Neurosurgical nursing postoperative management system and method

Abstract

The invention relates to the technical field of neurosurgical nursing postoperative management, and discloses a neurosurgical nursing postoperative management system and method, and the method comprises the steps: obtaining multi-source heterogeneous information, carrying out the unified evidence conversion processing of the multi-source heterogeneous information, and generating structured evidence data; identifying and associating a plurality of pieces of structured evidence data which are overlapped or adjacent in time in a preset sliding time window; initial weights of the multiple pieces of structured evidence data associated in a preset sliding time window are calculated, a comprehensive risk score value is generated, and the dynamic confidence degree of the comprehensive risk score value is synchronously calculated; and when the comprehensive risk score value reaches a preset risk level threshold value and the dynamic confidence coefficient is not lower than a lower limit of a corresponding preset risk level confidence coefficient, generating a risk early warning. According to the invention, through effective integration and risk assessment of multi-source heterogeneous information, the limitation of the prior art in the aspects of information islands and early warning is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of postoperative management technology in neurosurgical nursing, and more specifically, to a postoperative management system and method for neurosurgical nursing. Background Technology

[0002] In postoperative care of neurosurgery, close monitoring of patients' vital signs is crucial to ensuring safe recovery. While modern postoperative management systems can continuously collect patients' physiological data and analyze it according to pre-set rules, they often encounter challenges when processing information from various sources and of different natures. Especially in complex situations where changes in a patient's condition are subtle and scattered, and multiple pieces of information need to be correlated to detect early problems, existing systems often struggle to provide timely and effective warnings. This not only increases the workload of medical staff but may also lead to missed opportunities for optimal intervention.

[0003] Existing management systems lack the ability to integrate and analyze these different forms of information. The system treats continuous physiological data, intermittent quantitative scores, and unstructured text records as three separate information silos. It can display each type of information well, but it cannot understand the inherent connections between them. Therefore, the system cannot automatically identify dangerous patterns composed of multiple "subthreshold" changes. This lack of information integration shifts the entire burden of identifying early complications onto healthcare workers, making it easy for the window of opportunity for early intervention to be missed due to human negligence or fatigue. By the time the condition finally progresses to a point sufficient to trigger a clear alarm on a single indicator, it often means that the optimal treatment window has passed, and irreversible neurological damage may have already occurred. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention discloses a neurosurgical postoperative care management system and method. The aim is to solve the problem that existing neurosurgical postoperative care management systems struggle to effectively integrate and analyze multi-source heterogeneous information, leading to the inability to detect early changes in the condition in a timely manner and potentially missing the optimal intervention opportunity.

[0005] The technical solution of the present invention is as follows: In a first aspect, the present invention discloses a method for postoperative management of neurosurgical nursing, comprising the following steps: Acquire multi-source heterogeneous information about patients during postoperative monitoring, perform unified evidence transformation processing on the multi-source heterogeneous information, and generate structured evidence data; Within a preset sliding time window, identify and associate multiple structured evidence data that overlap or are adjacent in time; The initial weights of multiple structured evidence data associated within a preset sliding time window are calculated to generate a comprehensive risk score, and the dynamic confidence level of the comprehensive risk score is calculated simultaneously. A risk warning is generated when the comprehensive risk score reaches the preset risk level threshold and the dynamic confidence level is not lower than the corresponding preset risk level confidence level lower limit.

[0006] Through this technical solution, the present invention can transform multi-source heterogeneous information into unified structured evidence data, and perform correlation analysis through a sliding time window. Combined with comprehensive risk scores and dynamic confidence, it can achieve early and accurate warning of postoperative risks for patients, effectively solving the problem of insufficient information integration capabilities and difficulty in identifying early complications in existing systems.

[0007] Furthermore, multi-source heterogeneous information includes physiological parameters, neurological assessment results, nursing records, and examination results.

[0008] Based on this, the present invention further proposes a step for generating structured evidence data by performing unified evidence transformation processing on multi-source heterogeneous information, including: Physiological parameters are continuously collected, and the changing trend characteristics of physiological parameters within a preset monitoring time window are calculated. Based on the characteristics of change trends, structured evidence data of physiological trends are generated to characterize the changing trends of physiological parameters.

[0009] Based on the above, the present invention also proposes that the neurological assessment results include Glasgow Coma Scale score, pupil size and light reflex, and limb movement ability. The steps for performing unified evidence transformation on multi-source heterogeneous information to generate structured evidence data include: The Glasgow Coma Scale, pupil size and light reflex, and limb movement ability manually entered by medical staff were transformed into structured evidence data for neurological assessment. Semantic analysis of nursing records is performed to obtain original behavioral descriptions and corresponding time points; Physiological parameters within a preset time period before and after a given time point are acquired as real-time physiological background features; Track relevant clinical interventions performed by medical staff within a pre-defined time period before and after the specified time point; The original behavioral descriptions, real-time physiological background features, and clinical intervention actions are matched with a pre-set clinical scenario pattern library. Based on the matched clinical scenario patterns, structured evidence data of nursing events is generated.

[0010] Specifically, within a preset sliding time window, the steps of identifying and associating multiple structured evidence data that overlap or are adjacent in time include: Key numerical values ​​and descriptive information are extracted from the imaging data in the examination results and transformed into structured evidence data to support the examination.

[0011] Furthermore, within a pre-defined sliding time window, the steps of identifying and associating multiple structured evidence data that overlap or are adjacent in time include: Based on preset association rules, multiple structured evidence data appearing within a preset sliding time window are matched and associated using a preset event processing engine.

