Urinary post-operation sign monitoring data processing method and system

Abstract

The invention relates to the technical field of medical data processing, in particular to a urinary post-operation sign monitoring data processing method and system, and the method comprises the steps: obtaining original urinary post-operation sign time sequence flows of a plurality of monitoring devices, and carrying out the verification of data integrity and consistency. And creating a dynamic sign map for describing physiological state evolution, and injecting sign nodes and state transition edges into the dynamic sign map based on the verified time sequence flow. And performing staged partitioning on the dynamic sign map according to a postoperative rehabilitation stage division standard. And deploying independent abnormal probes matched with the stage characteristics in each partition, and performing parallel operation to scan and collect potential abnormal signals. And constructing an abnormal signal aggregation network, and carrying out fusion and priority ranking on the collected abnormal signals. And according to the network output, generating a differentiated graded early warning instruction, and calling a corresponding early warning response protocol for execution. According to the method, modeling of continuous evolution of the postoperative physiological status and staged adaptive anomaly monitoring are realized.

Description

Technical Field

[0001] This invention relates to the field of medical data processing technology, and in particular to a method and system for processing postoperative urological signs monitoring data. Background Technology

[0002] Current postoperative vital sign monitoring methods for urological surgeries largely rely on independent observation or simple trend analysis of vital sign data collected at discrete time points. A common approach involves setting fixed warning thresholds, triggering an alarm when a vital sign exceeds the threshold. These methods fragment the continuous postoperative recovery process into isolated monitoring points, failing to reflect the dynamic correlations between vital sign parameters and the coherent evolution of physiological states.

[0003] Fixed-threshold early warning systems lack consideration for the differences in physiological characteristics at different stages of postoperative recovery. The normal fluctuation range of vital signs and their clinical significance vary significantly between different stages, such as the early postoperative period and the recovery phase. Using a uniform standard for monitoring can easily lead to a higher false alarm rate at one stage and an increased risk of missed alarms at another. Current technology cannot adaptively adjust monitoring strategies according to the recovery process, limiting the accuracy and timeliness of early warnings.

[0004] This invention aims to address how to structurally integrate discrete time-series vital sign data to continuously and intuitively model the complete evolution path of postoperative physiological states. Simultaneously, it seeks to overcome the limitations of fixed monitoring modes and achieve targeted and adaptive anomaly detection based on the physiological characteristics of different recovery stages, thereby improving the accuracy of postoperative monitoring. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the existing technology and to propose a method and system for processing postoperative physical signs monitoring data in urology.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for processing postoperative physical signs monitoring data for urological surgery, comprising:

[0007] Acquire raw postoperative urological signs time-series data from multiple monitoring devices;

[0008] The original postoperative urological signs time-series stream is subjected to data integrity and consistency verification, and the verified time-series signs stream is output.

[0009] Create a dynamic atlas of physical signs to describe the evolution of physiological status after urological surgery;

[0010] Based on the verified temporal symptom flow, symptom nodes and state transition edges are injected into the dynamic symptom graph;

[0011] Based on the preset postoperative rehabilitation stage division criteria, the dynamic vital signs atlas is divided into stages to generate a staged vital signs atlas.

[0012] Within each region of the staged physiological characteristic map, independent abnormal probes matching the physiological stage characteristics of the region are deployed.

[0013] Each independent anomaly probe is run synchronously to perform parallel scanning of the staged vital signs map and collect potential anomaly signals in each partition.

[0014] An abnormal signal aggregation network is constructed, and the potential abnormal signals collected by each independent abnormal probe are input into the abnormal signal aggregation network for fusion and priority sorting.

[0015] Based on the output of the abnormal signal aggregation network, differentiated hierarchical early warning instructions are generated;

[0016] Based on the type and level of the tiered early warning instruction, the corresponding early warning response protocol is invoked and executed.

[0017] Preferably, the step of performing data integrity and consistency verification on the original post-urological urological signs time-series stream, and outputting the verified time-series signs stream includes:

[0018] Identify missing and conflicting signal segments in the time-series flow of the original postoperative urological signs;

[0019] The missing segments of the signal are interpolated and reconstructed by using the evolution trend of similar vital sign data in adjacent time periods to form continuous vital sign data;

[0020] Based on a preset physiological parameter mutual exclusion rule library, the conflicting signal segments are logically arbitrated, and vital sign data that conform to physiological logic are retained.

[0021] The data, after interpolation, reconstruction, and logical arbitration, are re-integrated to form a verified temporal characteristic stream that is continuous in the time dimension and self-consistent in the logical dimension.

[0022] Preferably, the creation of a dynamic atlas of signs describing the evolution of physiological state after urological surgery includes:

[0023] Define a graph node, which represents the combined state of multiple vital signs parameters at a specific time.

[0024] Define directed edges in the graph, where each directed edge represents a state transition path between nodes in the graph at different times.

[0025] For each of the directed edges of the graph, a transfer attribute is labeled, and the transfer attribute includes at least the transfer time and the main driving characteristic parameters;

[0026] Initialize a blank atlas structure containing a starting node as the framework for the dynamic symptom atlas.

[0027] Preferably, the step of injecting symptom nodes and state transition edges into the dynamic symptom map based on the verified temporal symptom stream includes:

[0028] The verified time-series vital sign data snapshots are segmented at fixed time intervals from the vital sign stream;

[0029] Each of the said vital sign data snapshots is mapped to a said atlas node, and the vital sign vector of the atlas node is calculated;

[0030] Based on the sequential order of the continuous vital sign data snapshots in the verified time-series vital sign stream, directed edges are established between the corresponding graph nodes;

[0031] The data injection of the dynamic vital signs graph is completed by using the actual data in the verified temporal vital signs stream to calculate and update the transition attributes of the directed edges of the graph.

