An intelligent management system and method for surgical anesthesia information

Abstract

The application provides a surgical anesthesia information intelligent management system and method, and relates to the technical field of data management.The system comprises a first acquisition module, which is used for acquiring preoperative standard body index data;an establishment module, which is used for establishing a support vector description model; a second acquisition module, which is used for acquiring electronic medical record data of a patient to be anesthetized and a target anesthesia operation category; a retrieval module, which is used for retrieving target preoperative body indexes; an extraction module, which is used for extracting original content related to each target preoperative body index retrieved; an input module, which is used for inputting the original content into the support vector description model corresponding to the target anesthesia operation category to output abnormal content; a first output module, which is used for outputting the original content and the abnormal content and issuing a prompt information in the case that the abnormal content exists; and a second output module, which is used for outputting the original content.The preoperative information retrieval efficiency and accuracy are improved, and abnormal data can be quickly warned.

Description

Technical Field

[0001] This invention relates to the field of data management technology, and in particular to an intelligent management system and method for surgical anesthesia information. Background Technology

[0002] Surgical anesthesia information refers to data and records related to anesthesia before surgery, including the patient's preoperative physical indicators (such as blood pressure, heart rate, and blood oxygen saturation). These preoperative physical indicators directly reflect the patient's health level and can also predict potential risks during surgery and anesthesia, thereby ensuring the safety of the surgery and the smoothness of postoperative recovery. This information is crucial for ensuring surgical safety and patient recovery.

[0003] Surgical anesthesia information management is a core element in ensuring patient safety. Effective management of surgical anesthesia information allows for real-time monitoring of patient physiological conditions, early warning of potential risks, optimization of anesthesia plans, and reduction of intraoperative emergencies. Simultaneously, management improves preoperative preparation and postoperative recovery efficiency, standardizes anesthesia records, and facilitates medical quality analysis and legal accountability. This not only increases surgical success rates but also reduces the workload of medical staff and enhances patient satisfaction.

[0004] However, the current surgical anesthesia information management mainly relies on medical staff manually flipping through the patient's electronic medical records page by page to record the patient's physical indicators. This is prone to subjective errors, which may lead to inaccurate anesthesia decisions and consequently medical accidents. In addition, in cases of emergency medical events requiring immediate anesthesia and surgery, manually flipping through information page by page can cause excessive delays in anesthesia and surgery, resulting in inaccurate anesthesia decisions and surgical delays, increasing the risk of medical accidents and patients. Summary of the Invention

[0005] To address the problems in existing surgical anesthesia information management technologies, which primarily involve medical staff manually flipping through patients' electronic medical records and recording their physical indicators page by page, leading to subjective errors and potentially inaccurate anesthesia decisions, thus causing medical accidents, and further, in cases of emergency medical events requiring immediate anesthesia, manually flipping through information page by page can cause excessive delays in anesthesia and surgery, resulting in inaccurate anesthesia decisions, surgical delays, and increased risks to both medical accidents and patients, this invention provides an intelligent surgical anesthesia information management system and method.

[0006] The technical solutions provided by the embodiments of the present invention are as follows:

[0007] First aspect

[0008] This invention provides an intelligent management system for surgical anesthesia information, the system comprising:

[0009] The first acquisition module is used to acquire preoperative standard physical indicator data for different types of anesthesia and surgery.

[0010] A module is established to incorporate preoperative abnormal physical index data using an extreme value algorithm, and to establish support vector description models belonging to different anesthesia and surgical categories by combining the preoperative abnormal physical index data and the preoperative standard physical index data.

[0011] The second acquisition module is used to acquire the electronic medical record data of the patient to be anesthetized and the target anesthesia surgery category;

[0012] The retrieval module is used to retrieve target preoperative physical indicators related to the target anesthesia surgery category from the electronic medical record data;

[0013] The extraction module is used to determine a dynamic context extraction window that is inversely proportional to the number of times the target preoperative physical indicators appear, and to extract the original text content related to each target preoperative physical indicator from the electronic medical record data through the dynamic context extraction window.

[0014] The input module is used to input the original text content into the support vector description model corresponding to the target anesthesia surgery category and output abnormal content.

[0015] The first output module is used to output the original text and the abnormal content when the abnormal content exists, and issue a reminder message about the abnormal content; otherwise, it executes the second output module.

[0016] The second output module is used to output the original text content.

[0017] Second aspect

[0018] This invention provides an intelligent management method for surgical anesthesia information, applied to an intelligent management system for surgical anesthesia information as described in the first aspect. The method includes:

[0019] S1: Obtain preoperative standard physical indicators for different types of anesthesia and surgery;

[0020] S2: Combine the extreme value algorithm to introduce preoperative abnormal physical index data, and combine the preoperative abnormal physical index data and preoperative standard physical index data to establish support vector description models belonging to different anesthesia and surgical categories.

