Neurosurgery patient postoperative care risk early warning system

Abstract

The invention discloses a neurosurgery patient postoperative care risk early warning system, and relates to the technical field of postoperative care risk early warning. According to the postoperative care risk early warning system for the neurosurgical patient, multi-source postoperative data such as brain waves, physiological indexes, behaviors and inflammation are integrated through the data acquisition module, single-index anomaly recognition is achieved through the sub-item risk assessment module, and then overall risk judgment is conducted by further fusing data when single indexes are normal through the comprehensive risk assessment module; finally, the early warning module performs graded early warning, so that multi-dimensional data fusion analysis is realized, abnormity of a single index can be found in time, potential overall risks can be mined through a comprehensive model, the comprehensiveness, accuracy and timeliness of early warning are improved, scientific and accurate risk judgment basis is provided for postoperative nursing, early intervention of medical staff is assisted, and the medical staff is prevented from suffering from early warning. Problems of single postoperative care risk assessment and lack of comprehensive judgment of existing neurosurgery patients are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of postoperative care risk warning, and specifically to a postoperative care risk warning system for neurosurgical patients. Background Art

[0002] Neurosurgical diseases often involve complex surgical interventions on the brain or spinal cord. These surgeries may bring various postoperative complications, including increased intracranial pressure, insufficient cerebral perfusion, abnormal cerebral oxygen saturation, etc. The characteristics of rapid and complex postoperative condition changes make it difficult for traditional nursing methods (relying on artificial experience and static scoring mechanisms) to achieve accurate monitoring and rapid response. This deficiency may lead to the failure to detect the deterioration of the condition in a timely manner, increasing the risk of complications. In addition, the medical staff in the neurosurgical ward are under greater pressure. Compared with the nursing in other departments, timeliness is more emphasized in the neurosurgical intensive care unit.

[0003] The Chinese patent application with the publication number CN119694597A discloses a neurosurgical nursing assessment method based on clinical warning. By obtaining the historical vital sign data of the monitored patient, the Isolation Forest algorithm is used to detect abnormal data points, the time series data is smoothed by local weighted regression to eliminate noise data; the interpolation method is used to fill in the missing data values; the sliding window method is used to segment the time series data to extract trend and fluctuation feature data; for each type of vital sign data, a time fluctuation curve is constructed respectively based on the fluctuation feature data, and the postoperative stable period is used as the critical point to calculate the fluctuation index of the time fluctuation curve before the critical point; based on the fluctuation index, it is compared with the threshold value output by the machine learning model to identify abnormal signals, significantly improving the accuracy of clinical warning.

[0004] However, the existing postoperative care risk assessment methods for neurosurgical patients mostly focus on single-dimensional data, do not integrate multi-source heterogeneous information such as electroencephalogram, physiology, and behavior, and are difficult to capture the risk of index linkage; moreover, the mining of data association and collaborative influence is lacking, and the comprehensive recovery status of patients cannot be characterized as a whole, and complex risks are easily missed. Therefore, there are problems of single assessment and lack of comprehensive judgment. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides a postoperative care risk warning system for neurosurgical patients, which solves the problems of single postoperative care risk assessment and lack of comprehensive judgment for existing neurosurgical patients.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: A postoperative nursing risk early warning system for neurosurgical patients, including a data acquisition module, a sub-item risk assessment module, a comprehensive risk assessment module, and an early warning module, wherein: The data acquisition module is used to acquire various index data of neurosurgical patients after surgery, including electroencephalogram signal data, physiological index data, behavior data, and inflammation data; The sub-item risk assessment module is used to conduct sub-item risk assessments on various index data to obtain various risk assessment results, including electroencephalogram assessment results, physiological index assessment results, behavior assessment results, and inflammation assessment results. Each risk assessment result includes an abnormal assessment result and a normal assessment result; The comprehensive risk assessment module is used to fuse various index data when all risk assessment results are normal assessment results, obtain a trained comprehensive risk assessment model, and process the fused various index data based on the trained comprehensive risk assessment model to obtain a comprehensive risk assessment result, including an abnormal comprehensive assessment result and a normal comprehensive assessment result; The early warning module is used to acquire various risk assessment results and the comprehensive risk assessment result, and issue an early warning based on various risk assessment results and the comprehensive risk assessment result.

