Method for dynamically evaluating awakening progress of patient in anesthesia recovery stage of operating room

Abstract

The invention relates to the technical field of vital sign monitoring and evaluation, in particular to an operating room anesthesia recovery stage patient wake-up progress dynamic evaluation method, which comprises the following steps of: obtaining anesthesia recovery stage intervention behaviors, performing time sequence arrangement, performing segmentation and statistics according to time dimensions, analyzing intervention density change, adjusting an evaluation time range, and obtaining an evaluation result. And performing association analysis in combination with the vital sign signals in the corresponding time periods to form a patient waking progress dynamic evaluation conclusion. According to the method, quantitative expression and change trend capture of intervention density are achieved by constructing a structured intervention operation sequence and performing time granularity segmentation statistics, the evaluation boundary is adjusted in combination with the density difference value direction, the focusing capacity for the state active period is improved, and on the basis, continuous electroencephalogram and blood oxygen signal sequences are extracted and subjected to pairing analysis; collaborative fluctuation characteristics of nerve and blood oxygen response are recognized, dynamic quantitative judgment on the awakening process is formed, and the precision and timeliness of state recognition are enhanced.

Description

Technical Field

[0001] This invention relates to the field of vital sign monitoring and assessment technology, and in particular to a method for dynamically assessing the recovery progress of patients during the anesthesia recovery phase in the operating room. Background Technology

[0002] The field of vital sign monitoring and assessment technology involves the continuous or intermittent acquisition, analysis, and assessment of key human physiological parameters such as heart rate, blood pressure, respiratory rate, blood oxygen saturation, and electroencephalogram (EEG) activity through sensors, monitoring equipment, and physiological signal acquisition methods. This data is used to assist in the diagnosis, treatment feedback, and rehabilitation management during the medical process. This field covers several core aspects, including physiological signal acquisition methods, physiological parameter assessment models, dynamic monitoring systems, human-computer interaction assessment interfaces, and disease trend recognition methods. It plays a crucial role in clinical applications such as surgery, intensive care, pre-hospital emergency care, and rehabilitation assessment. In particular, it is critical for the accurate identification and assessment of changes in patient condition during and after surgery, especially in the operating room and anesthesia management. The traditional dynamic assessment method for patient awakening progress during the recovery phase of anesthesia in the operating room refers to observing a series of physiological behaviors such as the recovery of spontaneous breathing, eye opening response, limb movement, and speech ability after the patient has undergone surgical anesthesia and entered the recovery phase. It is combined with static parameters such as electrocardiogram, blood oxygen, and blood pressure to make a phased judgment. Generally, it is done by medical staff to observe and record, supplemented by timed collection of single-point physiological parameters. It relies on experience for status assessment and lacks real-time quantitative means for continuous dynamic changes in the awakening process. The assessment granularity is relatively coarse, highly subjective, and the response lags behind the actual changes in the patient's status.

[0003] Existing technologies use phased physiological parameters and behavioral observations to assess patient awakening. However, the operational mode suffers from a lack of real-time correspondence between monitoring response and intervention behavior, failing to present the direct impact of intervention events on changes in patient status. This results in a lack of intervention context support for status identification. Furthermore, single-point collected physiological parameters lack the ability to describe fluctuation trends and do not have the value of tracking the evolution of abnormal states. The assessment results are difficult to reflect the continuous characteristics of status changes. Limited by the intermittent nature of observation records and the unstructured expression of assessment criteria, status judgment is subject to subjective interference, which can easily lead to feedback lag and make it difficult to meet the need for identification of high-frequency dynamic changes during the postoperative awakening stage. Summary of the Invention

[0004] To achieve the above objectives, the present invention employs the following technical solution: a method for dynamically assessing the patient's awakening progress during the anesthesia recovery phase in the operating room, comprising the following steps: S1: Obtain the intervention behaviors recorded during the anesthesia recovery phase, mark the trigger time, operation type and execution order, construct the intervention operation chain according to the time sequence, and generate the operation behavior sequence by segmenting according to the preset time interval granularity; S2: Call the operation behavior sequence, summarize the occurrence frequency of time segment intervention types, call the preset intervention level parameter set, perform hierarchical statistics on intervention types, perform multiplication operation according to preset level coefficients and sum them, and superimpose the number of intervention behaviors and intensity levels in a weighted manner. Combine the continuity of intervention operation time to construct a time scale mapping and generate an intervention frequency distribution map. S3: Call the intervention frequency distribution map, collect the intervention density change results within the sliding window, calculate the direction of the intervention density difference between adjacent windows, determine the trend change, extract the time nodes with density change tendency, and output the trend change marker. S4: Call the trend change marker, compare the density growth direction with the current evaluation time period, control the time boundary to shrink and extend towards the density concentration feature direction, and construct the time evaluation boundary segment; S5: Call the time assessment boundary segment, extract vital sign monitoring data, obtain EEG rhythm signal sequence and blood oxygen saturation signal sequence, analyze the period and fluctuation direction, identify the coordinated fluctuation phenomenon, and output the dynamic assessment conclusion of the patient's awakening progress.

[0005] As a further aspect of the present invention, the operational behavior sequence includes intervention trigger time markers, operation type codes, execution sequence numbers, and time granularity segments; the intervention frequency distribution map includes intervention type level stratification, weighted intervention intensity values, and time continuity mapping relationships; the trend change markers include intervention density difference direction, trend change continuity indicators, and change tendency time nodes; the time assessment boundary segments include boundary contraction segments, boundary extension segments, and adjusted boundary combinations; and the dynamic assessment conclusions of the patient's awakening progress include EEG rhythm change characteristics, blood oxygen saturation fluctuation characteristics, and neuro-blood oxygen synergistic fluctuation relationships.

[0006] As a further aspect of the present invention, the time nodes with density change trends refer to key time points in the intervention frequency distribution map where the intervention density changes with a trend of increase or decrease, calculated by a sliding window.