[0021] S3: Obtain the electronic medical record data of the patient to be anesthetized and the target anesthesia procedure category;

[0022] S4: Retrieve target preoperative physical indicators related to the target anesthesia procedure category from electronic medical record data;

[0023] S5: Determine the dynamic context extraction window that is inversely proportional to the number of times the target preoperative physical indicators appear, and extract the original text content related to each target preoperative physical indicator from the electronic medical record data through the dynamic context extraction window.

[0024] S6: Input the original text into the support vector description model corresponding to the target anesthesia surgery category and output the abnormal content;

[0025] S7: If abnormal content exists, output the original content and the abnormal content, and issue a reminder message about the abnormal content; otherwise, proceed to step S8.

[0026] S8: Output the original text.

[0027] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0028] In this invention, preoperative standard physical indicator data under different anesthesia and surgical categories are combined with extreme value algorithms to introduce preoperative abnormal physical indicator data. Then, a support vector description model is established based on the preoperative standard physical indicator data and the preoperative abnormal physical indicator data for detecting preoperative physical indicator anomalies in patients awaiting anesthesia. This effectively avoids the overfitting problem caused by overly narrow boundaries when using only normal data, improves the detection accuracy of data near the boundary, and enhances the generalization ability to samples without abnormalities. This results in more accurate detection of preoperative physical indicator anomalies, particularly effective in scenarios with complex data distributions or sparse anomalies, significantly improving the robustness and reliability of the detection. During the extraction of physical indicator data from patients awaiting anesthesia, a dynamic context extraction window, inversely proportional to the frequency of occurrence of the target preoperative physical indicator, is used to extract the original text content related to each retrieved target preoperative physical indicator from the electronic medical record data for anomaly detection and output. This allows for rapid location of key information, greatly improving retrieval efficiency and accuracy, while also enabling anomaly detection and alerts. This helps doctors optimize anesthesia decisions in a timely manner, ensuring patient safety and reducing preoperative preparation time. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a schematic diagram of the structure of an intelligent management system for surgical anesthesia information provided in an embodiment of the present invention;

[0031] Figure 2This is a flowchart illustrating an intelligent management method for surgical anesthesia information provided in an embodiment of the present invention. Detailed Implementation

[0032] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0033] Reference manual attached Figure 1 The diagram shows a structural schematic of an intelligent management system for surgical anesthesia information provided in an embodiment of the present invention.

[0034] This invention provides an intelligent management system for surgical anesthesia information, the system comprising:

[0035] The first acquisition module is used to acquire preoperative standard physical indicator data for different types of anesthesia and surgery.

[0036] Anesthesia and surgical category refers to the classification of surgeries based on their type, complexity, patient condition, and required anesthesia method. Examples include general surgery, cardiac surgery, neurosurgery, pediatric surgery, and surgery for the elderly. These categories directly affect the anesthesia method (e.g., general anesthesia, local anesthesia, spinal anesthesia) and preoperative preparation requirements. Preoperative standard physical indicators refer to a series of health indicators used to assess a patient's preoperative physical condition and anesthesia risks under different anesthesia and surgical categories. Common indicators include blood pressure, heart rate, blood oxygen saturation, blood glucose, blood analysis, respiratory rate, and pulmonary function. The scope and importance of these indicators may vary depending on the different anesthesia and surgical categories.

[0037] In practical applications, obtaining preoperative standard physical indicators for different types of anesthesia and surgery can provide personalized reference standards for each type of surgery, help assess the patient's health status and anesthesia risks, and provide a scientific basis for preoperative preparation and anesthesia planning, thereby improving the safety and efficiency of the surgery.

[0038] A module is established to incorporate preoperative abnormal physical index data using an extreme value algorithm, and to build support vector description models belonging to different anesthesia and surgical categories by combining preoperative abnormal physical index data and preoperative standard physical index data.

[0039] Among them, the extreme value algorithm is a method used to detect outliers or boundary values ​​in data. It typically determines the range of outliers by analyzing the maximum and minimum values ​​in the data distribution. The extreme value algorithm can calculate the extended range based on the upper and lower bounds or quantiles of preoperative standard physical indicators, generating abnormal physical indicator data to simulate possible anomalies. Support Vector Data Description (SVDD) is a machine learning algorithm for anomaly detection. Its goal is to construct a hypersphere that surrounds normal data points while excluding outliers. By optimizing the objective function, SVDD determines the center and radius of the hypersphere and introduces slack variables to handle boundary data points or a small number of outliers, thereby improving the model's generalization performance and robustness.