[0032] Preferably, the step of dividing the dynamic vital signs atlas into stages according to a preset postoperative rehabilitation stage division standard to generate a staged vital signs atlas includes:

[0033] The preset postoperative rehabilitation stage division criteria are based on the stable threshold and fluctuation pattern definition of key vital signs indicators. The vital sign vectors of each node in the dynamic vital sign map are scanned along the time axis direction. When the vital sign vector characteristics of multiple consecutive nodes in the map meet the definition of a new rehabilitation stage, a stage boundary marker is created in the dynamic vital sign map.

[0034] Based on all stage boundary markers, the dynamic vital signs map is divided into multiple continuous stage sub-maps, which together constitute the staged vital signs map.

[0035] Preferably, the deployment of independent abnormal probes matching the physiological stage characteristics of each partition of the staged vital sign map includes:

[0036] Analyze the statistical distribution characteristics of the eigenvectors of all nodes in each stage sub-map;

[0037] Based on the statistical distribution characteristics, a specific set of anomaly detection rules and sensitivity parameters are configured for the stage sub-map. The set of anomaly detection rules and sensitivity parameters are then encapsulated to form an independent anomaly probe specifically designed for scanning the stage sub-map.

[0038] Each of the independent anomaly probes is loaded into the storage space of its corresponding stage submap.

[0039] Preferably, the synchronous operation of each of the independent anomaly probes to perform parallel scanning of the staged vital signs map and collect potential anomaly signals within each partition includes:

[0040] Each of the independent anomaly probes traverses each node of the map in chronological order within its corresponding stage sub-map.

[0041] The feature vector of the currently traversed graph node is matched and calculated with the set of anomaly determination rules inside the independent anomaly probe.

[0042] When the matching calculation result exceeds the sensitivity parameter of the independent anomaly probe, a potential anomaly signal containing the anomaly type, node location, and degree of deviation is generated;

[0043] Each independent anomaly probe outputs the potential anomaly signals it generates to a unified signal buffer in real time.

[0044] Preferably, the step of constructing the abnormal signal aggregation network, which involves inputting the potential abnormal signals collected by each of the independent abnormal probes into the abnormal signal aggregation network for fusion and priority ranking, includes:

[0045] The abnormal signal aggregation network receives all potential abnormal signals from the signal buffer;

[0046] Based on the clinical urgency of the sub-map from which the potential abnormal signal originates, a base weight is assigned.

[0047] The clustering of different potential abnormal signals over time and their correlation with vital signs parameters are analyzed. Cluster fusion is performed on potential abnormal signals with strong correlation to generate composite abnormal signals.

[0048] Based on the total deviation between the basic weights and the composite abnormal signals, the final priority score of each abnormal signal is calculated and sorted accordingly.

[0049] Preferably, generating differentiated hierarchical early warning instructions based on the output of the abnormal signal aggregation network includes:

[0050] Different instruction content templates and target response terminals are preset for different levels of early warning instructions;

[0051] The list of abnormal signals with priority sorting output by the abnormal signal aggregation network is mapped to the corresponding warning level;

[0052] Based on the mapped warning level, select the corresponding instruction content template and fill in the specific abnormal signal details to form the complete graded warning instruction.

[0053] Preferably, the present invention also includes a post-urological examination vital signs monitoring data processing system, the system including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor, when executing the computer program, implements the steps of the post-urological examination vital signs monitoring data processing method described above.

[0054] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0055] By creating a dynamic vital sign atlas and injecting vital sign nodes and state transition edges, traditional discrete time-series data streams are transformed into a graph structure model. Nodes represent the comprehensive vital sign state at a specific moment, while edges define the temporal causal relationships and evolution paths between states. This technology enables a structured representation of the interaction of multiple vital sign parameters and the continuous evolution of physiological states, allowing monitoring systems to track state transition trajectories rather than merely responding to isolated data point anomalies, thereby revealing potential trends deviating from the normal rehabilitation path earlier.

[0056] The dynamic vital signs atlas was divided into zones based on postoperative rehabilitation stage criteria, and independent abnormality probes were deployed within each zone. The detection logic of each probe was strictly matched with the physiological stage characteristics of its respective zone. Multiple probes scanned in parallel within their respective zones, achieving synchronous adaptation between the monitoring strategy and the rehabilitation process. This stage-specific customized detection mechanism reduced misjudgments caused by physiological differences at each stage, while enhancing the sensitivity to identify stage-specific abnormalities.