[0007] As a further aspect of the present invention, the identification of coordinated fluctuation phenomenon refers to the phenomenon in which the periodic changes in EEG rhythm and blood oxygen saturation signals maintain a consistent and synchronous fluctuation direction.

[0008] As a further aspect of the present invention, the specific steps of S1 are as follows: S101: Obtain the intervention behaviors during the anesthesia recovery stage recorded in the nurse's operating terminal, extract the timestamp field from each intervention behavior record, use the timestamp field as the trigger time of the intervention behavior, call the operation field in the corresponding intervention record, obtain the operation instruction content, mark the operation sequence of the intervention behavior by traversing the sequence number field, and generate an intervention behavior parameter set; S102: Based on the trigger time, operation type and operation sequence extracted from the intervention behavior parameter set, the operation sequence is used as the sorting basis to sort all intervention behaviors in ascending order. An operation linked list structure is constructed according to the sorting result. Adjacent intervention behavior nodes are connected through the linked list structure to establish intervention operation linked list structure data. S103: Call the timestamp information of each node in the intervention operation chain list structure data, divide the nodes in the operation chain list into time periods according to the preset time interval granularity, and generate an operation behavior sequence according to the original order of the nodes included in each time period.

[0009] As a further aspect of the present invention, the specific steps of S2 are as follows: S201: Call the time segment data in the operation behavior sequence, traverse the intervention instruction content in each time segment, identify the intervention type identifier corresponding to the differentiated behavior instruction, and generate an intervention type frequency matrix by counting the number of intervention types in the same time segment and counting them independently according to the intervention type. S202: Based on the frequency matrix of intervention types, call the preset set of intervention level parameters, perform level matching on intervention types within the same time segment, extract the level coefficient corresponding to each intervention type, and perform a weighted superposition method to multiply the number of intervention types and the level coefficient within the time segment and then sum them to obtain the numerical distribution table of intervention intensity. S203: Based on the time segment sequence in the intervention intensity numerical distribution table, establish a time scale axis of equal length according to the start and end positions of each time segment, map the intervention intensity values ​​onto the time scale axis according to the time segment positions, connect adjacent data points in chronological order to form a continuous numerical curve, and generate an intervention frequency distribution map.

[0010] As a further aspect of the present invention, the specific steps of S3 are as follows: S301: Call the intervention intensity values ​​on the continuous time axis of the intervention frequency distribution map, set the equal-width sliding window parameters, and sequentially extract the set of intervention intensity values ​​in the corresponding time period by sliding the window. Calculate the arithmetic mean of all intervention intensity values ​​in each sliding window to generate a sliding window intervention density sequence. S302: Based on the density values ​​of two adjacent sliding window positions in the sliding window intervention density sequence, perform a difference operation between the current position density value and the previous window density value, extract the positive or negative sign of the difference to determine the direction of density change, and obtain the density change trend direction sequence by continuously judging whether the direction of the difference before and after is reversed. S303: Based on the sign inflection point position of the density direction change in the density change trend direction sequence, retrieve all time index nodes where the direction change occurs within the continuous detection period, extract the corresponding time position on the time scale axis, and output the trend change mark.

[0011] As a further aspect of the present invention, the specific steps of S4 are as follows: S401: Call the time identifier node in the trend change marker, extract the continuously arranged density growth direction marker, set the start and end boundary values ​​of the current evaluation time period, and index and compare the continuous density growth markers with the time axis scale, filter all growth marker nodes within the current time period range, and obtain the density concentration trend node set; S402: Based on the density concentration trend node set, analyze the order of time nodes and the boundary positions of the current evaluation time period, and determine whether the direction of change trend tends to be closer to the center position of the time period. If the judgment is true, adjust the start and end boundaries to decrease towards the center position in turn to obtain the boundary contraction interval parameter set. S403: Based on the boundary adjustment results of the boundary contraction interval parameter set, the current evaluation time period is replaced according to the adjusted start and end boundaries to construct a continuous change segment structure, and a combination aggregation operation is performed on all change segments to generate time evaluation boundary segments.

[0012] As a further aspect of the present invention, the specific steps of S5 are as follows: S501: Call the start and end boundaries in the time assessment boundary segment, extract vital sign monitoring data frames according to time period, and extract the original signals of the EEG channel and the blood oxygen monitoring channel at the corresponding sampling frequency within the extraction range. Separate the data according to the channel type to obtain the EEG and blood oxygen combined signal sequence set. S502: Based on the combined EEG and blood oxygenation signal sequence set, the EEG rhythm signal sequence and blood oxygenation signal sequence are paired point by point using a synchronous indexing method under a unified time axis. Local cycles and fluctuation trends are extracted respectively. The phase change direction and cycle duration in adjacent intervals are compared to generate a neural blood oxygenation fluctuation comparison result matrix. S503: Based on the phased periodic direction consistency characteristics extracted from the neural blood oxygen fluctuation comparison result matrix, count the number of time segments with synchronous upward and synchronous downward trends within the time period, and mark them as response event node sequences. Connect and merge adjacent response event nodes to output the dynamic assessment conclusion of the patient's awakening progress.

[0013] As a further aspect of the present invention, the response event node sequence refers to the set of consecutive time points corresponding to the same trend change of EEG rhythm and blood oxygen saturation signal on a unified time axis, which are consistent in phase direction and synchronized in period, marking the temporal nodes of the neuro-blood oxygen coordinated response during the patient's awakening process.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by constructing a structured intervention operation sequence and performing time-granular segmented statistics, the quantitative expression of intervention density and the capture of change trends are realized. Combined with the adjustment of the assessment boundary by density difference direction, the ability to focus on the active period of the state is improved. On this basis, continuous EEG and blood oxygen signal sequences are extracted and paired analysis is performed to identify the coordinated fluctuation characteristics of neural and blood oxygen responses, forming a dynamic quantitative judgment of the awakening process, and enhancing the accuracy and timeliness of state recognition. Attached Figure Description

[0015] 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.