[0040] It should be noted that by incorporating preoperative abnormal physical indicator data into the algorithm and building a support vector description model based on this data and preoperative standard physical indicator data, the model boundary can be optimized while simulating real abnormal situations. SVDD effectively optimizes the model boundary by dynamically adjusting the center and radius of the hypersphere, improving not only the model's ability to detect data points near the boundary but also its generalization performance on data without abnormalities. This avoids overfitting problems caused by using only normal data, thereby enhancing the model's ability to detect boundary points and samples without abnormalities, and improving the accuracy of anesthesia information monitoring and alerts. This method is particularly suitable for scenarios with complex data distributions or sparse anomalies, providing accurate and reliable references for anesthesia decisions.

[0041] In practical applications, without outliers, the hypersphere may overfit the distribution of the target data, causing the boundary to be too close to the target data and resulting in decreased model generalization performance. Introducing outliers allows adjustment of the boundary to better reflect the actual distribution of the target data. Introducing these outliers also optimizes the slack variable ε of the objective function. i This includes the influence of outliers. A larger slack variable for an outlier indicates that the outlier is located outside the hypersphere, prompting the model boundary to adjust outwards and avoid overfitting. If only the target data (number of normal points) is considered, the model tends to surround all data points, potentially leading to an overly narrow boundary and misclassification of new samples near the boundary. Introducing outliers balances the volume of the hypersphere and the weights of the slack variables in the objective function, preventing the boundary from becoming excessively close to the target data. This improves the model's generalization performance and reduces misclassification of data near the boundary, effectively enhancing the accuracy of monitoring surgical anesthesia information and providing anomaly alerts.

[0042] In one possible implementation, the module is specifically used for:

[0043] Obtain preoperative standard physical parameters for the type of anesthesia surgery to be modeled:

[0044] D standard ={[L k U k ]}

[0045] x i ~Random(L k U k ), i = 1, 2, ..., M std

[0046] Among them, L k U k D represents the lower limit and upper limit of the k-th body indicator, respectively. standard Indicates about L k U k The set of preoperative standard physical indicators, x i L represents the k-th body indicator. k U k The generated preoperative standard physical indicator data, M std x represents the total number of data points in the preoperative standard physical indicator dataset. i ~Random(L k U k ) indicates from the given range [L k U k x is generated randomly within the [inner] container. i .

[0047] It should be noted that direct random selection is a simple and efficient data generation method, suitable for scenarios where the distribution of outliers is unknown or the model is in its initial stages. It is easy to implement, has good coverage, and no prior dependencies, thus generating outliers that better reflect actual needs. Furthermore, some textual descriptions can be standardized according to the standards of their respective fields to obtain preoperative standard physical indicator data.

[0048] Generate an expanded range of data for each body metric used to generate preoperative abnormal body metric data:

[0049]

[0050] S k =α·(U k -L k )

[0051]

[0052] in, and S represents the upper limit and lower limit of the extended range of the k-th body indicator, respectively.k Let α represent the expansion amount of the k-th physical indicator, and let α represent the expansion coefficient. and Representing the intervals [U] k -L k The p-th quantile and the 1-p quantile of ].

[0053] It should be noted that determining the expansion coefficient through quantile differences avoids the interference of extreme values. Quantiles dynamically reflect the distribution characteristics of data within an interval, while avoiding the bias problems caused by directly relying on the mean or extreme values. Quantiles are unaffected by extreme outliers, making the calculation of the expansion coefficient more robust. Furthermore, this method can precisely adjust the expansion amount according to the local characteristics of the distribution, ensuring that the expansion range is neither too narrow to miss outliers nor too wide to reduce detection accuracy, thus exhibiting strong adaptability and practicality in different distribution scenarios. Ensuring the robustness of the expansion range, while combining the data distribution characteristics within the interval, allows the expansion range to dynamically adapt to local features, resulting in higher applicability and stability. Optionally, 0 < p < 0.5, specifically, values ​​can be 0.05 or 0.1. A larger expansion range can accommodate more potential outliers.

[0054] Generate abnormal preoperative physical indicators from an expanded range that does not include standard preoperative physical indicators:

[0055]

[0056] Among them, D outliers This represents the set of abnormal preoperative physical indicators, y j M represents the j-th abnormal preoperative physical indicator data. outliers This represents the total number of abnormal preoperative physical indicators generated. Indicates from a given range y is generated randomly. j ∪ represents the union symbol.

[0057] A support vector description model was constructed using a training set consisting of preoperative standard physical indicators and preoperative abnormal physical indicators.

[0058] It's important to note that introducing outlier data points when constructing a support vector model (SVM) offers the advantage of effectively optimizing the model's boundary, enabling it to more accurately distinguish between normal and outlier points. Typically, using only normal data points, the model's boundary tends to overfit the normal data distribution, resulting in insufficient ability to identify outliers, i.e., overfitting. By adding outlier data points, the model can dynamically adjust the position and shape of the boundary by optimizing the objective function and constraints, avoiding overfitting to normal points. Furthermore, the distribution characteristics of outliers help the model better expand its boundary, allowing it to accommodate more complex data distributions, improving its outlier detection capability and generalization performance, especially when the data distribution is complex or noisy.