[0057] An abnormal signal aggregation network fuses and prioritizes potential abnormal signals acquired in parallel. This network comprehensively evaluates abnormal signals from different stages and symptom patterns, integrating and judging them based on their clinical urgency and relevance, ultimately outputting differentiated, tiered early warning instructions. This transforms the early warning response from a single signal trigger to an intelligent decision based on a multi-stage, multi-dimensional chain of evidence, enhancing the clinical relevance and guiding value of the early warning instructions. Attached Figure Description

[0058] Figure 1 This is a flowchart of the postoperative urological examination data processing method according to the present invention;

[0059] Figure 2 A flowchart for data integrity and consistency verification;

[0060] Figure 3 A flowchart for creating a dynamic vital signs map;

[0061] Figure 4 A bar chart showing the number of abnormal signals at each stage of postoperative rehabilitation for urological surgery;

[0062] Figure 5A heat map showing the postoperative recovery stage and physiological characteristics of urological surgery. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0064] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0065] See Figure 1 This process acquires raw postoperative urological signs time-series data from multiple monitoring devices that continuously collect various physiological parameters from patients. The raw postoperative urological signs time-series data is validated for data integrity and consistency, and a validated time-series sign data stream is output. A dynamic sign atlas is created to describe the evolution of physiological states after urological surgery, formally representing the dynamic changes in physiological states. Based on the validated time-series sign data stream, sign nodes and state transition edges are injected into the dynamic sign atlas to concretize it. According to a pre-defined postoperative rehabilitation stage division standard, the dynamic sign atlas is divided into staged partitions, generating a staged sign atlas that divides the continuous rehabilitation process into discrete clinical stages. Within each partition of the staged sign atlas, independent abnormal probes matching the physiological stage characteristics of that partition are deployed, each probe responsible for monitoring the sign patterns of a specific stage. The independent abnormal probes are run synchronously to perform parallel scanning of the staged sign atlas, collecting potential abnormal signals within each partition. An abnormal signal aggregation network is constructed. Potential abnormal signals collected by individual abnormal probes are input into the network for fusion and prioritization to form a global risk assessment. Based on the output of the abnormal signal aggregation network, differentiated hierarchical early warning instructions are generated, with the instruction level corresponding to the severity of the abnormal risk. According to the type and level of the hierarchical early warning instruction, the corresponding early warning response protocol is invoked and executed, driving the corresponding clinical intervention process.

[0066] In one embodiment of the present invention, see [reference] Figure 2This study identifies missing and conflicting signal segments in the original post-urological urological examination data stream. Missing signal segments refer to time periods in the data stream where no valid records are found. The missing signal segments are then imputed and reconstructed using the evolution trend of similar vital signs within adjacent time periods, forming continuous vital sign data. The imputation process is based on the trend line of data before and after the missing point. Conflicting signal segments are logically arbitrated according to a pre-defined physiological parameter mutual exclusion rule base, retaining vital sign data that conforms to physiological logic. The rule base defines reasonable coexistence relationships between parameters. The data after imputation, reconstruction, and logical arbitration are then reintegrated to form a verified temporal vital sign stream that is continuous in the time dimension and self-consistent in the logical dimension.

[0067] In specific implementation, the original post-urological examination vital signs time-series stream contains continuous time-series data from multiple monitoring devices. The process involves identifying missing and conflicting signal segments within this stream. Missing segments refer to numerical gaps at consecutive time points in the data record, while conflicting segments refer to logically contradictory vital sign parameter values ​​reported by different monitoring devices at the same time point. In some embodiments, missing segments are identified by scanning the continuity of timestamps in the original post-urological examination vital signs time-series stream. A missing segment is defined as an interval between adjacent timestamps exceeding a preset threshold. For example, if heart rate data is recorded as 75 beats / min at time point 10:00, 78 beats / min at time point 10:05, but not recorded at time point 10:02, this is marked as a missing segment. Conflicting segments are identified by comparing vital sign parameter values ​​from different sources at the same time point. For example, if systolic blood pressure monitoring device A reports 120 mmHg at time point 10:10 while systolic blood pressure monitoring device B reports 90 mmHg, this is marked as a conflicting segment. The missing segments of the signal are interpolated and reconstructed using the evolution trend of similar vital sign data within adjacent time periods to form continuous vital sign data. The interpolation and reconstruction are calculated based on the trend of valid data points before and after the missing segment. In specific implementation, for missing segments of the heart rate signal, linear interpolation is performed using two known heart rate values ​​before and after the missing point. The interpolation formula is as follows:

[0068] ;

[0069] in: This represents the heart rate interpolation value at time t. Indicates the most recent time point before the missing segment. Heart rate value, Indicates the most recent time point after the missing segment. Heart rate value, This indicates a specific time point within the missing segment. It's understandable that linear interpolation assumes heart rate changes linearly over short time intervals, which is suitable for a stable recovery period. Optionally, for non-stationary trends, a moving average-based interpolation method is used, reconstructing the segment using a weighted average of multiple data points before and after the missing point; for example, calculating the average of three data points before and after the missing segment.

[0070] Based on a pre-defined physiological parameter mutual exclusion rule library, logical arbitration is performed on conflicting signal segments, retaining vital sign data that conforms to physiological logic. The physiological parameter mutual exclusion rule library defines the reasonable range of relationships between parameters. In specific implementations, the physiological parameter mutual exclusion rule library includes rules such as "the difference between systolic and diastolic blood pressure should be greater than 20 mmHg and less than 60 mmHg." When one of the two values ​​in a conflicting segment violates this rule, it is discarded through arbitration. For example, if at time 10:15 the systolic blood pressure is reported as 130 mmHg and the diastolic blood pressure as 120 mmHg, the difference of 10 mmHg violates the rule, and the data is marked as invalid. In some embodiments, logical arbitration employs a voting mechanism. When multiple monitoring devices report the same parameter, the value consistent with the majority of devices or closest to the patient's historical baseline is selected. For example, if three heart rate monitoring devices report 72 bpm, 73 bpm, and 85 bpm respectively at time 10:20, the first two values ​​are close and retained, while the third value is excluded as an outlier through arbitration. The data, after interpolation reconstruction and logical arbitration, is re-integrated to form a verified time-series vital signs stream that is continuous in the time dimension and self-consistent in the logical dimension. The re-integration process arranges all processed data points in chronological order. For example, the interpolated heart rate value and the arbitrated blood pressure value are merged into a unified data stream by timestamp.