[0016] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0019] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0020] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0021] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0022] Please see Figure 1 This invention provides a method for dynamically assessing the recovery progress of patients during the anesthesia recovery phase in the operating room, including the following steps: S1: The intervention behaviors recorded during the anesthesia recovery phase after surgery are obtained through the nurse's operating terminal. The trigger time, operation type and execution order of each intervention behavior are marked. An intervention operation chain is established according to the time sequence of the intervention behaviors. The operation chain is then segmented according to the time granularity to generate an operation behavior sequence. S2: Call the sequence of operational behaviors, summarize the number of occurrences of differentiated intervention types in each time segment, perform stratified statistics on intervention types according to preset levels, weight and superimpose the number and intensity levels of intervention behaviors in the same time segment, establish a corresponding time scale mapping based on the temporal continuity of intervention operations, and generate an intervention frequency distribution map. S3: Call the intervention frequency distribution map, collect the intervention density change results of continuous time periods within the equal-width sliding window, calculate the direction of the intervention density difference between two consecutive time windows, and determine whether the change trend is continuously rising and falling. Extract the time nodes with density change tendency within the detection period and output the trend change marker. S4: Call the trend change marker, compare the continuous density increase direction in the marker with the current assessment time period, and control the time boundary to shrink towards the direction of intervention density concentration based on the judgment result. If the density decrease direction appears continuously in the marker, the time boundary is extended backward, and the adjusted time boundary combination is constructed according to the change segment to generate the time assessment boundary segment. S5: Call the time assessment boundary segment, extract continuous vital sign monitoring data within the time segment, obtain EEG rhythm signal sequence and blood oxygen saturation signal sequence respectively, perform synchronous pairing of the two types of sequences according to the time axis and compare the change cycle and fluctuation direction, identify whether there is a staged coordinated fluctuation phenomenon in the neural and blood oxygen response dimensions, and output the dynamic assessment conclusion of the patient's awakening progress.

[0023] The sequence of operational behaviors includes intervention trigger time stamps, operation type codes, execution sequence numbers, and time granularity segments. The intervention frequency distribution map includes intervention type level stratification, weighted intervention intensity values, and temporal continuity mapping. The trend change markers include the direction of intervention density difference, trend change continuity indicators, and change tendency time nodes. The time assessment boundary segments include boundary contraction segments, boundary extension segments, and adjusted boundary combinations. The dynamic assessment conclusions of the patient's awakening progress include EEG rhythm change characteristics, blood oxygen saturation fluctuation characteristics, and neuro-blood oxygen synergistic fluctuation relationships.

[0024] Please see Figure 2 The specific steps of S1 are as follows: S101: Obtain the intervention behaviors during the anesthesia recovery stage recorded in the nurse's operating terminal, extract the timestamp field from each intervention behavior record, use the timestamp field as the trigger time of the intervention behavior, call the operation field in the corresponding intervention record, obtain the operation instruction content, mark the operation sequence of the intervention behavior by traversing the sequence number field, and generate an intervention behavior parameter set; First, initialize the communication connection with the underlying data interface of the hospital's anesthesia recovery room nursing records, and verify access permissions through the Secure Sockets Layer protocol. Initiate a data retrieval request to the server, setting the retrieval time range to the entire duration of the current anesthesia recovery cycle, for example, from 08:00:00 upon patient admission to 12:00:00 upon patient discharge. Receive the data stream returned by the server, which consists of a series of unstructured log entries, each stored in key-value pairs in JSON format. Start the log parsing engine to scan thousands of log records one by one, first identifying the "event attribute" key-value pair in each record, and filtering only records with the attribute value of "human intervention" or "nursing operation". For each filtered intervention record, perform field extraction and cleaning operations. Locate the timestamp field with the key name "Timestamp" and read the stored 13-digit long integer value (e.g., 1672531200000), which represents the number of milliseconds since the standard epoch. The time conversion service is invoked to convert the long integer value into the standard time format of hours, minutes, and seconds (HH:mm:ss), i.e., 08:00:00, and this is defined as the trigger time for the intervention. Next, the "OperationPayload" field in the same record is parsed to extract the specific text instructions, such as "suctioning operation," "gentle tap to wake up," and "adjust oxygen flow," removing leading and trailing whitespace and special escape characters. Simultaneously, the "SequenceID" field in the record is read. This field is an auto-incrementing integer automatically assigned by the terminal when the record is generated, used to distinguish the order of concurrent operations within the same millisecond. The parsed trigger time, the cleaned operation instructions, and the sequence number are associated and encapsulated into an independent intervention data object. The above extraction and encapsulation process is repeated for all compliant records within the retrieval period, ultimately aggregating all data objects into an in-memory list to generate an intervention parameter set. For example, the parameter set includes object A (08:15:30, airway clearing, ID 101) and object B (08:15:30, intravenous administration, ID 102), thereby transforming the original discrete log data into a structured set of parameters that can be processed by subsequent algorithms.