[0007] Further, the data acquisition module includes an electroencephalogram signal acquisition unit, a physiological index acquisition unit, a behavior data acquisition unit, and an inflammation data acquisition unit, wherein: The electroencephalogram signal acquisition unit is used to connect with a high-density electroencephalogram device, acquire initial electroencephalogram signals at multiple electrode sites of the whole brain, and process the initial electroencephalogram signals based on independent component analysis and band-pass filtering to obtain electroencephalogram signal data; The physiological index acquisition unit is used to acquire the basic physiological indexes of neurosurgical patients after surgery, preprocess the basic physiological indexes and then perform standardization processing to obtain physiological index data; The behavior data acquisition unit is used to acquire patient movement data based on a camera, identify the patient's behavior based on a target detection algorithm, determine the patient's behavior, and at the same time acquire behavior data based on an inertial measurement sensor; The inflammation data acquisition unit is used to acquire the postoperative clinical examination results of the patient, extract the initial inflammation data, and perform standardization processing on the initial inflammation data to obtain inflammation data.

[0008] Further, the electroencephalogram signal data includes electroencephalogram frequency band parameters, electroencephalogram time domain parameters, and electroencephalogram spatial parameters; The physiological index data includes basic physical sign parameters and neurosurgery special parameters; The behavior data includes the acceleration during movement and the angular velocity during movement corresponding to each patient's behavior; The inflammation data includes blood inflammation parameters and inflammatory mediator parameters.

[0009] Further, the sub-item risk assessment module includes an electroencephalogram signal risk assessment unit, a physiological index risk assessment unit, a behavior risk assessment unit, and an inflammation risk assessment unit, wherein: The electroencephalogram signal risk assessment unit is used to identify abnormalities in electroencephalogram signal data based on an autoencoder and output an electroencephalogram assessment result; The physiological index risk assessment unit is used to obtain a physiological index assessment coefficient based on physiological index data, and compare the physiological index assessment coefficient with the physiological index assessment threshold stored in the database: if the physiological index assessment coefficient is greater than the physiological index assessment threshold stored in the database, the output physiological index assessment result is that the assessment result is abnormal; if the physiological index assessment coefficient is not greater than the physiological index assessment threshold stored in the database, the output physiological index assessment result is that the assessment result is normal; The behavior risk assessment unit is used to obtain a behavior assessment coefficient based on behavior data, and compare the behavior assessment coefficient with the behavior assessment threshold stored in the database: if the behavior assessment coefficient is greater than the behavior assessment threshold stored in the database, the output behavior assessment result is that the assessment result is abnormal; if the behavior assessment coefficient is not greater than the behavior assessment threshold stored in the database, the output behavior assessment result is that the assessment result is normal; The inflammation risk assessment unit is used to obtain an inflammation assessment coefficient based on inflammation data, and compare the inflammation assessment coefficient with the inflammation assessment threshold stored in the database: if the inflammation assessment coefficient is greater than the inflammation assessment threshold stored in the database, the output inflammation assessment result is that the assessment result is abnormal; if the inflammation assessment coefficient is not greater than the inflammation assessment threshold stored in the database, the output inflammation assessment result is that the assessment result is normal.

[0010] Further, the process of the autoencoder for abnormal identification of electroencephalogram signal data is as follows: obtain the abnormal-free electroencephalogram signal data regularly updated in the database; train the autoencoder based on the abnormal-free electroencephalogram signal data. The process is: the encoder in the autoencoder compresses the input abnormal-free electroencephalogram signal data into a low-dimensional electroencephalogram signal feature vector, and the decoder in the autoencoder reconstructs the low-dimensional electroencephalogram signal feature vector back to the original dimension to obtain the reconstructed abnormal-free electroencephalogram signal data; calculate the mean square error based on the abnormal-free electroencephalogram signal data and the reconstructed abnormal-free electroencephalogram signal data, and through the backpropagation algorithm, backpropagate the value of the loss function to each parameter of the encoder and decoder, and use the gradient descent method to adjust the parameters to continuously reduce the mean square error. When the mean square error is less than the set mean square error threshold, the training ends, and the trained encoder is output, where the parameters include weights and biases; Input the electroencephalogram signal data into the trained encoder, and after encoding and decoding operations, obtain the reconstructed electroencephalogram signal data to be verified. Calculate the mean square error to be verified based on the reconstructed electroencephalogram signal data to be verified and the electroencephalogram signal data; if the mean square error to be verified is greater than the set error threshold, the output electroencephalogram assessment result is that the assessment result is abnormal; if the mean square error to be verified is not greater than the set error threshold, the output electroencephalogram assessment result is that the assessment result is normal.