[0071] In one embodiment of the present invention, see [reference] Figure 3The process involves defining graph nodes, each representing a combination of multiple vital sign parameters at a specific moment, defined by a combination of parameter values. Directed graph edges represent state transition paths between graph nodes at different times, indicating the direction of physiological state change. Transition attributes are labeled for each graph edge, including at least the transition time and the main driving vital sign parameter; the transition time is determined by the time difference between nodes. A blank graph structure containing a starting node is initialized as the framework for the dynamic vital sign graph, with the starting node corresponding to the state at the initial monitoring moment. Vital sign data snapshots are segmented from the validated time-series vital sign stream at fixed time intervals, each snapshot containing all vital sign parameters collected at the same moment. Each vital sign data snapshot is mapped to a graph node, and a vital sign vector is calculated for each node; the vital sign vector is a numerical representation of the parameter values. Directed graph edges are established between corresponding graph nodes based on the chronological order of consecutive vital sign data snapshots in the validated time-series vital sign stream. The data injection of the dynamic vital signs graph is completed by using the actual data in the verified temporal vital signs stream to calculate and update the transition attributes of the directed edges of the graph.

[0072] In practice, a dynamic vital sign atlas is created to describe the evolution of physiological state after urological surgery. Based on validated time-series vital sign flow injection data, atlas nodes are defined. Each atlas node represents a combination of multiple vital sign parameters at a specific time point. For example, at time point T1, an atlas node might contain a set of parameter values ​​such as heart rate 75 beats per minute, systolic blood pressure 120 mmHg, diastolic blood pressure 80 mmHg, and body temperature 36.5 degrees Celsius. Directed edges are defined in the atlas, representing state transition paths between atlas nodes at different times, indicating the direction of physiological state change from one time point to the next. For example, a directed edge from an atlas node at time point T1 to an atlas node at time point T2 represents the evolution of the patient's vital sign state from T1 to T2. Transition attributes are labeled for each directed edge in the atlas. These attributes include at least the transition time and the primary driving vital sign parameter. The transition time is calculated from the difference between the corresponding timestamps of the two atlas nodes, and the primary driving vital sign parameter is determined by comparing the magnitude of changes in various vital sign parameters between the two atlas nodes. Initialize a blank atlas structure containing a start node as the framework for the dynamic vital signs atlas. The start node corresponds to the state of vital signs parameters at the start of monitoring, such as the first data point at the start of postoperative monitoring.

[0073] The validated time-series vital sign stream is segmented into snapshots at fixed time intervals, with each interval set to five minutes. This means that every five minutes, the values ​​of all vital sign parameters are extracted from the validated time-series vital sign stream to form a snapshot. Each vital sign data snapshot is mapped to a map node, and a vital sign vector is calculated for each map node. The vital sign vector is a numerical representation of the vital sign parameter values. In practice, the vital sign vector is constructed by normalizing each parameter value to a uniform scale and then concatenating them. For example, for three parameters including heart rate, systolic blood pressure, and diastolic blood pressure, the vital sign vector is calculated as follows:

[0074] ;

[0075] in: Represents the symptom vector. This represents the raw heart rate value. This represents the historical average heart rate over the entire monitoring period. The standard deviation of heart rate, This represents the original value of systolic blood pressure. This represents the historical average systolic blood pressure. This represents the standard deviation of systolic blood pressure. This represents the raw diastolic blood pressure value. This represents the historical average diastolic blood pressure. This represents the standard deviation of diastolic blood pressure. It can be understood that normalization eliminates the influence of different dimensions of vital sign parameters, making subsequent state transition analysis more balanced. In some embodiments, the fixed time interval can be adjusted to one minute or ten minutes according to clinical needs to accommodate monitoring requirements of different precision. Based on the sequential order of continuous vital sign data snapshots in the validated time-series vital sign stream, directed graph edges are established between the corresponding graph nodes. For example, if the vital sign data snapshot at time point T1 is mapped to graph node N1, and the vital sign data snapshot at time point T2 is mapped to graph node N2, then a directed graph edge from N1 to N2 is established between graph nodes N1 and N2.

[0076] The transfer attributes of directed edges in the dynamic vital sign graph are calculated and updated using actual data from the validated temporal vital sign stream, completing the data injection into the dynamic vital sign graph. For transfer time, the difference between the timestamps of the two graph nodes is directly calculated. For example, if graph node N1 corresponds to time 10:00 and graph node N2 corresponds to time 10:05, the transfer time is 5 minutes. For the primary driving vital sign parameter, the absolute value of the change in the normalized value of each vital sign parameter between the two graph nodes is calculated, and the parameter with the largest change is selected as the primary driving vital sign parameter. For example, if the vital sign vector of graph node N1 is [0.1, -0.2, 0.05] and the vital sign vector of graph node N2 is [0.3, -0.1, 0.04], then the change in heart rate is 0.2, the change in systolic blood pressure is 0.1, and the change in diastolic blood pressure is 0.01. Therefore, the primary driving vital sign parameter is determined to be heart rate. In some embodiments, the calculation of key driving signs parameters can employ weighted changes, assigning clinical importance weights to different parameters, but the basic logic remains based on comparing the magnitude of change. Optionally, the transition attributes of directed edges in the atlas can also include state transition intensity, defined as the Euclidean distance between the sign vectors of two atlas nodes, used to quantify the degree of state change. It can be understood that by continuously injecting atlas nodes and directed edges, the dynamic sign atlas gradually expands, forming a network graph reflecting the evolution of the patient's physiological state over time. The dynamic sign atlas after data injection contains a series of atlas nodes connected in chronological order, with each directed edge carrying specific transition time and labels for key driving signs parameters.