[0059] Specifically, by introducing abnormal preoperative physical indicators and combining them with standard preoperative physical indicators to construct a support vector description model, the model boundary can be effectively optimized, avoiding overfitting caused by using only normal data. After adding outliers, the model can dynamically adjust the position and shape of the boundary, better adapting to complex data distributions and improving the ability to identify outliers and the model's generalization performance. In scenarios with complex data distributions or high noise levels, this method can significantly improve the accuracy and robustness of anomaly detection, providing more reliable support for anesthesia protocols.

[0060] In one possible implementation, a support vector description model is constructed using a training set consisting of preoperative standard physical indicator data and preoperative abnormal physical indicator data, specifically including:

[0061] Create a training set:

[0062] D train =D standard ∪D outliers

[0063] Among them, D train This represents the training set.

[0064] Determine the constraints for each training data point in the training set:

[0065] ||z t -a|| 2 ≤R 2 +ξ t ,z t ∈D train ξ t ≥0, t=1,2,...,M std +M outliers

[0066] Among them, z t Let R and a represent the training data, respectively, and let ξ represent the radius and center point of the hypersphere used to describe the support vector model.t This indicates that z is allowed. t The distance from 'a' is the slack variable, ||z t -a|| 2 Indicate z t The square of the distance to a.

[0067] Under the constraints, the training data in the training set is input into the support vector description model, and the objective function for the model parameters of the support vector description model is solved. The model parameters include the hypersphere center point, the hypersphere radius, and relaxation variables. The specific objective function is as follows:

[0068]

[0069] Where min represents taking the minimum value, and C represents the penalty coefficient for adjusting the relationship between the hypersphere volume and the sum of slack variables.

[0070] Specifically, the objective function can be solved using the Lagrange multiplier method or quadratic programming, adjusting the center of the hypersphere based on the distribution of the training data points. The final model boundary is defined by the support vectors z and (points on the boundary), where the points on the boundary satisfy ||z, -a||. 2 =R 2 .

[0071] It's important to note that the advantage of this penalty coefficient setting lies in dynamically combining the distribution characteristics of target data points and outliers. The numerator represents the total deviation of the target data points from the center, while the denominator is the deviation of outliers. When target points are more concentrated, the penalty coefficient increases, making the model fit the target data more tightly. Conversely, when outliers are more dispersed, the penalty coefficient decreases, making the model more flexible in accommodating outliers. This mechanism ensures that the model can balance overfitting and underfitting under complex distributions, significantly improving anomaly detection accuracy, especially in scenarios with uneven dataset distribution or a high proportion of outliers, thereby optimizing the model's robustness and generalization ability. By comparing the relative distributions of the two, the model can adaptively adjust the boundary during optimization, ensuring it tightly surrounds the target data points while effectively excluding outliers, thus enhancing the model's robustness and anomaly identification ability. This setting avoids the limitations of a fixed penalty coefficient, improving the model's generalization performance, especially when the distributions of target and outliers differ significantly.

[0072] Specifically, this process combines preoperative standard physical indicator data and abnormal physical indicator data, using a support vector model to dynamically adjust the center and boundary of the hypersphere, optimizing the penalty coefficient in the objective function, and achieving a balance in the distribution of target and abnormal data. While adapting to complex data distributions, the model effectively avoids overfitting and underfitting, improving the detection accuracy and generalization ability of anomalies. Especially in scenarios where target points are concentrated or anomalies are scattered, it can adaptively adjust the model boundary, significantly enhancing robustness and the reliability of anomaly detection, providing precise support for anesthesia information monitoring and decision support.

[0073] The second acquisition module is used to acquire the electronic medical record data of the patient to be anesthetized and the target anesthesia surgery category.

[0074] Electronic Medical Records (EMR) data refers to patient health information and medical history recorded electronically, including basic patient information (name, gender, age, etc.), past medical history, surgical records, medication history, preoperative examination results, diagnosis, and treatment plans. This data is crucial for hospital management and physician clinical decision-making. The target anesthesia surgery category refers to the type of surgery and its corresponding anesthesia plan determined for a specific patient. Obtaining the electronic medical record data and target anesthesia surgery category of patients awaiting anesthesia provides anesthesiologists with complete preoperative information and targeted anesthesia needs for the specific surgical type, helping to develop personalized anesthesia plans, improve the efficiency and safety of anesthesia decisions, and reduce intraoperative risks.

[0075] In one possible implementation, the electronic medical record data of the patient to be anesthetized is obtained, specifically as follows:

[0076] Obtain electronic medical record data of patients awaiting anesthesia from the HIS system.