[0025] S102: Based on the trigger time, operation type and operation sequence extracted from the intervention behavior parameter set, the operation sequence is used as the sorting basis to sort all intervention behaviors in ascending order. An operation linked list structure is constructed according to the sorting result. Adjacent intervention behavior nodes are connected through the linked list structure to establish the intervention operation linked list structure data. First, space is allocated in the memory heap to construct a doubly linked list structure, and the head and tail pointers of the linked list are initialized to null. The intervention behavior parameter set is loaded, and all data objects in the set are rearranged using multi-key sorting logic. "Trigger time" is set as the first sorting key, and "sequence number" as the second sorting key. The parameter set is traversed, first comparing the trigger times of two objects. If the times are different, they are sorted in ascending order of time; if the trigger times are exactly the same (e.g., both are 09:10:00), their sequence numbers are further compared, placing the object with the smaller number first to ensure the absolute logical order of the operations. After completing the ascending sort of all data, the linked list nodes are constructed. Each sorted intervention behavior object is read sequentially, and a linked list node is created for it. Each node contains three main areas: a data field to store the object's trigger time, operation content, and sequence number; a predecessor pointer field to store the memory address of the previous node in the sequence; and a successor pointer field to store the memory address of the next node in the sequence. The first sorted object is used as the head node, and subsequent objects are linked sequentially. The predecessor pointer of the current node points to the previous node, and the successor pointer of the previous node points to the current node, until all objects are linked. In this way, a linked list structure for intervention operations is established. This structure not only maintains the linear relationship of data in physical storage, but also solidifies the temporal connection between adjacent intervention actions through pointer logic. This allows for quick backtracking or look-ahead of the context of any operation by traversing the linked list. For example, it can be found that the operation immediately preceding "suspension" is "auscultation," and the operation immediately following "recording vital signs" is "auscultation."

[0026] S103: Call the timestamp information of each node in the intervention operation chain list structure data, divide the nodes in the operation chain list into time periods according to the preset time interval granularity, and generate an operation behavior sequence according to the original order of the nodes included in each time period. The system reads the preset time interval granularity parameter, which defines the standard length for discretizing the continuous time axis into slices, for example, setting the granularity to 300 seconds (i.e., 5 minutes). It accesses the head node of the intervention operation linked list structure data and extracts its trigger time (e.g., 08:00:00) as the starting reference point for the entire time axis. Starting from this reference point, the entire recovery monitoring cycle is divided into consecutive time buckets in 300-second increments. For example, the first time bucket covers 08:00:00 to 08:05:00, the second time bucket covers 08:05:00 to 08:10:00, and so on. The system starts the linked list traversal program, visiting each node in the linked list starting from the head node. For each visited node, its trigger time is read and compared with the start and end boundaries of each time bucket. If the trigger time of a node is greater than or equal to the start time of a time bucket and less than the end time of that bucket, the reference address of that node is added to the member list of that time bucket. During the inclusion process, the relative order of nodes in the original linked list is strictly maintained, and no secondary sorting is performed. If there are no intervention records in a time bucket, the member list of that bucket is empty. After traversal, the nodes collected in each time bucket are arranged in their original order to form the subsequence corresponding to that time period. The subsequences corresponding to all time buckets are concatenated in chronological order, with time period index information attached, to finally generate an operation behavior sequence. This sequence maps the streaming operation records to a fixed time window grid, providing a standardized data foundation for subsequent frequency statistics and intensity analysis based on time periods.

[0027] Please see Figure 3 The specific steps of S2 are as follows: S201: Call the time segment data in the operation behavior sequence, traverse the intervention instruction content in each time segment, identify the intervention type identifier corresponding to the differentiated behavior instruction, and generate an intervention type frequency matrix by counting the number of intervention types in the same time segment and counting them independently according to the intervention type. The system retrieves data for each time segment in the sequence of operational behaviors sequentially. For the currently processed time segment (e.g., segment with index 5, corresponding to 08:20:00 to 08:25:00), it obtains the text content of all intervention instructions contained within that segment. A pre-built intervention behavior classification dictionary is loaded, which establishes mapping rules between clinical operational terms and standard intervention type identifiers. Keyword matching and semantic classification are performed on each instruction text. For example, the instructions "call out loudly" and "ask for name" are mapped to the "verbal arousal" type identifier; "pat the shoulder" and "press the tiger's mouth" are mapped to the "tactile stimulation" type identifier; and "suction" and "lift the chin" are mapped to the "airway management" type identifier. After completing the type mapping, a one-dimensional counter array is initialized in memory, with the array index corresponding to different intervention types. All mapping results within the current segment are traversed, and for each type encountered, the counter value at the corresponding index is incremented by 1. If 4 instances of verbal arousal and 2 instances of tactile stimulation are identified within the segment, the corresponding counter values ​​are 4 and 2, respectively. The above traversal and counting operations are repeated for each time segment in the sequence of intervention behaviors. The counter arrays generated for each segment are then concatenated in chronological order as column vectors to ultimately generate a two-dimensional intervention type frequency matrix. This matrix fully quantifies the temporal distribution density of different types of intervention behaviors throughout the recovery process.

[0028] S202: Based on the frequency matrix of intervention types, call the preset set of intervention level parameters, perform level matching on intervention types within the same time segment, extract the level coefficient corresponding to each intervention type, and perform a weighted superposition method to multiply the number of intervention types and the level coefficient within the time segment and then sum them to obtain the numerical distribution table of intervention intensity. The frequency matrix of intervention types is read, and a preset set of intervention level parameters is invoked simultaneously. This parameter set is pre-set based on clinical expert consensus, assigning a quantified level coefficient to each intervention type to characterize the intensity of stimulation of the patient's nerves by the behavior. For example, the parameter set sets the level coefficient for "verbal arousal" to 1.5, the level coefficient for "tactile stimulation" to 3.0, and the level coefficient for "airway management" to 5.0. A weighted summation operation is performed for each time segment (column vector) in the frequency matrix. A specific calculation example is as follows: Assuming the current treatment time segment is T5 (08:20 to 08:25), the frequency statistics for this segment show: verbal arousal 4 times, tactile stimulation 2 times, and airway management 0 times. First, the component value of verbal arousal is calculated: frequency 4 is multiplied by the level coefficient 1.5, resulting in a value of 6.0. Then, the component value of tactile stimulation is calculated: frequency 2 is multiplied by the level coefficient 3.0, resulting in a value of 6.0. Finally, the airway management component value was calculated: frequency 0 was multiplied by the grade coefficient 5.0, resulting in a value of 0.0. All the above component values ​​were then summed: 6.0 + 6.0 + 0.0, yielding a comprehensive intervention intensity value of 12.0 for this time segment. This process was repeated for all time segments in the matrix to calculate the intensity value for each segment, and these values ​​were arranged chronologically to construct an intervention intensity value distribution table. As shown in Table 1, this table records the calculation results for each time period.