[0077] In one embodiment of the present invention, the preset criteria for dividing postoperative rehabilitation stages are defined based on the stability thresholds and fluctuation patterns of key vital signs, such as the range of variation of parameters like heart rate and blood pressure. Along the time axis of the dynamic vital sign atlas, the vital sign vectors of each atlas node are scanned. When the vital sign vector characteristics of multiple consecutive atlas nodes meet the definition of a new rehabilitation stage, stage boundary markers are created in the dynamic vital sign atlas. Based on all stage boundary markers, the dynamic vital sign atlas is divided into multiple consecutive stage sub-atlases, collectively forming a staged vital sign atlas. The statistical distribution characteristics of the vital sign vectors of all atlas nodes within each stage sub-atlas are analyzed. These statistical distribution characteristics include the mean, variance, and covariance relationship. Based on the statistical distribution characteristics, an anomaly detection rule set and sensitivity parameters are configured for the stage sub-atlas. The anomaly detection rule set and sensitivity parameters are encapsulated to form an independent anomaly probe specifically for scanning that stage sub-atlas. Each independent anomaly probe is loaded into the storage space of its corresponding stage sub-atlas.

[0078] In practice, the pre-defined criteria for dividing postoperative rehabilitation stages are based on the stability thresholds and fluctuation patterns of key vital signs. For example, the first postoperative stage is defined as the "anesthesia recovery period," characterized by heart rate fluctuations greater than 20 beats per minute and blood pressure fluctuations greater than 30 mmHg. The second stage is defined as the "initial stabilization period," characterized by heart rate fluctuations between 10 and 20 beats per minute and blood pressure fluctuations between 15 and 30 mmHg. Along the time axis of the dynamic vital signs map, the vital signs vectors of each map node are scanned. When the vital signs vector characteristics of multiple consecutive map nodes meet the definition of a new rehabilitation stage, a stage boundary marker is created in the dynamic vital signs map. For example, if the heart rate fluctuations of five consecutive map nodes are all less than 10 beats per minute and the blood pressure fluctuations are less than 15 mmHg, then the patient is determined to have entered the "stable recovery period," and a stage boundary marker is created at the fifth map node.

[0079] Based on all stage boundary markers, the dynamic vital sign atlas is segmented into multiple continuous stage sub-atlases, collectively forming a staged vital sign atlas. The segmentation operation is performed at the atlas nodes corresponding to the stage boundary markers, ensuring that each stage sub-atlas contains continuous atlas nodes belonging to the same rehabilitation stage and the directed edges between them. The statistical distribution characteristics of the vital sign vectors of all atlas nodes within each stage sub-atlas are analyzed. The calculation of the statistical distribution characteristics includes the mean, variance, and covariance between different vital sign parameters. In specific implementation, for the "initial stabilization period" stage sub-atlas, the mean of the heart rate vector components of all atlas nodes within it is calculated. With variance The mean of the systolic pressure vector components With variance and the covariance between the vector components of heart rate and systolic blood pressure. The calculation method is as follows:

[0080] ;

[0081] in: This represents the covariance between heart rate and systolic blood pressure. This represents the total number of graph nodes within the stage subgraph. Indicates the first Heart rate vector component values ​​of each graph node This represents the mean of the heart rate vector components. Indicates the first The contraction pressure vector component values ​​of each spectral node. This represents the mean value of the systolic pressure vector components.

[0082] Based on statistical distribution characteristics, an anomaly detection rule set and sensitivity parameters are configured for each stage sub-map. These rules and parameters are then encapsulated to form an independent anomaly probe specifically designed for scanning the stage sub-map. In some embodiments, the statistical distribution characteristics are used to display heart rate variance. The smaller "stable recovery period" sub-map may have an anomaly detection rule set that includes "heart rate exceeding the mean". The rule states that "an anomalies are triggered within a range of plus or minus two standard deviations," and the sensitivity parameter is set to a threshold multiple of the standard deviation in this rule. Optionally, the sensitivity parameter can be fine-tuned based on the experience of medical staff; for example, the heart rate anomaly threshold for the "initial stabilization period" can be adjusted from two standard deviations to 1.5 standard deviations to improve detection sensitivity. Each independent anomaly probe is loaded into the storage space of its corresponding stage sub-map, thus binding the probe to the partitioned data. Each independent anomaly probe can only access and scan the map nodes and directed edges within the specific stage sub-map it is loaded into. This design ensures that an independent anomaly probe targeting the "anesthesia recovery period" will not mistakenly use the rules for the "stable recovery period" to scan the data. This ensures the stage-specificity of anomaly detection. In some embodiments, the storage space for the stage sub-map can be an independent data structure in memory or a partitioned table in a database, where independent anomaly probes are deployed as a piece of program code or a computational process. It can be understood that by customizing and loading independent anomaly probes for each stage sub-map, the system can adapt to the physiological characteristic changes at different stages of patient recovery, achieving dynamic adaptation of the monitoring strategy. Optionally, when the dynamic vital sign map expands with the injection of new data, generating new stage boundary markers and forming new stage sub-maps, the system will analyze the statistical distribution characteristics of the new stage sub-maps in real time and automatically generate and load corresponding new independent anomaly probes.