[0077] The HIS (Hospital Information System) is a management information system for hospitals, primarily used to integrate and manage various business information of medical institutions. By acquiring electronic medical record data of patients awaiting anesthesia through the HIS system, patients' health information and medical history can be extracted quickly and accurately, providing efficient data support for preoperative assessment and anesthesia plan development, avoiding omissions and delays caused by manual retrieval.

[0078] The retrieval module is used to retrieve target preoperative physical indicators related to the target anesthesia category from electronic medical record data.

[0079] Understandably, by extracting preoperative physical indicators related to the target anesthesia surgery category from electronic medical record data, it is possible to efficiently locate and accurately retrieve all relevant information, reduce omissions in manual searches, significantly improve information retrieval efficiency, shorten decision-making time, and provide a guarantee for the rapid development of anesthesia plans.

[0080] In one possible implementation, the retrieval module is specifically used for:

[0081] Retrieve each target preoperative physical indicator from electronic medical record data using an inverted index algorithm:

[0082]

[0083] Wherein, G(o) represents the set of search results for the target preoperative physical indicator o, A1, A2, A3, A4 represent the document ID, page number, line number, and page time tag of each search result, i.e., the target preoperative physical indicator, respectively, syn(o) represents the set of synonyms for o, o` represents the synonyms for the target preoperative physical indicator o, Sim(o`, w) represents the similarity between o` and the keyword w in the electronic medical record data, θ represents the similarity threshold, w∈A1 means that w belongs to the document corresponding to A1, EditDist(o`, w) represents the edit distance between o` and w, max means take the maximum value, and |o`|, |w| represent the text lengths of o` and w, respectively.

[0084] Here, syn(o) represents the set of synonyms for o, which can be retrieved from the Wordnet vocabulary database.

[0085] It should be noted that those skilled in the art can set the similarity threshold according to actual needs, and this invention does not limit it.

[0086] Specifically, this retrieval process uses an inverted index algorithm to locate target preoperative physical indicators from electronic medical record data, and leverages edit distance and lexical similarity calculations to improve the flexibility and accuracy of keyword matching. By combining the synonym set of the target indicator, it ensures that the search results comprehensively cover related information, while filtering results based on similarity thresholds to remove low-relevance data. This process can quickly and efficiently extract the document ID, page number, line number, and time tag corresponding to each indicator, providing accurate basic information for subsequent data processing and analysis, and improving retrieval efficiency and accuracy.

[0087] The extraction module is used to determine a dynamic context extraction window that is inversely proportional to the frequency of occurrence of the target preoperative physical indicators, and to extract the original text content related to each target preoperative physical indicator from the electronic medical record data through the dynamic context extraction window.

[0088] The dynamic context extraction window is an extraction range that automatically adjusts based on the frequency of the target preoperative physical indicator in the electronic medical record. When the target indicator appears frequently, the window shrinks to avoid information redundancy. When the frequency is low, the window expands to ensure sufficient contextual information is extracted. The original text content refers to the textual information extracted from the electronic medical record that is directly related to the target preoperative physical indicator. This content includes indicator values, examination reports, and medical history descriptions, and is an important data source for doctors' analysis and decision-making.

[0089] It's important to note that accurate extraction of the original text not only provides the data for the target indicator itself but also includes contextual information (such as time labels, testing conditions, and related descriptions), thus offering doctors more complete background information. Dynamically adjusting the window avoids information omissions and redundancy, ensuring the extracted data is highly relevant and usable. This helps doctors quickly understand the meaning behind the indicators, optimize anesthesia plans, improve the comprehensiveness and accuracy of preoperative assessments, and save time on manual screening and judgment, reducing the risk of misjudgment.

[0090] In one possible implementation, the extraction module is specifically used for:

[0091] Using Chebyshev's inequality, the dynamic context extraction window is determined based on the frequency of occurrence of each target preoperative physical indicator within its respective document:

[0092]

[0093] Where n represents the dynamic context extraction window, and N(o) represents the number of times the target preoperative physical indicator o appears in the relevant document, i.e., the electronic medical record data. This indicates rounding down, where τ represents the coverage of the context content related to o in the document.

[0094] Chebyshev's inequality is an important tool in probability theory used to measure the upper bound of the probability of a random variable deviating from its mean. In the dynamic adjustment of the context extraction window, Chebyshev's inequality is used to estimate the distribution characteristics of the target preoperative physical indicators appearing in the document. By treating the frequency of occurrence of the target indicator as a statistical distribution, Chebyshev's inequality can be used to determine a reasonable range of the context window. Through the application of Chebyshev's inequality, the context extraction module can more intelligently balance the coverage and accuracy of the extracted information, ensuring that the extracted original text content is both comprehensive and avoids interference from redundant noise, thus improving the efficiency and accuracy of extracting information related to preoperative physical indicators.