[0029] Table 1. Example of numerical calculation results for intervention intensity. Time Segment Index Time range Frequency of each type (verbal / tactile / aura) Strength calculation logic Total Intervention Intensity T_05 08:20:00-08:25:00 4 / 2 / 0 4×1.5+2×3.0+0×5.0 12.0 T_06 08:25:00-08:30:00 2 / 5 / 1 2×1.5+5×3.0+1×5.0 23.0 T_07 08:30:00-08:35:00 1 / 0 / 0 1×1.5+0×3.0+0×5.0 1.5 The advantage of this operational logic is that by introducing a grading coefficient, the number of operations is transformed into a physical quantity that reflects the actual physiological stimulus load, thus giving higher weight to high-intensity operations in the evaluation.

[0030] S203: Based on the time segment sequence in the intervention intensity numerical distribution table, establish a time scale axis of equal length according to the start and end positions of each time segment, map the intervention intensity values ​​onto the time scale axis according to the time segment position, connect adjacent data points in chronological order to form a continuous numerical curve, and generate an intervention frequency distribution map. Read the data items from the intervention intensity value distribution table. Construct a virtual two-dimensional coordinate system in the memory drawing buffer, with the horizontal axis representing continuous time scales and the vertical axis representing intervention intensity values. Establish time scale points of equal length on the horizontal axis according to the start and end positions of each time segment; for example, use the center time of each time segment as the mapping reference point. Traverse the distribution table, mapping the intervention intensity value calculated for each time segment to the corresponding time reference point. For example, map the intensity value of 12.0 for time segment T5 to 08:22:30 on the time axis, and map the intensity value of 23.0 for T6 to 08:27:30. After completing the spatial mapping of all numerical points, use a cubic spline interpolation algorithm to connect adjacent data points. Construct a cubic polynomial function between every two adjacent points to ensure that the curve at the connection point is smooth and the first derivative is continuous, thereby eliminating discrete jumps caused by time slices. In this way, discrete intensity value points are fitted into a continuous numerical curve fluctuating along the time axis; this curve object is defined as the intervention frequency distribution map. This diagram visually illustrates the changing trajectory of the intensity of interventions applied to patients by healthcare professionals over time, providing a visual geometric basis for identifying peak intervention periods.

[0031] Please see Figure 4 The specific steps of S3 are as follows: S301: Call the intervention intensity values ​​on the continuous time axis of the intervention frequency distribution map, set the equal-width sliding window parameters, and sequentially extract the set of intervention intensity values ​​in the corresponding time period by sliding the window. Calculate the arithmetic mean of all intervention intensity values ​​in each sliding window to generate a sliding window intervention density sequence. Retrieve the sequence of intervention intensity values ​​from the continuous time axis of the intervention frequency distribution chart. Set a sliding window parameter of equal width, for example, a window width of 3 time units (corresponding to 15 minutes) and a sliding step size of 1 time unit. Place the sliding window at the beginning of the sequence, covering the first 3 intensity values. Calculate the arithmetic mean of the values ​​within the window. A specific calculation example is as follows: Assume the intensity values ​​of the three consecutive time points covered by the current window are 12.0, 23.0, and 1.5. First, perform an addition operation on these three values ​​to obtain a sum: 12.0 + 23.0 + 1.5 = 36.5. Then, divide the sum 36.5 by the window width of 3, performing a division operation to obtain an average value of 12.17 (rounded to two decimal places). Record this average value of 12.17 as the smooth density value at the center position of the current window. Next, move the sliding window one step forward, moving out of the first value 12.0, and read in the next value in the sequence (e.g., 8.0). The values ​​within the new window become 23.0, 1.5, and 8.0. Perform the summation and averaging operation again: (23.0 + 1.5 + 8.0) divided by 3 equals 10.83. Repeat the above sliding and calculation process until the window has traversed the entire sequence. Arrange all the calculated averages in order of window position to generate a sliding window intervention density sequence. This process suppresses local noise in the data through smoothing and highlights the macroscopic trend of intervention intensity.

[0032] S302: Based on the density values ​​of two adjacent sliding window positions in the density sequence, perform a difference operation between the density value at the current position and the density value of the previous window, extract the positive or negative sign of the difference to determine the direction of density change, and obtain the density change trend direction sequence by continuously judging whether the direction of the difference before and after is reversed. The sliding window intervention density sequence is read, and point-by-point analysis is performed starting from the second data point in the sequence. The density value at the current position is extracted and compared with the density value at the previous window position, and a difference calculation is performed. A specific example is as follows: assuming the density value at the previous position is 12.17 and the density value at the current position is 10.83. The difference is calculated as 10.83 minus 12.17, resulting in -1.34. The direction of density change is determined based on the sign of the difference: a positive difference is marked as "increasing"; a negative difference is marked as "decreasing"; and a zero difference is marked as "unchanged". In this example, it is determined to be "decreasing". This operation is performed on each pair of adjacent data points in the sequence, generating a direction label sequence. Subsequently, the direction label sequence is continuously scanned to identify nodes where the direction of change reverses. Two consecutive direction labels are checked; if the previous one is "increasing" and the next one is "decreasing", a "peak reversal" is determined at that position; if the previous one is "decreasing" and the next one is "increasing", a "valley reversal" is determined. Record all identified reversal states and their index positions in the sequence to ultimately obtain the density change trend direction sequence.