[0083] In one embodiment of the present invention, each independent anomaly probe traverses each map node in its corresponding stage sub-map in chronological order. The feature vector of the currently traversed map node is matched against an anomaly determination rule set within the independent anomaly probe, where the rule set defines the range or pattern of normal values. When the matching result exceeds the sensitivity parameter of the independent anomaly probe, a potential anomaly signal containing the anomaly type, node location, and degree of deviation is generated. Each independent anomaly probe outputs its generated potential anomaly signal to a unified signal buffer in real time.

[0084] In practical implementation, each independent anomaly probe runs synchronously and performs parallel scanning of the staged vital sign map to collect potential abnormal signals. Within its corresponding stage sub-map, each independent anomaly probe traverses each map node in chronological order. For example, an independent anomaly probe deployed in the "stable recovery period" stage sub-map starts from the map node with the earliest timestamp in that partition and visits each subsequent map node sequentially. The vital sign vector of the currently traversed map node is matched against the anomaly judgment rule set within the independent anomaly probe. The matching calculation process compares each component of the vital sign vector with the normal range or expected pattern of the corresponding parameter defined in the anomaly judgment rule set. In some embodiments, the anomaly judgment rule set may include rules such as "heart rate vector component value exceeds 0.5" or "systolic blood pressure vector component value is below -1.2." The independent anomaly probe calculates the degree of conformity between the current vital sign vector and these rules. It can be understood that the matching calculation can be quantified as a distance or deviation score, for example, calculating the absolute difference between the current heart rate vector component value of 0.8 and the rule threshold of 0.5.

[0085] When the matching result exceeds the sensitivity parameter of the independent anomaly probe, a potential anomaly signal is generated, containing the anomaly type, node location, and deviation degree. The anomaly type is identified by the specific rule that triggered it, and the node location is determined by the stage sub-map identifier and the sequence number of the map node within the partition. The deviation degree is the quantized value of the matching result exceeding the sensitivity parameter threshold. In some embodiments, the sensitivity parameter is set such that the deviation score must be greater than 1.0 to trigger signal generation; if the calculated deviation score is 1.5, the condition is met and a signal is generated. The generation process of the potential anomaly signal follows a unified data structure; see Table 1 for an example of the signal buffer contents.

[0086] Table 1: List of Potential Abnormal Signals in the Signal Buffer

[0087] Each independent anomaly probe outputs its generated potential anomaly signals in real time to a unified signal buffer, which is a shared memory area where all independent anomaly probes have permission to write data. In practice, the degree of deviation... The calculation can be performed using the following formula:

[0088] ;

[0089] in: Indicates the degree of deviation. This represents the component value of a monitored parameter in the feature vector of the currently traversed graph node. This represents the threshold value set for this parameter in the set of anomaly detection rules. This represents the baseline value of the sensitivity parameter for independent anomaly probes. This formula makes it possible for... The difference is equal to the baseline value of the sensitivity parameter. At that time, the degree of deviation A value of 1 indicates that the threshold for triggering the signal has just been reached. This can be understood as the degree of deviation. The magnitude of the deviation directly reflects the relative severity of the vital sign parameter's deviation from the normal threshold. Optionally, for matching calculations involving multiple parameter composite rules, the degree of deviation... This could be the square root of the weighted sum of squares of the deviation components of each parameter. Each independent anomaly probe appends a timestamp accurate to milliseconds when writing potential anomaly signals into a unified signal buffer. This timestamp marks the time the signal was generated, rather than the physiological time corresponding to the map node. This helps in subsequent analysis of the generation order and density of anomaly signals. The signal buffer is designed as a first-in, first-out queue or a time-indexed buffer to ensure that subsequent anomaly signal aggregation networks can read all potential anomaly signals sequentially or efficiently.

[0090] See Figure 4 This is a bar chart showing the number of abnormal signals at different stages of urological surgery recovery. The functional recovery period has the highest number of abnormal signals, while the discharge observation period has the lowest. The number of abnormal signals varies significantly across stages, reflecting different levels of fluctuation in vital signs at different recovery stages. This type of chart is commonly used in medical monitoring scenarios to help healthcare professionals quickly identify high-risk stages after surgery (such as the functional recovery period) and adjust monitoring frequency and intervention strategies. By comparing numerical values, the "risk level of the recovery stage" is transformed from a qualitative description into quantitative data, facilitating standardized archiving of medical data.

[0091] In one embodiment of the invention, an abnormal signal aggregation network receives all potential abnormal signals from a signal buffer. Based on the clinical urgency of the stage sub-map from which the potential abnormal signal originates, a base weight is assigned; the clinical urgency is predefined by the recovery stage. The temporal clustering and correlation of different potential abnormal signals with corresponding vital signs are analyzed. Strongly correlated potential abnormal signals are clustered and fused to generate composite abnormal signals. Based on the total deviation between the base weight and the composite abnormal signal, a final priority score is calculated for each abnormal signal, and they are ranked accordingly. Different instruction content templates and target response terminals are preset for different levels of warning instructions. The priority-ranked list of abnormal signals output by the abnormal signal aggregation network is mapped to the corresponding warning level. Based on the mapped warning level, the corresponding instruction content template is selected, and the specific abnormal signal details are filled in to form a complete hierarchical warning instruction.