[0012] To enhance functionality, the steps for calculating initial weights for multiple structured evidence data associated within a preset sliding time window and generating a comprehensive risk score include: The initial weight of each structured evidence data is multiplied by its corresponding preset strength factor and then summed to generate a comprehensive risk score.

[0013] To improve the plan, the steps for simultaneously calculating the dynamic confidence level of the comprehensive risk score include: Based on the quantity of structured evidence data, the average strength of structured evidence data, the consistency score of structured evidence data, the quality score of structured evidence data, and the time decay factor of structured evidence data, the dynamic confidence level is calculated according to a preset confidence level comprehensive calculation function.

[0014] To further address the issue, when the comprehensive risk score reaches a preset risk level threshold and the dynamic confidence level is not lower than the corresponding preset lower limit of risk level confidence, the steps for generating a risk warning include: Generate risk warnings that include a visual view of the chain of evidence, which displays detailed information about all structured evidence data in the form of a timeline or list. Simultaneously, based on the comprehensive risk score and multiple structured evidence data, and based on preset clinical decision-making rules, decision support suggestions are generated and pushed to medical staff.

[0015] Secondly, the present invention also discloses a postoperative neurosurgical nursing management system, the system comprising: The structured evidence data generation module is used to acquire multi-source heterogeneous information of patients during postoperative monitoring, perform unified evidence transformation processing on the multi-source heterogeneous information, and generate structured evidence data. The association identification module is used to identify and associate multiple structured evidence data that overlap or are adjacent in time within a preset sliding time window; The risk assessment module is used to calculate the initial weights of multiple structured evidence data associated within a preset sliding time window, generate a comprehensive risk score, and simultaneously calculate the dynamic confidence level of the comprehensive risk score. The early warning generation module is used to generate a risk warning when the comprehensive risk score reaches the preset risk level threshold and the dynamic confidence level is not lower than the corresponding preset risk level confidence level lower limit.

[0016] Through this technical solution, the present invention provides a system for implementing the above-mentioned method. Through modular design, it ensures the coordinated work of each functional unit, provides reliable hardware and software support for postoperative management of neurosurgery, and realizes the effective implementation of the method.

[0017] In summary, this invention provides a postoperative management system and method for neurosurgical nursing. The method acquires multi-source heterogeneous information from patients during postoperative monitoring and performs unified evidence transformation processing to generate structured evidence data, effectively solving the problem that existing systems cannot process and integrate information from different sources and of different natures. Furthermore, within a preset sliding time window, this method identifies and associates multiple structured evidence data that overlap or are adjacent in time, overcoming the deficiency of existing systems that treat information as isolated islands and fail to discover their inherent connections. By calculating the initial weights of the associated evidence data, a comprehensive risk score is generated, and dynamic confidence is calculated simultaneously. This invention can comprehensively, quantitatively, and reliably assess the patient's risk status, avoiding the limitations of single-indicator judgment. Finally, when the comprehensive risk score reaches a preset risk level threshold and the dynamic confidence meets the requirements, the system can generate a timely risk warning, thus effectively solving the problems of difficulty in detecting early changes in the condition and the easy loss of optimal intervention opportunities in existing technologies. This method can combine scattered, subthreshold information to identify potential danger patterns, significantly improving the early warning capability of postoperative complications in neurosurgery, reducing the workload of medical staff, and buying valuable treatment time for patients, thus avoiding irreversible neurological damage. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a postoperative management method for neurosurgical nursing, as provided in an embodiment of the present invention.

[0019] Figure 2 This is a schematic diagram of the structure of a postoperative neurosurgical nursing management system provided in an embodiment of the present invention.

[0020] Labeling Explanation: 210, Structured Evidence Data Generation Module; 220, Association Identification Module; 230, Risk Assessment Module; 240, Early Warning Generation Module. Detailed Implementation

[0021] The technical solutions of this invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some, not all, of the embodiments of this invention. The components of this invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0022] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0023] In postoperative neurosurgical care, close monitoring of patients' vital signs is crucial for ensuring safe recovery. However, existing postoperative management systems often face challenges in processing information from diverse sources and of varying natures. Particularly in complex situations where changes in a patient's condition are subtle and scattered, requiring the correlation of multiple pieces of information to detect early problems, existing systems often struggle to provide timely and effective warnings. This not only increases the workload of medical staff but may also lead to missed opportunities for optimal intervention. Existing systems primarily rely on fixed thresholds for single parameters, lacking the ability to integrate and analyze different forms of information, and are unable to automatically identify dangerous patterns composed of combinations of multiple "subthreshold" changes. This lack of information integration capability makes the window of opportunity for early intervention easily missed due to human negligence or fatigue.

[0024] Firstly, please see Figure 1 This invention proposes a postoperative management method for neurosurgical nursing, comprising: S1. Obtain multi-source heterogeneous information about the patient during postoperative monitoring, perform unified evidence transformation processing on the multi-source heterogeneous information, and generate structured evidence data. S2. Within a preset sliding time window, identify and associate multiple structured evidence data that overlap or are adjacent in time. S3. Calculate the initial weights of multiple structured evidence data associated within a preset sliding time window, generate a comprehensive risk score, and simultaneously calculate the dynamic confidence level of the comprehensive risk score. S4. When the comprehensive risk score reaches the preset risk level threshold and the dynamic confidence level is not lower than the corresponding preset risk level confidence level lower limit, a risk warning is generated.