[0033] S303: Based on the sign inflection point position of density direction change in the density change trend direction sequence, retrieve all time index nodes where direction change occurs within the continuous detection period, extract the corresponding time position on the time scale axis, and output the trend change marker. Based on the density change trend sequence, a key event point retrieval operation is performed. The retrieval scope is set to the entire monitoring period, focusing on finding all index nodes where sign reversals (i.e., peak reversals or trough reversals) occur. For each retrieved reversal index, the specific time point corresponding to that index is calculated backwards based on the sliding window step size and start time. For example, if a "trough reversal" (from decreasing to increasing) occurs at the 10th window position, and each window step size is 5 minutes, the time corresponding to this reversal point is calculated to be 08:50:00. This time position is extracted and marked as a "trend reversal point." The entire sequence is traversed, collecting the time information of all reversal points, and they are organized into a list structure according to chronological order, outputting trend change markers. This list clearly indicates the key moments when the intervention intensity fundamentally changes, such as the starting point of the intervention's strengthening or the ending point of weakening after reaching its peak, providing precise time anchors for subsequently determining key assessment periods.

[0034] Please see Figure 5 The specific steps of S4 are as follows: S401: Call the time marker node in the trend change marker, extract the continuously arranged density growth direction marker, set the start and end boundary values ​​of the current evaluation time period, and index and compare the continuous density growth markers with the time axis scale, filter all growth marker nodes within the current time period range, and obtain the density concentration trend node set; The process involves calling trend change markers and extracting all consecutive marker segments identified as "density growth direction" (i.e., between a trough reversal and the subsequent most recent peak reversal). The current assessment timeframe is defined, for example, an assessment window from 09:00:00 to 10:00:00. The timestamp of each growth marker node in the list is compared to the start and end boundaries of this assessment window. A filtering logic is then executed: if the timestamp of a marker node falls within the closed interval of 09:00:00 to 10:00:00, the node is considered valid and retained; otherwise, it is discarded. This filtering process removes fluctuation data outside the current focus period, retaining only nodes within the assessment window that show a significant increase in intervention density. These retained nodes constitute a density concentration trend node set, reflecting the specific time clusters of intensive interventions performed on patients by healthcare professionals within a specific timeframe.

[0035] S402: Based on the density concentration trend node set, analyze the order of time nodes and the boundary positions of the current evaluation time period, and determine whether the direction of change trend tends to approach the center position of the time period. If the judgment is true, adjust the start and end boundaries to decrease towards the center position in turn to obtain the boundary contraction interval parameter set. Based on the temporal distribution characteristics of the node concentration trend, the boundary of the evaluation time period is dynamically optimized. First, the average time position of all time points in the node set is calculated as the intervention focus. Then, the deviation of this focus from the geometric center of the current evaluation time period is analyzed, and it is determined whether the trend shows a tendency towards the center. The specific logic is as follows: Assuming the evaluation window is from 09:00 to 10:00 (center 09:30), the earliest time of the node set is 09:15, and the latest time is 09:45. It is determined that the time span of the node set (30 minutes) is significantly smaller than the evaluation window (60 minutes) and is located within the window, thus confirming the judgment of "tending towards the center." A boundary contraction operation is performed: a new starting boundary is set as the earliest time point of the node set minus a preset buffer margin (e.g., 120 seconds): 09:15:00 minus 120 seconds yields 09:13:00. Set a new end boundary as the latest time point of the node set plus a preset buffer margin (e.g., 120 seconds): 09:45:00 plus 120 seconds equals 09:47:00. Through this step, the evaluation range is precisely adjusted from the broad 09:00-10:00 to 09:13-09:47, eliminating the blank periods at the beginning and end where there is no substantial intervention, thus obtaining the parameter set of the boundary contraction interval.

[0036] S403: Based on the boundary adjustment results of the boundary contraction interval parameter set, the current evaluation time period is replaced according to the adjusted start and end boundaries to construct a continuous change segment structure, and a combination aggregation operation is performed on all change segments to generate time evaluation boundary segments. The system reads the modulation results from the parameter set of the boundary contraction interval and replaces the original assessment time period with the adjusted start and end boundaries. A continuously changing segment structure object is constructed for each adjusted time period. Since multiple adjacent intervention peaks may exist, a combination aggregation operation is performed to handle overlaps or proximity between segments. All segments are sorted by start time, and adjacent segments A and B are checked one by one. If the end time of segment A is greater than or equal to the start time of segment B, or the time interval between them is less than a preset small threshold (e.g., 60 seconds), these two segments are determined to be merged. They are merged into a new large segment, with the start time being the minimum of the two and the end time being the maximum. After multiple rounds of iterative checking and merging, until the time interval between all segments is greater than the threshold, a set of independent time assessment boundary segments is finally generated. These segments precisely lock each independent active cycle during the patient's awakening process, serving as a precise time filter for subsequent extraction of physiological signals.

[0037] Please see Figure 6 The specific steps of S5 are as follows: S501: Call the start and end boundaries in the time assessment boundary segment, extract vital sign monitoring data frames by time period, and extract the original signals of the EEG channel and the blood oxygen monitoring channel at the corresponding sampling frequency within the extraction range. Separate the data according to the channel type to obtain the EEG and blood oxygen combined signal sequence set. The system retrieves the aggregated time assessment boundary segment (e.g., 09:13:00 to 09:47:00) and sends a data request to the bedside monitor via the internal data bus, capturing the full range of vital sign data frames within that time frame. Multi-channel signals are parsed from the data frames, and the raw EEG signal and blood oxygen saturation (SpO2) monitoring signal are separated based on channel identifiers. The raw sampling frequencies of the two types of signals are identified, for example, 250Hz for EEG signals and 1Hz for SpO2 signals. Digital filtering is performed on the EEG signal. The bandpass filter frequency range is set to 13Hz to 30Hz (Beta band). The raw EEG sequence is input, low-frequency artifacts and high-frequency noise are filtered out, and a rhythmic signal reflecting the activity of the cerebral cortex is output. Noise reduction processing is performed on the SpO2 signal to remove mutation artifacts. Finally, the processed EEG and SpO2 sequences are classified and stored according to channel type to obtain a combined EEG and SpO2 signal sequence set. This collection contains high-quality synchronous physiological signals during the intensive intervention period, laying the data foundation for subsequent analysis.