[0092] In practice, the abnormal signal aggregation network receives all potential abnormal signals from the signal buffer, which serves as a centralized storage area for the output results of independent abnormal probes and is continuously filled with new potential abnormal signals. Based on the clinical urgency of the sub-map from which the potential abnormal signal originates, a base weight is assigned. The clinical urgency is predefined by medical knowledge during system initialization. For example, the "anesthesia recovery period" sub-map has a high clinical urgency and its base weight is set to 1.5, while the "stable recovery period" sub-map has a relatively low clinical urgency and its base weight is set to 1.0.

[0093] This study analyzes the temporal clustering and correlation of different potential abnormal signals with corresponding vital signs. Strongly correlated potential abnormal signals are clustered and fused to generate composite abnormal signals. In practice, temporal clustering refers to the timestamps of multiple potential abnormal signals falling within a preset time window, such as five minutes. Correlation with vital signs refers to the physiological coupling relationship between the vital signs involved in these signals, such as heart rate and blood pressure. Cluster fusion can be understood as merging multiple independent potential abnormal signals into a single composite abnormal signal. This composite abnormal signal includes the set of abnormality types, overall deviation, and temporal range of all original signals. In some embodiments, if three potential abnormal signals are received consecutively from the "initial stabilization period" sub-map within five minutes, with abnormality types of "regular high heart rate," "regular low blood pressure," and "regular low blood oxygen," and these three vital signs are pathologically correlated, the abnormal signal aggregation network fuses them into a single composite abnormal signal, labeled as "composite circulatory and respiratory abnormality."

[0094] Based on the base weights and the total deviation of the composite anomalous signals, a final priority score is calculated for each anomalous signal, and they are then ranked accordingly. The total deviation refers to the deviation of a single potential anomalous signal or the weighted sum of the deviations of all original signals in a composite anomalous signal. For example, calculating the final priority score... The formula is:

[0095] ;

[0096] in: Indicates the final priority score. Indicates the basic weight. The aggregation coefficient represents the degree of deviation. This represents the sum of all original deviation values ​​included in this abnormal signal. It is understandable that... It is an adjustable coefficient used to control the contribution ratio of the total deviation degree to the final score. Different instruction content templates and target response terminals are preset for different levels of warning instructions. The warning levels can be divided into three levels: "low", "medium" and "high". Each level corresponds to a different instruction content template. For example, the high-level warning instruction template includes a red warning sign and an emergency contact number list, and the target response terminals are the doctor's mobile terminal and the nurse station main control panel; the low-level warning instruction template only includes a yellow warning sign, and the target response terminal is only the nurse station secondary monitoring screen.

[0097] The list of priority-ranked anomalous signals output by the anomalous signal aggregation network is mapped to corresponding warning levels, with the mapping based on the final priority score. The numerical range is used, for example Mapped to "low" level, Mapped to "medium" level. Mapped to a "high" level. In some embodiments, the list of anomalous signals is a list ranked by final priority. The sequence is arranged in descending order. Based on the mapped warning level, the corresponding instruction content template is selected, and the specific abnormal signal details are filled in to form a complete graded warning instruction. The abnormal signal details must include at least the abnormality type, the stage of occurrence, the specific deviation value, and time information. Optionally, when filling in the specific abnormal signal details into the instruction content template, a standardized field filling method is used. For example, "Complex Circulatory Respiratory Abnormality" is entered in the "{Abnormality Type}" position of the template, and "2026-01-14 11:05 to 11:10" is entered in the "{Occurrence Time}" position. It can be understood that the formation of graded warning instructions is a process of transforming data into clinically actionable information. The final generated graded warning instructions are transmitted to the scheduling system, which then executes the notification or initiates the intervention process according to the warning response protocol.

[0098] See Figure 5 This is a heatmap showing the recovery stages and physiological characteristics after urological surgery. The "Clinical Urgency" value is highest (darkest color) during the anesthesia recovery period, indicating this is the highest-risk period post-surgery. The "Number of Abnormal Signals" value is relatively high during the stable recovery period, consistent with the previous bar chart conclusions. The values ​​of various physiological characteristics are generally lower during the discharge preparation period, reflecting that the patient's vital signs are approaching stability. Visualizing multi-dimensional correlations and presenting the correspondence between "stages" and "physiological characteristics" facilitates the rapid identification of core monitoring indicators at each stage. The color intensity visually distinguishes high / low-risk stages and characteristics, assisting medical staff in developing differentiated monitoring plans. The color intensity quantifies the risk level, accurately identifying high-risk stages and characteristics, and assisting in the early deployment of intervention measures.

[0099] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for processing postoperative physical signs monitoring data after urological surgery, characterized in that, The method comprises: Obtaining original postoperative urological sign time series streams from multiple monitoring devices; Verifying the data integrity and consistency of the original postoperative urological sign time series streams, and outputting the verified time series sign streams; Creating a dynamic sign atlas for describing the evolution of postoperative urological physiological states; Based on the verified time series sign streams, injecting sign nodes and state transition edges into the dynamic sign atlas; According to the preset postoperative rehabilitation stage division standard, partitioning the dynamic sign atlas into stages to generate a staged sign atlas; In each partition of the staged sign atlas, deploying an independent anomaly probe matched with the partition physiological stage characteristics; Synchronously running each independent anomaly probe to perform parallel scanning on the staged sign atlas and collect potential abnormal signals in each partition; Building an abnormal signal aggregation network and inputting the potential abnormal signals collected by each independent anomaly probe into the abnormal signal aggregation network for fusion and priority sorting; According to the output of the abnormal signal aggregation network, generating a differentiated hierarchical early warning instruction; According to the type and level of the hierarchical early warning instruction, calling and executing the corresponding early warning response protocol.