[0025] "Multi-source heterogeneous information" refers to various types of data acquired from different sources and in different formats during postoperative patient monitoring. "Heterogeneous" is reflected in the fact that this data may exist in multiple forms such as numerical values, text, and images, and the collection frequency and recording methods vary. "Unified evidence transformation processing" refers to the process of transforming the aforementioned multi-source heterogeneous information into unified, standardized "structured evidence data" that can be calculated and analyzed by the system. "Structured evidence data" is an information unit after unified evidence transformation processing; each piece of evidence data contains clearly defined attributes such as type, value, timestamp, and source, facilitating subsequent association, weight calculation, and risk assessment. "Preset sliding time window" refers to a fixed-length time period that continuously moves along a timeline. "Comprehensive risk score" is a quantitative indicator calculated based on the initial weights of multiple structured evidence data associated within the sliding time window, used to comprehensively assess the patient's current risk level. "Dynamic confidence level" is an indicator calculated synchronously with the comprehensive risk score, used to measure the reliability and credibility of the risk score; it changes dynamically based on factors such as the quantity, strength, and consistency of evidence. "Risk level threshold" and "risk level confidence level lower limit" are preset values ​​used to determine when a risk warning is triggered.

[0026] The postoperative management method for neurosurgical nursing proposed in this invention works by constructing a dynamic, multi-dimensional risk assessment model to address the problems of information silos and insufficient early warning in existing technologies.

[0027] Specifically, this method first acquires multi-source heterogeneous information about the patient during postoperative monitoring, such as physiological parameters, neurological assessment results, nursing records, and examination results. This information comes from a wide range of sources and is diverse in form, making it difficult to effectively integrate using traditional methods. To overcome this challenge, this method performs a unified evidence transformation process on this multi-source heterogeneous information, converting it into standardized structured evidence data. This step is fundamental to all subsequent analyses; it unifies previously scattered and difficult-to-compare information, enabling the system to perform unified calculations and analyses.

[0028] Subsequently, within a pre-defined sliding time window, the system continuously identifies and correlates multiple structured pieces of evidence that overlap or are adjacent in time. This mechanism simulates the thought process of clinicians observing patients, connecting information from different time points and of different types to make a comprehensive judgment. By using the sliding time window, the system can capture the dynamic trends in the development of a patient's condition, rather than relying solely on static data from a single point in time. For example, a slight, continuous increase in a patient's heart rate, a slight sluggish pupillary light reflex, and restlessness mentioned in the nursing record—these seemingly independent events, when correlated within the time window, will have significantly enhanced clinical significance.

[0029] Next, the initial weights of multiple structured evidence data associated within a preset sliding time window are calculated to generate a comprehensive risk score, and the dynamic confidence level of the comprehensive risk score is calculated simultaneously. The comprehensive risk score is a quantitative assessment of the patient's overall risk level, taking into account the cumulative effect of multiple pieces of evidence. The introduction of dynamic confidence level further enhances the reliability of the risk assessment. The confidence level is dynamically adjusted based on factors such as the quantity, strength, and consistency of evidence, avoiding false alarms when evidence is insufficient or conflicting. For example, when multiple strong pieces of evidence appear simultaneously, even if the comprehensive risk score is high, if the confidence level is low, the system will not immediately issue a warning, but will instead suggest further observation or obtaining more evidence.

[0030] Ultimately, when the comprehensive risk score reaches the preset risk level threshold and the dynamic confidence level is not lower than the corresponding preset lower limit of risk level confidence, the system generates a risk warning. This dual judgment mechanism ensures the timeliness and accuracy of the warning. The system will only issue an alert to medical staff when the risk level is sufficiently high and the assessment results are sufficiently reliable, thus avoiding over-warning or under-warning. In this way, this method can effectively integrate scattered and subtle changes in the patient's condition, identify potential risks in the early stages, and provide timely and accurate decision support for medical staff, thereby significantly improving the efficiency and safety of postoperative neurosurgical management.

[0031] The postoperative management method for neurosurgical nursing proposed in this invention demonstrates significant progress and innovation in addressing the challenges of integrating multi-source heterogeneous information and early risk warning in postoperative neurosurgical management, compared to existing technologies.

[0032] Traditional postoperative management systems primarily rely on fixed thresholds for single parameters; for example, the system issues an alarm when a physiological indicator exceeds a preset warning line. This approach falls short when dealing with complex and multifactorial postoperative conditions in neurosurgery. Many serious complications often present with subtle and diverse early signs, scattered across different types of information, which existing systems cannot connect and comprehensively analyze. This leads to information silos, making it difficult for medical staff to detect potential problems early and thus missing the optimal intervention window.

[0033] The core innovation of this invention lies in its unique multi-source heterogeneous information integration and dynamic risk assessment mechanism. The postoperative management method for neurosurgical nursing proposed in this invention effectively overcomes the limitations of existing technologies in terms of information silos and early warning through innovative multi-source heterogeneous information integration, dynamic time window correlation, and a dual risk assessment mechanism. It can identify changes in the patient's condition earlier and more accurately, providing timely and reliable decision support for medical staff, thereby significantly improving the quality of postoperative management in neurosurgery and patient prognosis.

[0034] In some embodiments of the present invention, in order to more specifically illustrate the composition of multi-source heterogeneous information and thus ensure that the acquired data can comprehensively reflect the postoperative status of the patient, the present invention further clarifies the specific content of multi-source heterogeneous information.

[0035] Specifically, the above-mentioned postoperative management methods for neurosurgical nursing include multi-source heterogeneous information such as physiological parameters, neurological assessment results, nursing records, and examination results.