[0038] S502: Based on the combined EEG and blood oxygenation signal sequence set, the EEG rhythm signal sequence and blood oxygenation saturation signal sequence are paired point by point using a synchronous indexing method under a unified time axis. Local cycles and fluctuation trends are extracted respectively. The phase change direction and cycle duration in adjacent intervals are compared to generate a neural blood oxygenation fluctuation comparison result matrix. The system processes a set of combined EEG and blood oxygenation signal sequences. Due to the different sampling rates of the two signal types, a downsampling algorithm is first used to unify the high-frequency EEG signals to the same low-frequency time axis (e.g., 1Hz) as the blood oxygenation signals, achieving point-by-point pairing. The system extracts the signal fluctuation characteristics. For the EEG sequence, its energy amplitude per second is calculated to reflect the degree of neural excitation; for the blood oxygenation sequence, its first-order difference is calculated to reflect the trend of oxygenation changes. The phase change direction and period duration are compared within adjacent intervals. A specific example is as follows: within a 10-second interval, if the EEG energy amplitude continuously increases (positive), and the blood oxygen value also continuously increases (positive), the two are determined to be in phase, with a "positive" direction and a duration of 10 seconds. If the EEG increases while the blood oxygen decreases, the phases are determined to be opposite. A neural blood oxygenation fluctuation comparison matrix is ​​generated by traversing all intervals, recording the phase relationship (same direction / opposite direction) and duration for each time period.

[0039] S503: Based on the consistent characteristics of the phased cycle direction extracted from the comparison matrix of nerve blood oxygen fluctuations, the number of time segments with synchronous rising and synchronous falling trends within the time period is counted and marked as response event node sequences. Adjacent response event nodes are connected and merged to output the dynamic assessment conclusion of the patient's awakening progress. Statistical analysis was performed based on the comparison matrix of neuro-oxygen fluctuations. All time periods marked "phase-consistent" (i.e., synchronous rise or synchronous fall) in the matrix were selected, and their occurrence and cumulative total duration were counted. A specific calculation example is as follows: assuming a total assessment duration of 2040 seconds. 15 "synchronous rise" segments were found, totaling 600 seconds; 10 "synchronous fall" segments were found, totaling 400 seconds. The total effective coupling time was calculated as 600 + 400 = 1000 seconds. The coupling ratio was calculated as 1000 divided by 2000, resulting in 50%. A conclusion was output based on preset assessment criteria: if the ratio exceeds 40%, it is judged as "good awakening progress, normal neuro-respiratory coupling"; otherwise, it is judged as "delayed awakening or separation". In this example, the output is "Dynamic assessment conclusion of patient's awakening progress: the nervous and respiratory systems show a benign synchronous response to the intervention, and the awakening quality meets expectations." As shown in Table 2, this conclusion quantifies the patient's physiological recovery status.

[0040] Table 2. Conclusion of Dynamic Assessment of Patient Awakening Progress Evaluation indicators Statistical results Threshold Standard Judgment Conclusion Synchronous rise duration 600 seconds - - Synchronous descent duration 400 seconds - - Coupling time percentage 50.0% >40% good Comprehensive assessment The patient's nervous and respiratory systems responded synchronously to the intervention, resulting in excellent quality of awakening. excellent

[0041] The experimental results show that the calculated 50% coupling degree, compared with traditional methods, can more accurately reflect the patient's physiological integration ability during the recovery period, and effectively assist medical staff in making extubation decisions.

[0042] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the technical solution.

Claims

1. A method for dynamically assessing the recovery progress of patients during the anesthesia recovery phase in the operating room, characterized in that... Includes the following steps: S1: Obtain the intervention behaviors recorded during the anesthesia recovery phase, mark the trigger time, operation type and execution order, construct the intervention operation chain according to the time sequence, and generate the operation behavior sequence by segmenting according to the preset time interval granularity; S2: Call the operation behavior sequence, summarize the occurrence frequency of time segment intervention types, call the preset intervention level parameter set, perform hierarchical statistics on intervention types, perform multiplication operation according to preset level coefficients and sum them, and superimpose the number of intervention behaviors and intensity levels in a weighted manner. Combine the continuity of intervention operation time to construct a time scale mapping and generate an intervention frequency distribution map. S3: Call the intervention frequency distribution map, collect the intervention density change results within the sliding window, calculate the direction of the intervention density difference between adjacent windows, determine the trend change, extract the time nodes with density change tendency, and output the trend change marker. S4: Call the trend change marker, compare the density growth direction with the current evaluation time period, control the time boundary to shrink and extend towards the density concentration feature direction, and construct the time evaluation boundary segment; S5: Call the time assessment boundary segment, extract vital sign monitoring data, obtain EEG rhythm signal sequence and blood oxygen saturation signal sequence, analyze the period and fluctuation direction, identify the coordinated fluctuation phenomenon, and output the dynamic assessment conclusion of the patient's awakening progress.

2. The method for dynamically assessing patient awakening progress during the anesthesia recovery phase in the operating room according to claim 1, characterized in that, The operational behavior sequence includes intervention trigger time stamps, operation type codes, execution sequence numbers, and time granularity segments. The intervention frequency distribution map includes intervention type level stratification, weighted intervention intensity values, and time continuity mapping relationships. The trend change markers include intervention density difference direction, trend change continuity indicators, and change tendency time nodes. The time assessment boundary segments include boundary contraction segments, boundary extension segments, and adjusted boundary combinations. The dynamic assessment conclusions of the patient's awakening progress include EEG rhythm change characteristics, blood oxygen saturation fluctuation characteristics, and neuro-blood oxygen synergistic fluctuation relationships.

3. The method for dynamically assessing patient awakening progress during the anesthesia recovery phase in the operating room according to claim 1, characterized in that, The time points where density changes tend to occur refer to key time points in the intervention frequency distribution map where the trend of change in intervention density, calculated through a sliding window, includes both increases and decreases.