[0095] Understandably, when keyword density is high, the window size is controlled to avoid redundancy in context extraction; when density is low, the window size is expanded to ensure sufficient information is extracted. Simultaneously, dynamically adapting to target coverage and data distribution allows the window size to be intelligently adjusted. During context extraction, this ensures sufficient information coverage while reducing redundancy, making it suitable for complex data analysis and content extraction scenarios.

[0096] Extract the contextual information corresponding to each target preoperative physical indicator from the electronic medical record data page by page:

[0097]

[0098] in, This represents the total number of lines in page A2, Context(o) represents the context content of the target preoperative physical indicator o, and C(A1,A2,l) represents the content of line l on page A2 in document A1.

[0099] In this case, max(1,A3-n) ensures that the starting line number of the context window is not less than 1 (the first line of the page). Ensure that the end line number of the context window is not greater than the maximum number of lines on the page, i.e., the total number of lines.

[0100] The contextual content corresponding to the latest page's timestamp is output as the original text content of the target preoperative physical indicators.

[0101] It should be noted that this process dynamically determines the context extraction window for the target preoperative physical indicators by combining Chebyshev's inequality, and intelligently adjusts the window size based on the frequency of the indicator's occurrence in the document. When keyword density is high, the window shrinks to avoid redundant information; when density is low, the window expands to ensure sufficient information extraction. During extraction, the start and end line numbers of the window are kept within the page range. Contextual content is extracted page by page from the electronic medical record, and the latest content is located using time tags as the original text output. This method reduces redundant information while ensuring complete coverage, improving extraction efficiency and accuracy, and providing precise data support for subsequent analysis and decision-making.

[0102] The input module is used to input the original text into the support vector description model corresponding to the target anesthesia surgery category and output abnormal content.

[0103] In one possible implementation, the input module is specifically used for:

[0104] The preoperative preoperative physical index values ​​of each target preoperative physical index are extracted from the original text using a pre-trained natural language processing model.

[0105] In one possible implementation, the natural language processing model includes the BERT model and the NER model.

[0106] The target preoperative physical indicators are mapped to a high-dimensional feature space using a Gaussian kernel function.

[0107] The mapping result is input into the support vector description model, and the Euclidean distance between the mapping result and the hypersphere center of the support vector description model is calculated.

[0108] If the Euclidean distance is greater than the hypersphere radius of the support vector description model, the target preoperative physical index value and the corresponding target preoperative physical index will be output as abnormal content.

[0109] Specifically, this process first uses pre-trained natural language processing models (such as BERT and NER models) to accurately extract numerical data of target preoperative physical indicators from the original text. Then, these indicator values ​​are mapped to a high-dimensional feature space using a Gaussian kernel function to better capture the nonlinear characteristics of the data. Next, the mapped results are input into a support vector description model. By calculating the Euclidean distance between the mapped result and the center of the hypersphere, it is determined whether the indicator exceeds the model's normal range. If the distance is greater than the hypersphere radius, the indicator is considered abnormal, and the corresponding indicator and value are output as abnormal content. This efficiently achieves anomaly detection in complex medical data, providing accurate support for anesthesia decision-making.

[0110] The first output module is used to output the original content and the abnormal content if abnormal content exists, and to issue a reminder message about the abnormal content; otherwise, the second output module is executed.

[0111] It can quickly extract relevant original text content from electronic medical record data, and the electronic medical record data has time tags and provides early warnings for abnormal content.

[0112] The second output module is used to output the original text content.

[0113] In one possible implementation, it also includes:

[0114] The update module is used to update the support vector description model at preset intervals.

[0115] Understandably, updating the support vector description model at preset intervals allows for dynamic adaptation to changes in new data, improving the model's real-time performance and accuracy. Those skilled in the art can adjust the update frequency as needed to ensure the model always reflects the latest distribution of body indicator data, providing more reliable support for anomaly detection.

[0116] It should be noted that those skilled in the art can set the preset duration according to actual needs, and this invention does not limit this.

[0117] In practical applications, preoperative standard physical indicators for different types of anesthesia and surgery are acquired, and abnormal indicator data are generated using extreme value algorithms. An anomaly detection model is then constructed based on a support vector model. The system quickly retrieves preoperative physical indicators related to the target anesthesia and surgery category from the patient's electronic medical record. It also obtains the complete original text of the target indicators through a dynamic context extraction window, combining time tags to ensure data relevance and accuracy. Subsequently, the extracted original text is input into the SVDD model for anomaly detection. If anomalies are found, the system outputs the original text and anomaly alerts, providing doctors with precise preoperative analysis support, improving anesthesia decision-making efficiency and surgical safety.