[0036] Physiological parameters refer to objective data related to the patient's vital signs, such as heart rate, blood pressure, respiratory rate, body temperature, and blood oxygen saturation. These parameters are usually automatically collected through continuous monitoring equipment. Neurological assessment results refer to clinical data obtained by medical staff after assessing the patient's neurological function, such as the Glasgow Coma Scale score, pupil size and light reflex, and limb movement ability. These assessment results reflect the patient's level of consciousness and nervous system function. Nursing records are written or structured records kept by medical staff during the nursing process regarding the patient's condition, nursing interventions, and patient responses. They contain rich clinical context and behavioral information. Examination results refer to data obtained from various medical examinations performed on the patient after surgery, such as blood tests and imaging examinations (e.g., CT, MRI). These results provide objective evidence of the patient's internal physiological state and pathological changes.

[0037] This invention addresses the issue by explicitly including multi-source heterogeneous information such as physiological parameters, neurological assessment results, nursing records, and examination results. This allows the system to acquire postoperative monitoring data from multiple dimensions and from a more comprehensive perspective. Physiological parameters provide real-time changes in vital signs, neurological assessment results reflect key states of neurological function, nursing records supplement information on medical staff's observations and interventions, and examination results provide in-depth diagnostic evidence. This integration of multi-source information helps construct a more complete and accurate patient profile, laying a solid data foundation for subsequent unified evidence transformation and risk assessment.

[0038] Specifically, in the steps described above for performing unified evidence transformation on multi-source heterogeneous information to generate structured evidence data, the processing of physiological parameters may include the following:

[0039] The steps for performing unified evidence transformation on multi-source heterogeneous information to generate structured evidence data include: Physiological parameters are continuously collected, and the changing trend characteristics of physiological parameters within a preset monitoring time window are calculated. Based on the characteristics of change trends, structured evidence data of physiological trends are generated to characterize the changing trends of physiological parameters.

[0040] Physiological parameters can be understood as the patient's vital signs data, such as heart rate, blood pressure, respiratory rate, body temperature, and blood oxygen saturation. Continuous data acquisition refers to acquiring these physiological parameter data in real time and without interruption through various medical sensors or monitoring devices connected to the patient, to ensure the integrity and timeliness of the data.

[0041] Furthermore, the preset monitoring time window refers to a configurable time period, such as 5 minutes, 10 minutes, or 30 minutes. Within this time window, continuously collected physiological parameter data are analyzed to calculate their trend characteristics. These trend characteristics may include, but are not limited to, statistical indicators such as average, maximum, minimum, standard deviation, rate of change, and slope, aiming to reflect the dynamic changes of physiological parameters within a specific time period. For example, the average rate of change of blood pressure over the past 10 minutes, or the fluctuation range of heart rate over the past 5 minutes, can be calculated to capture their upward, downward, or fluctuating trends.

[0042] Therefore, based on the calculated trend characteristics, structured evidence data representing the changing trends of physiological parameters can be generated. Structured evidence data of physiological trends is a standardized, computer-processable data format that transforms raw, continuous physiological parameter data into information with clear semantics and structure. For example, if blood pressure shows a continuous upward trend within a preset monitoring time window, a structured evidence data point for "continuously rising blood pressure" can be generated, along with specific numerical values ​​and timestamps, to facilitate subsequent risk assessment.

[0043] This invention captures subtle dynamic changes in a patient's physiological state by continuously collecting physiological parameters and calculating their trend characteristics within a preset monitoring time window. Traditional single-point physiological parameter values ​​often fail to fully reflect a patient's true condition, while analyzing trends allows for earlier and more accurate detection of potential abnormalities. For example, even if the instantaneous value of a physiological parameter remains within the normal range, a sustained upward or downward trend may indicate an impending risk. By transforming these trends into structured evidence data of physiological trends, this dynamic information can be effectively identified, correlated, and further evaluated by the system, thus providing richer and more forward-looking evidence for subsequent risk scoring and early warning.

[0044] In some embodiments of the present invention described above, a unified evidence transformation process is proposed to convert multi-source heterogeneous information into structured evidence data. Specifically, the unified evidence transformation process for the aforementioned neurological assessment results and nursing records can be performed in the following manner.

[0045] Neurological assessment results included Glasgow Coma Scale score, pupil size and light reflex, and limb movement ability; The steps for performing unified evidence transformation on multi-source heterogeneous information to generate structured evidence data include: The Glasgow Coma Scale, pupil size and light reflex, and limb movement ability manually entered by medical staff were transformed into structured evidence data for neurological assessment. Semantic analysis of nursing records is performed to obtain original behavioral descriptions and corresponding time points; Physiological parameters within a preset time period before and after a given time point are acquired as real-time physiological background features; Track relevant clinical interventions performed by medical staff within a pre-defined time period before and after the specified time point; The original behavioral descriptions, real-time physiological background features, and clinical intervention actions are matched with a pre-set clinical scenario pattern library. Based on the matched clinical scenario patterns, structured evidence data of nursing events is generated.

[0046] Specifically, neurological assessment results are key indicators reflecting the functional status of a patient's nervous system. The Glasgow Coma Scale (GCS) assesses the patient's level of consciousness, pupil size and pupillary light reflex assess brainstem function, and limb movement reflects the function of the motor cortex and conduction pathways. These assessment results are typically manually entered by healthcare professionals in clinical practice. To integrate them into a unified risk assessment framework, these manually entered Glasgow Coma Scale scores, pupil size and light reflex, and limb movement data are transformed into standardized, structured evidence data of neurological assessment. This transformation aims to eliminate the heterogeneity of the original data format, enabling it to be effectively processed by subsequent risk assessment models.

[0047] Semantic analysis of nursing records involves using natural language processing (NLP) technology to extract meaningful clinical information from unstructured nursing record text. Through semantic analysis, raw behavioral descriptions of patient conditions, interventions, and observations by healthcare professionals can be obtained, and the corresponding time points can be precisely identified. These raw behavioral descriptions are crucial clues for understanding the patient care process and potential risks.