4. The method for dynamically assessing patient awakening progress during the anesthesia recovery phase in the operating room according to claim 1, characterized in that, The aforementioned phenomenon of coordinated fluctuation refers to the phenomenon in which the periodic changes in EEG rhythm and blood oxygen saturation signals maintain a consistent and synchronous direction.

5. The method for dynamically assessing patient awakening progress during the anesthesia recovery phase in the operating room according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Obtain the intervention behaviors during the anesthesia recovery stage recorded in the nurse's operating terminal, extract the timestamp field from each intervention behavior record, use the timestamp field as the trigger time of the intervention behavior, call the operation field in the corresponding intervention record, obtain the operation instruction content, mark the operation sequence of the intervention behavior by traversing the sequence number field, and generate an intervention behavior parameter set; S102: Based on the trigger time, operation type and operation sequence extracted from the intervention behavior parameter set, the operation sequence is used as the sorting basis to sort all intervention behaviors in ascending order. An operation linked list structure is constructed according to the sorting result. Adjacent intervention behavior nodes are connected through the linked list structure to establish intervention operation linked list structure data. S103: Call the timestamp information of each node in the intervention operation chain list structure data, divide the nodes in the operation chain list into time periods according to the preset time interval granularity, and generate an operation behavior sequence according to the original order of the nodes included in each time period.

6. The method for dynamically assessing patient awakening progress during the anesthesia recovery phase in the operating room according to claim 1, characterized in that, The specific steps of S2 are as follows: S201: Call the time segment data in the operation behavior sequence, traverse the intervention instruction content in each time segment, identify the intervention type identifier corresponding to the differentiated behavior instruction, and generate an intervention type frequency matrix by counting the number of intervention types in the same time segment and counting them independently according to the intervention type. S202: Based on the frequency matrix of intervention types, call the preset set of intervention level parameters, perform level matching on intervention types within the same time segment, extract the level coefficient corresponding to each intervention type, and perform a weighted superposition method to multiply the number of intervention types and the level coefficient within the time segment and then sum them to obtain the numerical distribution table of intervention intensity. S203: Based on the time segment sequence in the intervention intensity numerical distribution table, establish a time scale axis of equal length according to the start and end positions of each time segment, map the intervention intensity values ​​onto the time scale axis according to the time segment positions, connect adjacent data points in chronological order to form a continuous numerical curve, and generate an intervention frequency distribution map.

7. The method for dynamically assessing patient awakening progress during the anesthesia recovery phase in the operating room according to claim 1, characterized in that, The specific steps for S3 are as follows: S301: Call the intervention intensity values ​​on the continuous time axis of the intervention frequency distribution map, set the equal-width sliding window parameters, and sequentially extract the set of intervention intensity values ​​in the corresponding time period by sliding the window. Calculate the arithmetic mean of all intervention intensity values ​​in each sliding window to generate a sliding window intervention density sequence. S302: Based on the density values ​​of two adjacent sliding window positions in the sliding window intervention density sequence, perform a difference operation between the current position density value and the previous window density value, extract the positive or negative sign of the difference to determine the direction of density change, and obtain the density change trend direction sequence by continuously judging whether the direction of the difference before and after is reversed. S303: Based on the sign inflection point position of the density direction change in the density change trend direction sequence, retrieve all time index nodes where the direction change occurs within the continuous detection period, extract the corresponding time position on the time scale axis, and output the trend change mark.

8. The method for dynamically assessing patient awakening progress during the anesthesia recovery phase in the operating room according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Call the time identifier node in the trend change marker, extract the continuously arranged density growth direction marker, set the start and end boundary values ​​of the current evaluation time period, and index and compare the continuous density growth markers with the time axis scale, filter all growth marker nodes within the current time period range, and obtain the density concentration trend node set; S402: Based on the density concentration trend node set, analyze the order of time nodes and the boundary positions of the current evaluation time period, and determine whether the direction of change trend tends to be closer to the center position of the time period. If the judgment is true, adjust the start and end boundaries to decrease towards the center position in turn to obtain the boundary contraction interval parameter set. S403: Based on the boundary adjustment results of the boundary contraction interval parameter set, the current evaluation time period is replaced according to the adjusted start and end boundaries to construct a continuous change segment structure, and a combination aggregation operation is performed on all change segments to generate time evaluation boundary segments.

9. The method for dynamically assessing patient awakening progress during the anesthesia recovery phase in the operating room according to claim 1, characterized in that, The specific steps of S5 are as follows: S501: Call the start and end boundaries in the time assessment boundary segment, extract vital sign monitoring data frames according to time period, and extract the original signals of the EEG channel and the blood oxygen monitoring channel at the corresponding sampling frequency within the extraction range. Separate the data according to the channel type to obtain the EEG and blood oxygen combined signal sequence set. S502: Based on the combined EEG and blood oxygenation signal sequence set, the EEG rhythm signal sequence and blood oxygenation signal sequence are paired point by point using a synchronous indexing method under a unified time axis. Local cycles and fluctuation trends are extracted respectively. The phase change direction and cycle duration in adjacent intervals are compared to generate a neural blood oxygenation fluctuation comparison result matrix. S503: Based on the phased periodic direction consistency characteristics extracted from the neural blood oxygen fluctuation comparison result matrix, count the number of time segments with synchronous upward and synchronous downward trends within the time period, and mark them as response event node sequences. Connect and merge adjacent response event nodes to output the dynamic assessment conclusion of the patient's awakening progress.

10. The method for dynamically assessing patient awakening progress during the anesthesia recovery phase in the operating room according to claim 9, characterized in that, The response event node sequence refers to the set of consecutive time points where the EEG rhythm and blood oxygen saturation signal show the same trend on a unified time axis, including the phase direction and period synchronization, marking the temporal nodes of the neuro-blood oxygen coordinated response during the patient's awakening process.

Images