2. The method of claim 1, wherein the method further comprises: The data integrity and consistency verification of the original postoperative urological sign time series streams comprises: Identifying signal missing segments and signal conflict segments in the original postoperative urological sign time series streams; Using the evolution trend of similar sign data in adjacent time periods to interpolate and reconstruct the signal missing segments to form continuous sign data; According to the preset physiological parameter mutual exclusion rule library, logically arbitrating the signal conflict segments to retain sign data consistent with physiological logic; Reintegrating the data after interpolation and reconstruction and logical arbitration to form a verified time series sign stream that is continuous in time dimension and self-consistent in logic dimension.

3. The method of claim 1, wherein the method further comprises: The creation of a dynamic sign atlas for describing the evolution of postoperative urological physiological states comprises: Defining atlas nodes, which represent the combined state of multiple sign parameters at a specific time; Defining atlas directed edges, which represent the state transition path between the atlas nodes at different times; Labeling transition attributes for each atlas directed edge, which at least include transition time consumption and main driving sign parameters; Initializing a blank atlas structure containing a starting node as the framework of the dynamic sign atlas.

4. The method of urinary post-operative sign monitoring data processing of claim 3, wherein, The injection of sign nodes and state transition edges into the dynamic sign atlas based on the verified time series sign streams comprises: Dividing sign data snapshots at fixed time intervals from the verified time series sign streams; Mapping each sign data snapshot to an atlas node and calculating the sign vector of the atlas node; According to the order of the continuous sign data snapshots in the verified time series sign streams, establishing the atlas directed edges between the corresponding atlas nodes; Using the actual data in the verified time series sign streams to calculate and update the transition attributes of the atlas directed edges, and completing the data injection of the dynamic sign atlas.

5. The method of urinary post-operative sign monitoring data processing of claim 1, wherein, The stage division standard comprises: The preset postoperative rehabilitation stage division standard is defined based on a stable threshold and a fluctuation mode of a key sign index, and each sign vector of each graph node is scanned along a time axis direction of the dynamic sign graph. When the sign vector features of a plurality of continuous graph nodes meet the definition of a new rehabilitation stage, a stage boundary marker is created in the dynamic sign graph. According to all stage boundary markers, the dynamic sign graph is cut into a plurality of continuous stage sub-graphs, which together constitute the stage division sign graph.

6. A method of processing post-operative urinary signs monitoring data according to claim 5, characterized in that, The independent anomaly probe matched with the physiological stage features of each division comprises: The statistical distribution features of the sign vectors of all graph nodes in each stage sub-graph are analyzed. According to the statistical distribution features, a specific anomaly judgment rule set and a sensitivity parameter are configured for the stage sub-graph, the anomaly judgment rule set and the sensitivity parameter are encapsulated to form an independent anomaly probe specially used for scanning the stage sub-graph, and each independent anomaly probe is loaded into the storage space of the corresponding stage sub-graph. The independent anomaly probe matched with the physiological stage features of each division comprises:

7. The method of urinary post-operative sign monitoring data processing of claim 1, wherein, Each independent anomaly probe traverses each graph node in the corresponding stage sub-graph in time sequence. The sign vector of the currently traversed graph node is matched and calculated with the anomaly judgment rule set inside the independent anomaly probe. When the matching calculation result exceeds the sensitivity parameter of the independent anomaly probe, a potential abnormal signal containing the abnormal type, node position and deviation degree is generated. Each independent anomaly probe outputs the potential abnormal signal generated by it to a unified signal buffer area in real time. The anomaly signal aggregation network comprises:

8. A method of processing post-operative urinary signs monitoring data according to claim 7, characterized in that, The anomaly signal aggregation network receives all potential abnormal signals from the signal buffer area. According to the clinical emergency degree of the stage sub-graph from which the potential abnormal signal is derived, a basic weight is given to it. The clustering and correlation of different potential abnormal signals in time are analyzed, the potential abnormal signals with strong correlation are cluster fused to generate composite abnormal signals, and the final priority score of each abnormal signal is calculated based on the basic weight and the total amount of deviation degree of the composite abnormal signal, and is sorted accordingly. According to the output of the anomaly signal aggregation network, the differential hierarchical early warning instruction is generated. Different instruction content templates and target response terminals are preset for different levels of early warning instructions.

9. The method of urinary post-operative sign monitoring data processing of claim 1, wherein, The anomaly signal list with priority ranking output by the anomaly signal aggregation network is mapped to the corresponding early warning level. According to the mapped early warning level, the corresponding instruction content template is selected, and the specific abnormal signal details are filled in to form a complete hierarchical early warning instruction. ​ ​ 10. A post-urological procedure sign monitoring data processing system comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, The processor, when executing the computer program, implements the steps of the method for processing postoperative sign monitoring data of the urinary system according to any one of claims 1 to 9. The processor, when executing the computer program, implements the steps of the method for processing postoperative sign monitoring data of the urinary system according to any one of claims 1 to 9.

Images