[0048] Furthermore, to provide more comprehensive contextual information, the patient's physiological parameters are acquired within a preset time period before and after the specified time point and used as real-time physiological background features. These physiological parameters may include heart rate, blood pressure, respiratory rate, body temperature, etc., which can reflect the patient's physiological state at the time of a specific nursing event, providing objective evidence for understanding the background and impact of nursing behaviors.

[0049] In addition, to gain a comprehensive understanding of the clinical situation, the relevant clinical interventions performed by medical staff within a predetermined time period before and after the specified time point will also be tracked. These interventions may include medication administration, postural adjustments, wound care, etc., which, together with the behavioral descriptions and physiological background characteristics in the nursing records, constitute a complete chain of clinical events.

[0050] Therefore, the original behavioral descriptions, real-time physiological background characteristics, and clinical intervention actions are integrated and matched with a pre-defined clinical scenario pattern library. This library stores various known typical scenario patterns related to post-neurosurgery risks, such as "decreased level of consciousness with pupillary changes" and "dyspnea with postural adjustment." By matching the currently integrated clinical information with the pattern library, the specific clinical situation of the patient can be identified, and standardized structured evidence data of nursing events can be generated based on the matched clinical scenario patterns.

[0051] This invention transforms fragmented and heterogeneous clinical information into a unified, structured data format through refined evidence transformation of neurological assessment results and nursing records. Specifically, Glasgow Coma Scale scores, pupil size and light reflex, and limb movement ability are converted into structured evidence data for neurological assessment, enabling the quantification of these key neurological function indicators and their direct use in risk assessment. Simultaneously, semantic analysis of nursing records, combined with real-time physiological context features and clinical interventions, allows for the extraction of clinically significant nursing events from complex unstructured text. These events are then matched with pre-defined clinical scenario patterns to generate structured evidence data for nursing events with rich contextual information. This approach enables the system to more comprehensively and accurately understand the patient's clinical state and nursing process, providing high-quality input for subsequent risk identification and assessment.

[0052] Specifically, in some implementations of the above-mentioned postoperative management methods for neurosurgical nursing, the step of identifying and associating multiple structured evidence data that overlap or are adjacent in time within a preset sliding time window can be further refined.

[0053] Within a preset sliding time window, the steps for identifying and associating multiple structured evidence data that overlap or are adjacent in time include: Key numerical values ​​and descriptive information are extracted from the imaging data in the examination results and transformed into structured evidence data to support the examination.

[0054] Specifically, examination results can take various forms, such as medical imaging data like X-rays, CT scans, MRI images, and ultrasound images. Imaging data typically contains rich visual information that needs to be effectively processed and structured for subsequent risk assessment. Extracting key numerical and descriptive information from imaging data involves identifying and quantifying important indicators in the images, such as numerical data like lesion size, location, morphological characteristics, and density changes, as well as textual descriptions of these characteristics, such as "intracranial hemorrhage," "cerebral edema," and "mass lesion effect," through image processing technology, computer vision algorithms, or manual interpretation. This extracted information is the result of preliminary interpretation and quantification of the original imaging data. Further, transforming this extracted key numerical and descriptive information into structured evidence data for auxiliary examinations involves organizing and encoding unstructured imaging reports or image analysis results according to a preset data model and format, making them standardized data units that can be automatically identified, processed, and analyzed by the system. Structured evidence data used in auxiliary examinations can include fields such as image type, examination time, key findings, quantitative indicators, and diagnostic conclusions. Its purpose is to transform complex image information into a unified and standardized form of evidence, which facilitates integration and association with other types of structured evidence data.

[0055] The present invention structures the imaging data in examination results, transforming image information that is difficult to use directly for automated analysis into quantifiable and comparable structured evidence data for auxiliary examinations. This structured evidence data is then incorporated into a preset sliding time window and identified and correlated with other structured evidence data such as physiological parameters, neurological assessment results, and nursing records. As a result, the system can acquire more comprehensive clinical information about the patient, especially the pathophysiological changes reflected from an imaging perspective, thus providing richer and more accurate evidence support for subsequent comprehensive risk score calculations. In this way, imaging data is no longer merely a reference for auxiliary diagnosis, but becomes effective evidence that can be directly involved in the calculation of the risk assessment model.

[0056] Specifically, the steps described above for identifying and associating multiple structured evidence data that overlap or are adjacent in time within a preset sliding time window can be further implemented in the following ways.

[0057] Within a preset sliding time window, the steps of identifying and associating multiple structured evidence data that overlap or are adjacent in time include: matching and associating multiple structured evidence data that appear within the preset sliding time window using a preset event processing engine according to preset association rules.

[0058] The pre-defined association rules can be understood as a set of conditions that define the logical relationships between different types of structured evidence data. These rules can be set based on temporal relationships (e.g., event A occurs within a specific time frame of event B), semantic relationships (e.g., events A and B describe an abnormality in the same physiological system), causal relationships (e.g., a certain intervention leads to a change in physiological parameters), or clinical experience (e.g., a specific combination of symptoms predicts a certain complication). For example, an association rule could stipulate that when a patient's blood pressure is persistently elevated and their heart rate is simultaneously increased, these two physiological trend structured evidence data should be associated. Furthermore, the pre-defined event processing engine is a software module or system specifically designed for real-time analysis and processing of complex event streams. This engine is configured to receive and process structured evidence data from different sources and perform real-time matching and association of this data according to the pre-defined association rules. Specifically, the event processing engine can employ Complex Event Processing (CEP) technology, using mechanisms such as pattern recognition, sequence detection, and aggregation to identify combinations of structured evidence data that satisfy specific association patterns within a sliding time window.