[0118] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0119] In this invention, preoperative standard physical indicator data under different anesthesia and surgical categories are combined with extreme value algorithms to introduce preoperative abnormal physical indicator data. Then, a support vector description model is established based on the preoperative standard physical indicator data and the preoperative abnormal physical indicator data for detecting preoperative physical indicator anomalies in patients awaiting anesthesia. This effectively avoids the overfitting problem caused by overly narrow boundaries when using only normal data, improves the detection accuracy of data near the boundary, and enhances the generalization ability to samples without abnormalities. This results in more accurate detection of preoperative physical indicator anomalies, particularly effective in scenarios with complex data distributions or sparse anomalies, significantly improving the robustness and reliability of the detection. During the extraction of physical indicator data from patients awaiting anesthesia, a dynamic context extraction window, inversely proportional to the frequency of occurrence of the target preoperative physical indicator, is used to extract the original text content related to each target preoperative physical indicator retrieved from the electronic medical record data for anomaly detection and output. This allows for rapid location of key information, greatly improving retrieval efficiency and accuracy, while also enabling anomaly detection and alerts. This helps doctors optimize anesthesia decisions in a timely manner, ensuring patient safety and reducing preoperative preparation time.

[0120] Reference manual attached Figure 2 The diagram shows a flowchart of an intelligent management method for surgical anesthesia information provided by an embodiment of the present invention.

[0121] This invention also provides an intelligent management method for surgical anesthesia information, applied to the aforementioned intelligent management system for surgical anesthesia information, the method comprising:

[0122] S1: Obtain preoperative standard physical indicators for different types of anesthesia and surgery.

[0123] S2: Incorporate abnormal preoperative physical indicators data using the extreme value algorithm, and establish support vector description models belonging to different anesthesia and surgical categories by combining abnormal preoperative physical indicators data with standard preoperative physical indicators data.

[0124] S3: Obtain the electronic medical record data of the patient to be anesthetized and the target anesthesia procedure category.

[0125] S4: Retrieve target preoperative physical indicators related to the target anesthesia category from electronic medical record data.

[0126] S5: Determine a dynamic context extraction window that is inversely proportional to the frequency of occurrence of the target preoperative physical indicators, and extract the original text content related to each target preoperative physical indicator from the electronic medical record data through the dynamic context extraction window.

[0127] S6: Input the original text into the support vector description model corresponding to the target anesthesia surgery category and output the abnormal content.

[0128] S7: If abnormal content exists, output the original content and the abnormal content, and issue a reminder message about the abnormal content; otherwise, proceed to step S8.

[0129] S8: Output the original text.

[0130] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0131] In this invention, preoperative standard physical indicator data under different anesthesia and surgical categories are combined with extreme value algorithms to introduce preoperative abnormal physical indicator data. Then, a support vector description model is established based on the preoperative standard physical indicator data and the preoperative abnormal physical indicator data for detecting preoperative physical indicator anomalies in patients awaiting anesthesia. This effectively avoids the overfitting problem caused by overly narrow boundaries when using only normal data, improves the detection accuracy of data near the boundary, and enhances the generalization ability to samples without abnormalities. This results in more accurate detection of preoperative physical indicator anomalies, particularly effective in scenarios with complex data distributions or sparse anomalies, significantly improving the robustness and reliability of the detection. During the extraction of physical indicator data from patients awaiting anesthesia, a dynamic context extraction window, inversely proportional to the frequency of occurrence of the target preoperative physical indicator, is used to extract the original text content related to each target preoperative physical indicator retrieved from the electronic medical record data for anomaly detection and output. This allows for rapid location of key information, greatly improving retrieval efficiency and accuracy, while also enabling anomaly detection and alerts. This helps doctors optimize anesthesia decisions in a timely manner, ensuring patient safety and reducing preoperative preparation time.

[0132] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage system such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0133] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0134] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0135] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0136] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0137] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0138] In the embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0139] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0140] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0141] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer system (which may be a personal computer, server, or network system, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0142] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0143] The following points need to be explained:

[0144] (1) The accompanying drawings of the embodiments of the present invention only involve the structures involved in the embodiments of the present invention. Other structures can refer to the general design.

[0145] (2) For clarity, the thickness of layers or regions is enlarged or reduced in the drawings used to describe embodiments of the present invention; that is, these drawings are not drawn to actual scale. It is understood that when an element such as a layer, film, region, or substrate is referred to as being “above” or “below” another element, the element may be “directly” located “above” or “below” the other element, or there may be intermediate elements.

[0146] (3) Where there is no conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other to obtain new embodiments.