[0011] Further, the basic physical sign parameters include heart rate, blood pressure, respiratory rate, and body temperature; the special neurosurgery parameters include intracranial pressure and blood oxygen saturation. The process of obtaining the physiological index evaluation coefficient based on the physiological index data is as follows: Obtain the patient's pathological basic information, including age, gender, weight, height, and basic disease information; obtain the basic information of historical patient cases who have undergone the same neurosurgery stored in the case database, analyze the similarity between the patient's pathological basic information and the basic information of historical patient cases based on cosine similarity, and determine the basic information of the historical patient case that is most similar to the patient's pathological basic information; obtain the corresponding postoperative physiological index record data stored in the database based on the most similar historical patient case basic information, including the postoperative physiological index data for each day after surgery; perform a difference process on the physiological index data of the current day and the corresponding postoperative physiological index data in the patient's postoperative physiological index record data to obtain the physiological index evaluation coefficient.

[0012] Further, the process of obtaining the behavior evaluation coefficient based on the behavior data is as follows: Obtain the postoperative behavior data of patients undergoing the same type of neurosurgery; perform a standardization process on the behavior data and the postoperative behavior data and then perform a cosine similarity value analysis to obtain the behavior evaluation coefficient.

[0013] Further, the blood inflammation parameters include the content of various white blood cells, C-reactive protein content, and procalcitonin content; the inflammatory mediator parameters include interleukin-6 content and tumor necrosis factor-α content. The process of obtaining the inflammation evaluation coefficient based on the inflammation data is as follows: Obtain the normal human inflammation data, including normal human blood inflammation parameters and normal human inflammatory mediator parameters; obtain the postoperative inflammation data of patients undergoing the same type of neurosurgery, including postoperative blood inflammation parameters and postoperative inflammatory mediator parameters; obtain the first inflammation evaluation factor based on the normal human inflammation data and the inflammation data; obtain the second inflammation evaluation factor based on the postoperative inflammation data of patients undergoing the same type of neurosurgery and the inflammation data; perform a weighted sum of the first inflammation evaluation factor and the second inflammation evaluation factor to obtain the inflammation evaluation coefficient.

[0014] Further, the comprehensive risk assessment model is a gradient boosting decision tree model. The process of obtaining the comprehensive risk assessment result based on the comprehensive risk assessment model is as follows: Perform a dimensionless process on each index data and splice it into a comprehensive index vector; input the comprehensive index vector into the comprehensive risk assessment model to obtain the predicted values output by each decision tree, accumulate the predicted values according to the set weights to obtain the probability that the comprehensive index vector belongs to the normal; if the probability is greater than the set normal probability threshold, output the comprehensive risk assessment result as the comprehensive assessment result is normal; if the probability is not greater than the set normal probability threshold, output the comprehensive risk assessment result as the comprehensive assessment result is abnormal.

[0015] Furthermore, the process of issuing early warnings based on the results of each risk assessment and the comprehensive risk assessment is as follows: Count the number of abnormal assessment results obtained from the sub-risk assessments. If the number is not greater than two, issue a first-level warning; if the number is greater than two, issue a second-level warning. If an abnormal comprehensive assessment result is received, issue a second-level warning directly.

[0016] The present invention has the following beneficial effects: The postoperative care risk warning system for neurosurgery patients integrates multi-source postoperative data such as brain waves, physiological indicators, behavior and inflammation through the data acquisition module, and uses the sub-item risk assessment module to realize the identification of single indicator abnormalities. Then, the comprehensive risk assessment module further integrates the data to make an overall risk judgment when the single indicators are normal. Finally, the warning module uses graded warnings to realize multi-dimensional data fusion analysis. It can not only detect abnormalities of single indicators in a timely manner, but also explore potential overall risks through comprehensive models, improve the comprehensiveness, accuracy and timeliness of warnings, provide scientific and accurate risk judgment basis for postoperative care, help medical staff intervene early, improve patient prognosis, and solve the problem of single risk assessment and lack of comprehensive judgment in postoperative care for existing neurosurgery patients.