[0059] The present invention, by introducing preset association rules and a preset event processing engine, can systematically and efficiently identify and associate multiple structured evidence data that overlap or are adjacent in time within a preset sliding time window. The event processing engine continuously monitors and analyzes the incoming structured evidence data in real time according to the preset association rules. When a combination of evidence data conforming to a specific pattern or logical relationship is detected, these data are automatically matched and associated. This mechanism ensures that evidence data from different sources and of different types can be integrated into a meaningful clinical context, thereby providing a comprehensive and logically consistent data foundation for subsequent risk assessment.

[0060] Specifically, in some implementations of the above-mentioned postoperative management methods for neurosurgical nursing, the step of calculating the initial weights of multiple structured evidence data associated within a preset sliding time window to generate a comprehensive risk score can be further refined.

[0061] The steps for calculating the initial weights of multiple structured evidence data associated within a preset sliding time window and generating a comprehensive risk score include: The initial weight of each structured evidence data is multiplied by its corresponding preset strength factor and then summed to generate a comprehensive risk score.

[0062] Structured evidence data refers to evidence information with a unified format and semantics obtained through standardized evidence transformation of multi-source heterogeneous information acquired during postoperative patient monitoring. This structured evidence data can include physiological trend structured evidence data, neurological assessment structured evidence data, nursing event structured evidence data, and auxiliary examination structured evidence data. Each piece of structured evidence data is assigned an initial weight, reflecting its inherent importance or potential impact in assessing patient risk. A pre-defined intensity factor is a multiplier factor pre-set for each piece of structured evidence data or its type, used to adjust or amplify the influence of that evidence data in a specific clinical context. For example, certain key physiological parameter abnormalities or specific nursing events may be assigned higher intensity factors to give them a more significant role in risk assessment. The overall risk score is obtained by multiplying the initial weight of each piece of structured evidence data by its corresponding pre-defined intensity factor, and then summing all these products. This score comprehensively reflects the cumulative contribution of all associated structured evidence data to the patient's current risk status within a pre-defined sliding time window.

[0063] The present invention enables a quantitative assessment of a patient's risk status by summing the initial weights of each structured piece of evidence against its corresponding preset strength factor. The initial weights ensure that the fundamental importance of each piece of evidence is taken into account, while the preset strength factor allows for dynamic adjustment of the influence of different pieces of evidence based on clinical experience or specific contexts. This summing-by-product approach allows evidence with higher initial weights and / or higher strength factors to have a greater impact on the final comprehensive risk score, thus more accurately reflecting the patient's actual risk level.

[0064] Specifically, the steps for simultaneously calculating the dynamic confidence level of the comprehensive risk score include: Based on the quantity of structured evidence data, the average strength of structured evidence data, the consistency score of structured evidence data, the quality score of structured evidence data, and the time decay factor of structured evidence data, the dynamic confidence level is calculated according to a preset confidence level comprehensive calculation function.

[0065] The quantity of structured evidence data refers to the total number of structured evidence data identified and associated within the current preset sliding time window. Generally, a larger quantity of evidence means a more robust chain of evidence supporting the risk assessment, thus helping to increase confidence. The average strength of the structured evidence data refers to the average of the initial weights or strength factors of all associated structured evidence data within the preset sliding time window. This average strength reflects the overall severity or influence of the evidence; higher strength usually indicates a clearer risk indication, and the confidence level is correspondingly higher.

[0066] Furthermore, the consistency score of structured evidence data refers to the degree of convergence among multiple related structured evidence data points indicating the same risk direction or type within a pre-defined sliding time window. For example, when multiple independent structured evidence data points to increased intracranial pressure or brain herniation risk, their consistency score is high, thus enhancing the confidence of the risk assessment. The quality score of structured evidence data refers to the evaluation of the reliability, completeness, source credibility, and collection accuracy of each structured evidence data point. For example, physiological parameter data from real-time monitoring devices typically have a high quality score, while manually entered and unverified data may have a lower score; the higher the quality score, the higher the confidence level.

[0067] Furthermore, the time decay factor of structured evidence data refers to a parameter used to measure the timeliness of structured evidence data. Generally, evidence more recent to the present time has greater relevance and influence, and a smaller decay factor; conversely, evidence older than the present time has a gradually weakening influence over time, thus reducing its contribution to the confidence level of the current risk assessment. The preset confidence level calculation function is a predefined mathematical model or algorithm used to weight, combine, or perform other logical operations on the above factors (including the quantity, average strength, consistency score, quality score, and time decay factor of structured evidence data) to ultimately derive a quantified dynamic confidence level value. This function can be optimized and adjusted based on clinical experience and historical data analysis to ensure the accuracy and clinical applicability of its calculation results.

[0068] The solution of this invention comprehensively considers the quantity, average strength, consistency score, quality score, and time decay factor of structured evidence data, and utilizes a preset confidence level calculation function to perform a multi-dimensional and dynamic assessment of the comprehensive risk score. Therefore, the system can not only identify potential risks but also assess the reliability of the risk assessment, avoiding misjudgments due to single or insufficient evidence. For example, when the amount of evidence is insufficient, the strength of the evidence is low, there are contradictions among the evidence, the quality of the evidence is questionable, or the timeliness of the evidence is poor, even if the comprehensive risk score reaches a preset threshold, the dynamic confidence level may still be low, thus prompting medical personnel to treat the warning with caution.

