination conditions and exclusion conditions, so as to determine whether the target anesthesia quality control event is established, and record the corresponding state transition path and determination timestamp. When the target anesthesia quality control event is established, the corresponding quality control alarm is output according to the alarm rule, and the evidence fragment corresponding to the target anesthesia quality control event is extracted according to the evidence extraction rule to generate the quality control record of the target anesthesia quality control event.

2. The rule pack and state machine based anesthetic quality control event identification method of claim 1, wherein, The multi-source anesthesia data are correlated based on a unified time reference to form a unified event sequence, including: Extract the original time stamps corresponding to each of the multi-source anesthesia data; Convert the original time identifier into a standard time identifier under a unified time base; The multi-source anesthesia data are sorted and merged according to the standard time identifier, data source identifier, and event category identifier to form the unified event sequence.

3. The rule pack and state machine based anesthetic quality control event recognition method of claim 1, wherein, A structured rule package is constructed for the target anesthesia quality control events, including: Define the event observation object corresponding to the target anesthesia quality control event; Define corresponding time anchor points around the event observation object; Define corresponding triggering conditions, duration conditions, termination conditions, and exclusion conditions around the aforementioned time anchor point; Based on the event observation object and the time anchor point, the corresponding evidence extraction rules and alarm rules are defined to form the structured rule package.

4. The rule pack and state machine based anesthetic quality control event recognition method of claim 1, wherein, Determining the target event judgment interval from the unified event sequence based on the time anchor point includes: Locate the anchor event corresponding to the time anchor point in the unified event sequence; Based on the anchor event, the target event judgment interval is extracted according to the pre-observation range and post-observation range defined by the structured rule package; The multi-source anesthesia data that falls within the target event determination range are determined as the determination data for the target anesthesia quality control event.

5. The rule pack and state machine based anesthetic quality control event recognition method of claim 1, wherein, Extracting event determination elements corresponding to the event observation object from the target event determination interval includes: Extract the monitoring parameter records, anesthesia record content, and anesthesia treatment records corresponding to the event observation object from the target event determination interval; The monitoring parameter records, the anesthesia record content, and the anesthesia treatment records are converted into event-based data to form corresponding judgment items. The event determination elements are generated by sequentially combining each of the determination items according to the structured rule package.

6. The anesthesia quality control event identification method based on rule packages and state machines according to claim 1, characterized in that, Based on the structured rule package, an event state machine corresponding to the target anesthesia quality control event is constructed, including: The process of determining whether the triggering condition is not met is defined as a pending determination state; The initial determination process after the triggering condition is met is defined as the triggering determination state; The process of continuing to determine the condition that is met is defined as the continuous determination state; The determination result that meets the requirements for the establishment of the target anesthesia quality control event is defined as the confirmation status; The determination result after satisfying the termination condition or the exclusion condition is defined as the termination state.

7. The anesthesia quality control event identification method based on rule packages and state machines according to claim 1, characterized in that, The event state machine is driven to perform state transitions according to the aforementioned triggering conditions, duration conditions, termination conditions, and exclusion conditions, including: In the pending determination state, when the event determination element meets the triggering condition, the event state machine is transitioned to the trigger determination state; In the triggered determination state, if the event determination element continues to meet the continuous condition, the event state machine is transitioned to the continuous determination state or the confirmation state. In the triggered determination state or the continuous determination state, when the event determination element satisfies the termination condition or the exclusion condition, the event state machine is transitioned to the termination state. Record the preceding state, following state, and decision timestamp corresponding to each state transition to form the state transition path.

8. The anesthesia quality control event identification method based on rule packages and state machines according to claim 1, characterized in that, When the target anesthesia quality control event is established, a corresponding quality control alarm is output according to the alarm rules, including: Read the alarm level determination conditions corresponding to the target anesthesia quality control event from the alarm rules; The alarm level of the quality control alarm is determined based on the judgment result corresponding to the confirmation status; The alarm level, the target anesthesia quality control event, and the judgment timestamp are associated and output as the quality control alarm.

9. The anesthesia quality control event identification method based on rule packages and state machines according to claim 1, characterized in that, Evidence fragments corresponding to the target anesthesia quality control event are extracted according to the evidence extraction rules, and a quality control record for the target anesthesia quality control event is generated, including: Based on the evidence extraction rules, key record segments corresponding to the event observation object are extracted from the target event determination interval to form the evidence fragment; The evidence fragments are associated with the target anesthesia quality control event, the time anchor, the state transition path, the judgment timestamp, and the quality control alarm; The output includes the evidence fragment, the target anesthesia quality control event, the time anchor, the state transition path, the judgment timestamp, and the quality control alarm in the quality control record.

10. An anesthesia quality control event recognition system based on rule packages and state machines, characterized in that, include: The multi-source anesthesia data association module is used to acquire multi-source anesthesia data from the anesthesia information management system, monitoring equipment, and anesthesia treatment records, and associate the multi-source anesthesia data based on a unified time reference to form a unified event sequence. The structured rule package building module is used to build structured rule packages for target anesthesia quality control events; The structured rule package includes event observation objects, time anchors, triggering conditions, duration conditions, termination conditions, exclusion conditions, evidence extraction rules, and alarm rules; The event determination element extraction module is used to determine the target event determination interval from the unified event sequence based on the time anchor point, and extract the event determination elements corresponding to the event observation object from the target event determination interval; An event state machine construction module is used to construct an event state machine corresponding to the target anesthesia quality control event based on the structured rule package; the event state machine includes a pending judgment state, a triggered judgment state, a continuous judgment state, a confirmed state, and a termination state; The state transition determination module is used to input the event determination elements into the event state machine, drive the event state machine to perform state transition according to the triggering condition, the duration condition, the termination condition and the exclusion condition, so as to determine whether the target anesthesia quality control event is established, and record the corresponding state transition path and determination timestamp. The quality control record generation module is used to output a corresponding quality control alarm according to the alarm rules when the target anesthesia quality control event is established, and to extract evidence fragments corresponding to the target anesthesia quality control event according to the evidence extraction rules, and generate a quality control record for the target anesthesia quality control event.

Images