[0147] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A surgical anesthesia information intelligent management system, characterized in that, include: The first acquisition module is used to acquire preoperative standard physical indicator data for different types of anesthesia and surgery. A module is established to incorporate preoperative abnormal physical index data using an extreme value algorithm, and to establish support vector description models belonging to different anesthesia and surgical categories by combining the preoperative abnormal physical index data and the preoperative standard physical index data. The second acquisition module is used to acquire the electronic medical record data of the patient to be anesthetized and the target anesthesia surgery category; The retrieval module is used to retrieve target preoperative physical indicators related to the target anesthesia surgery category from the electronic medical record data; The extraction module is used to determine a dynamic context extraction window that is inversely proportional to the number of times the target preoperative physical indicators appear, and to extract the original text content related to each target preoperative physical indicator from the electronic medical record data through the dynamic context extraction window. The input module is used to input the original text content into the support vector description model corresponding to the target anesthesia surgery category and output abnormal content. The first output module is used to output the original text and the abnormal content when the abnormal content exists, and issue a reminder message about the abnormal content; otherwise, it executes the second output module. The second output module is used to output the original text content.

2. The intelligent management system for surgical anesthesia information according to claim 1, characterized in that, The establishment module is specifically used for: Obtain preoperative standard physical indicators for the type of anesthesia surgery to be modeled; Generate an extended range of data for each body indicator to generate preoperative abnormal body indicator data; Preoperative abnormal physical indicator data are generated from an extended range that does not include the aforementioned standard preoperative physical indicator data; A support vector description model is constructed using a training set consisting of the preoperative standard physical index data and the preoperative abnormal physical index data.

3. The intelligent management system for surgical anesthesia information according to claim 2, characterized in that, The construction of a support vector description model using a training set composed of the preoperative standard physical indicator data and the preoperative abnormal physical indicator data specifically includes: Establish the training set; Determine the constraints for each training data point in the training set; Under the constraints, the training data in the training set is input into the support vector description model to solve the objective function of the model parameters of the support vector description model, wherein the model parameters include the hypersphere center point, the hypersphere radius, and relaxation variables.

4. The intelligent management system for surgical anesthesia information according to claim 1, characterized in that, The acquisition of the electronic medical record data of the patient to be anesthetized specifically includes: Obtain the electronic medical record data of the patient to be anesthetized from the HIS system.

5. The intelligent management system for surgical anesthesia information according to claim 1, characterized in that, The retrieval module is specifically used for: Each target preoperative physical indicator is retrieved from the electronic medical record data using an inverted index algorithm.

6. The intelligent management system for surgical anesthesia information according to claim 5, characterized in that, The extraction module is specifically used for: Combining Chebyshev's inequality, the dynamic context extraction window is determined based on the number of times each target preoperative physical indicator appears in the document. Extract the contextual content corresponding to each target preoperative physical indicator from the electronic medical record data page by page; The contextual content corresponding to the latest page's timestamp is output as the original text content of the target preoperative physical indicators.

7. The intelligent management system for surgical anesthesia information according to claim 1, characterized in that, The input module is specifically used for: The target preoperative physical index values ​​of each target preoperative physical index are extracted from the original text content using a pre-trained natural language processing model. The target preoperative physical indicators are mapped to a high-dimensional feature space using a Gaussian kernel function; The mapping result is input into the support vector description model, and the Euclidean distance between the mapping result and the hypersphere center of the support vector description model is calculated. If the Euclidean distance is greater than the hypersphere radius of the support vector description model, the target preoperative physical index value and the corresponding target preoperative physical index are output as the abnormal content.

8. The intelligent management system for surgical anesthesia information according to claim 7, characterized in that, The natural language processing models include the BERT model and the NER model.

9. The intelligent management system for surgical anesthesia information according to claim 1, characterized in that, Also includes: The update module is used to update the support vector description model at preset intervals.

10. A method for intelligent management of surgical anesthesia information, characterized in that, The method of the intelligent management system for surgical anesthesia information applied to any one of claims 1 to 9 includes: S1: Obtain preoperative standard physical indicators for different types of anesthesia and surgery; S2: Incorporate abnormal preoperative physical index data using the extreme value algorithm, and establish support vector description models belonging to different anesthesia and surgical categories by combining the abnormal preoperative physical index data and the standard preoperative physical index data. S3: Obtain the electronic medical record data of the patient to be anesthetized and the target anesthesia procedure category; S4: Retrieve target preoperative physical indicators related to the target anesthesia surgery category from the electronic medical record data; S5: Determine a dynamic context extraction window that is inversely proportional to the number of times the target preoperative physical indicators appear, and extract the original text content related to each target preoperative physical indicator from the electronic medical record data through the dynamic context extraction window. S6: Input the original text into the support vector description model corresponding to the target anesthesia surgery category, and output the abnormal content; S7: If the abnormal content exists, output the original text and the abnormal content, and issue a reminder message about the abnormal content; otherwise, proceed to step S8. S8: Output the original text content.

Images