[0069] In some embodiments of the present invention described above, a risk warning is generated when the comprehensive risk score reaches a preset risk level threshold and the dynamic confidence level is not lower than the corresponding preset risk level confidence level lower limit. However, in actual clinical applications, simply generating a risk warning may not provide medical staff with sufficiently detailed background information or direct action guidance. This may lead to medical staff needing to spend additional time tracing the source of the risk, assessing its severity, and deciding on subsequent intervention measures. In time-sensitive scenarios such as postoperative neurosurgical monitoring, this may delay the optimal intervention time.

[0070] In response, this invention further proposes a step for generating a risk warning when the comprehensive risk score reaches a preset risk level threshold and the dynamic confidence level is not lower than the corresponding preset risk level confidence level lower limit, including: Generate risk warnings that include a visual view of the chain of evidence, which displays detailed information about all structured evidence data in the form of a timeline or list. Simultaneously, based on the comprehensive risk score and multiple structured evidence data, and based on preset clinical decision-making rules, decision support suggestions are generated and pushed to medical staff.

[0071] Specifically, when generating a risk warning, the warning is designed to include a visual representation of the chain of evidence. This visual representation aims to clearly present detailed information about all structured evidence data that led to the risk warning. The view can be displayed as a timeline, visually showing the chronological order and evolution of each piece of evidence; alternatively, it can be presented as a list, allowing healthcare professionals to quickly browse the specific content of each piece of evidence. Its purpose is to provide healthcare professionals with a comprehensive and intuitive risk context, enabling them to quickly understand the basis for the risk warning.

[0072] In addition to generating risk warnings, the system also generates and pushes decision support suggestions to medical staff based on a comprehensive risk score and multiple structured evidence data, according to preset clinical decision-making rules. These decision support suggestions are specific and actionable intervention or examination recommendations provided by the system based on the current risk situation and existing clinical experience knowledge base. For example, when the system identifies a risk of increased intracranial pressure, it may suggest "immediately repeat a head CT scan" or "adjust the dosage of sedative medication." The aim is to assist medical staff in making rapid and accurate clinical decisions, reducing their decision-making burden, and improving the timeliness and effectiveness of interventions.

[0073] This invention addresses the limitations of basic risk warning solutions, which lack sufficient information and action guidance, by providing not only warning information but also a visualized chain of evidence and decision support suggestions when generating risk warnings. Specifically, the visualized chain of evidence allows healthcare professionals to quickly trace and understand the structured evidence data leading to the risk warning, avoiding the time-consuming process of manually searching and analyzing data. Simultaneously, the decision support suggestions are generated based on a comprehensive risk score and multiple structured evidence data, combined with pre-defined clinical decision-making rules for intelligent reasoning. This directly provides healthcare professionals with targeted intervention plans, significantly shortening their decision-making time and improving decision accuracy. It is precisely because of these additional, highly relevant, and actionable pieces of information that healthcare professionals can more efficiently and accurately address potential postoperative risks for patients.

[0074] In postoperative neurosurgical care, close monitoring of patients' vital signs is crucial for ensuring safe recovery. However, existing postoperative management systems often face challenges in processing information from diverse sources and of varying natures. Particularly in complex situations where changes in a patient's condition are subtle and scattered, requiring the correlation of multiple pieces of information to detect early problems, existing systems often struggle to provide timely and effective warnings. This not only increases the workload of medical staff but may also lead to missed opportunities for optimal intervention. Existing systems primarily rely on fixed thresholds for single parameters, lacking the ability to integrate and analyze different forms of information, and are unable to automatically identify dangerous patterns composed of combinations of multiple "subthreshold" changes. This lack of information integration capability makes the window of opportunity for early intervention easily missed due to human negligence or fatigue.

[0075] Secondly, see Figure 2 This invention proposes a postoperative management system for neurosurgical nursing, the system comprising: The structured evidence data generation module 210 is used to acquire multi-source heterogeneous information of patients during postoperative monitoring, perform unified evidence transformation processing on the multi-source heterogeneous information, and generate structured evidence data. The association identification module 220 is used to identify and associate multiple structured evidence data that overlap or are adjacent in time within a preset sliding time window. The risk assessment module 230 is used to calculate the initial weights of multiple structured evidence data associated within a preset sliding time window, generate a comprehensive risk score, and simultaneously calculate the dynamic confidence level of the comprehensive risk score. The early warning generation module 240 is used to generate a risk warning when the comprehensive risk score reaches the preset risk level threshold and the dynamic confidence level is not lower than the corresponding preset risk level confidence level lower limit.

[0076] The system proposed in this invention, through its modular design, effectively integrates, dynamically correlates, assesses risks, and provides intelligent early warnings for multi-source heterogeneous information from neurosurgical patients. The structured evidence data generation module 210 is responsible for uniformly transforming patient information from different sources and in different formats into analyzable structured evidence data, solving the problem of information heterogeneity. The correlation identification module 220 intelligently identifies and correlates temporally overlapping or adjacent evidence data within a dynamic sliding time window, thereby capturing early and subtle clues to changes in the patient's condition. Based on this, the risk assessment module 230 calculates a comprehensive risk score and dynamic confidence level to quantitatively assess the patient's overall risk level and ensure the reliability of the assessment results. Finally, the early warning generation module 240 generates and pushes risk warnings in a timely manner based on preset risk level thresholds and confidence level lower limits. Through this collaborative working mechanism, this system overcomes the limitations of existing technologies in information integration and early warning, providing timely and accurate decision support for medical staff and significantly improving the efficiency and safety of postoperative neurosurgical management.

[0077] Traditional postoperative management systems primarily rely on fixed thresholds for single parameters; for example, the system issues an alarm when a physiological indicator exceeds a preset warning line. This approach falls short when dealing with complex and multifactorial postoperative conditions in neurosurgery. Many serious complications often present with subtle and diverse early signs, scattered across different types of information, which existing systems cannot connect and comprehensively analyze. This leads to information silos, making it difficult for medical staff to detect potential problems early, thus missing the optimal intervention window. In summary, the neurosurgical postoperative care management system proposed in this invention effectively addresses the limitations of existing technologies in terms of information silos and early warning through innovative multi-source heterogeneous information integration, dynamic time window correlation, and a dual risk assessment mechanism. It can identify changes in patient conditions earlier and more accurately, providing timely and reliable decision support for medical staff, thereby significantly improving the quality of postoperative neurosurgical management and patient prognosis.

[0078] The above description is merely an embodiment of the present invention and is not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for postoperative management of neurosurgical nursing, characterized in that, include: Obtain multi-source heterogeneous information about the patient during postoperative monitoring, and perform unified evidence transformation processing on the multi-source heterogeneous information to generate structured evidence data; Within a preset sliding time window, identify and associate multiple structured evidence data that overlap or are adjacent in time; The initial weights of the multiple structured evidence data associated within the preset sliding time window are calculated to generate a comprehensive risk score, and the dynamic confidence level of the comprehensive risk score is calculated simultaneously. A risk warning is generated when the comprehensive risk score reaches a preset risk level threshold and the dynamic confidence level is not lower than the corresponding preset risk level confidence level lower limit.

2. The postoperative management method for neurosurgical nursing according to claim 1, characterized in that, The multi-source heterogeneous information includes physiological parameters, neurological assessment results, nursing records, and examination results.

3. The postoperative management method for neurosurgical nursing according to claim 2, characterized in that, The step of performing unified evidence transformation processing on the multi-source heterogeneous information to generate structured evidence data includes: The physiological parameters are continuously collected, and the trend characteristics of the physiological parameters within a preset monitoring time window are calculated. Based on the aforementioned trend characteristics, structured evidence data representing the trend of physiological parameter changes is generated.

4. A method for postoperative management of neurosurgical nursing according to claim 2, characterized in that, The neurological assessment results included Glasgow Coma Scale score, pupil size and light reflex, and limb movement ability; The step of performing unified evidence transformation processing on the multi-source heterogeneous information to generate structured evidence data includes: The Glasgow Coma Scale, pupil size and light reflex, and limb movement ability manually entered by medical staff were converted into structured evidence data for neurological assessment. Semantic analysis was performed on the nursing records to obtain the original behavioral descriptions and corresponding time points; The physiological parameters within a preset time period before and after the specified time point are obtained as real-time physiological background features. Track relevant clinical intervention actions performed by medical staff within a preset time period before and after the stated time point; The original behavioral description, the real-time physiological background features, and the clinical intervention actions are matched with a preset clinical scenario pattern library. Based on the matched clinical scenario patterns, structured evidence data of nursing events is generated.

5. A method for postoperative management of neurosurgical nursing according to claim 2, characterized in that, The step of identifying and associating multiple structured evidence data that overlap or are adjacent in time within a preset sliding time window includes: Key numerical values ​​and descriptive information are extracted from the image data in the examination results and transformed into structured evidence data to assist in the examination.

6. A method for postoperative management of neurosurgical nursing according to claim 1, characterized in that, The step of identifying and associating multiple structured evidence data that overlap or are adjacent in time within a preset sliding time window includes: According to preset association rules, multiple structured evidence data appearing within the preset sliding time window are matched and associated through a preset event processing engine.

7. A method for postoperative management of neurosurgical nursing according to claim 1, characterized in that, The step of calculating the initial weights of the multiple structured evidence data associated within the preset sliding time window to generate a comprehensive risk score includes: The initial weight of each structured evidence data is multiplied by its corresponding preset strength factor and then summed to generate the comprehensive risk score.

8. A method for postoperative management of neurosurgical nursing according to claim 1, characterized in that, The step of simultaneously calculating the dynamic confidence level of the comprehensive risk score includes: Based on the quantity of the structured evidence data, the average strength of the structured evidence data, the consistency score of the structured evidence data, the quality score of the structured evidence data, and the time decay factor of the structured evidence data, the dynamic confidence level is calculated according to a preset confidence level comprehensive calculation function.

9. A method for postoperative management of neurosurgical nursing according to claim 1, characterized in that, The step of generating a risk warning when the comprehensive risk score reaches a preset risk level threshold and the dynamic confidence level is not lower than the corresponding preset risk level confidence level lower limit includes: Generate a risk warning that includes a visual view of the chain of evidence, which displays detailed information about all the structured evidence data in the form of a timeline or list. Simultaneously, based on the comprehensive risk score and multiple structured evidence data, and according to preset clinical decision-making rules, decision support suggestions are generated and pushed to medical staff.

10. A postoperative management system for neurosurgical nursing, characterized in that, The system includes: The structured evidence data generation module is used to acquire multi-source heterogeneous information of patients during postoperative monitoring, perform unified evidence transformation processing on the multi-source heterogeneous information, and generate structured evidence data. The association identification module is used to identify and associate multiple structured evidence data that overlap or are adjacent in time within a preset sliding time window; The risk assessment module is used to calculate the initial weights of multiple structured evidence data associated within the preset sliding time window, generate a comprehensive risk score, and simultaneously calculate the dynamic confidence level of the comprehensive risk score. The early warning generation module is used to generate a risk warning when the comprehensive risk score reaches a preset risk level threshold and the dynamic confidence level is not lower than the corresponding preset risk level confidence level lower limit.

Images