[0017] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flowchart of the risk warning system for postoperative care of neurosurgery patients of the present invention.

[0019] Figure 2 A flow chart of the process of obtaining inflammation assessment coefficients from inflammation data of the postoperative care risk warning system for neurosurgery patients of the present invention. DETAILED DESCRIPTION

[0020] See also Figure 1 The embodiment of the present invention provides a technical solution: a risk warning system for postoperative care of neurosurgery patients, including a data acquisition module, a sub-item risk assessment module, a comprehensive risk assessment module and a warning module, wherein: the data acquisition module is used to obtain various indicator data of postoperative neurosurgery patients, including brain wave signal data, physiological indicator data, behavioral data and inflammation data.

[0021] The data acquisition module includes an electroencephalogram (EEG) signal acquisition unit, a physiological index acquisition unit, a behavior data acquisition unit, and an inflammation data acquisition unit, where: The EEG signal acquisition unit is used to connect to a high-density electroencephalogram device to obtain the initial EEG signals at multiple electrode sites of the whole brain, such as the standard sites of the international 10-20 system, and capture the EEG signals in real time at a high sampling frequency of ≥1000 Hz. The initial EEG signals are processed based on independent component analysis (to separate and remove bioelectric artifacts such as electrooculogram and electromyogram to avoid interfering with the authenticity of the EEG signals) and band-pass filtering to obtain EEG signal data. Band-pass filters are designed for different frequency bands (alpha wave 8-13 Hz, beta wave 14-30 Hz, gamma wave 30-100 Hz, delta wave 0.5-4 Hz, theta wave 4-8 Hz) to extract the effective signal components related to functions such as brain arousal, sleep, and cognition.

[0022] The physiological index acquisition unit is used to obtain the basic physiological indexes of neurosurgical patients after surgery. After preprocessing the basic physiological indexes (removing outliers through the 3σ principle and filling in missing data using linear interpolation), they are then normalized (through Z-score or Min-Max normalization) to obtain physiological index data; The behavior data acquisition unit is used to collect the patient's movement data based on a camera and identify the patient's behavior (such as turning over, sitting up, getting out of bed, etc.) based on a target detection algorithm (such as YOLOv5). At the same time, behavior data is obtained based on an inertial measurement unit (IMU, including an accelerometer and a gyroscope); The inflammation data acquisition unit is used to obtain the patient's postoperative clinical examination results, extract the initial inflammation data, and obtain inflammation data after normalizing the initial inflammation data.

[0023] The EEG signal data includes EEG frequency band parameters (energy ratio and power spectral density of each frequency band), EEG time domain parameters (amplitude and latency), and EEG spatial parameters (phase difference and correlation between electrodes); The physiological index data includes basic physical sign parameters and special neurosurgical parameters; The behavior data includes the acceleration and angular velocity during movement corresponding to each patient's behavior; The inflammation data includes blood inflammation parameters and inflammatory mediator parameters.

[0024] The EEG signals directly reflect the neurological function status (such as epileptiform discharges and ischemic slow waves in the brain), the physiological indexes reveal the body's stress and organ functions (such as increased intracranial pressure indicating cerebral edema), the behavior data reflects the motor function and nursing compliance (such as reduced activity may lead to pressure ulcers), and the inflammation data warns of complications such as infections (such as PCT > 0.5 ng / mL indicating bacterial infection). Through technical integration and precise processing, the data acquisition module constructs a three-dimensional data network covering neurological function, systemic status, behavior patterns, and inflammatory responses, providing underlying data support for solving the problems of "fragmentation, lag, and simplification" in existing postoperative risk assessment.

[0025] The sub - item risk assessment module is used to conduct sub - item risk assessments on various index data, obtaining various risk assessment results, including electroencephalogram assessment results, physiological index assessment results, behavior assessment results, and inflammation assessment results. Each risk assessment result includes abnormal assessment results and normal assessment results.

[0026] The sub - item risk assessment module includes an electroencephalogram signal risk assessment unit, a physiological index risk assessment unit, a behavior risk assessment unit, and an inflammation risk assessment unit. Among them: The electroencephalogram signal risk assessment unit is used to identify abnormalities in electroencephalogram signal data based on an auto - encoder and output electroencephalogram assessment results.

[0027] The process of the auto - encoder identifying abnormalities in electroencephalogram signal data is as follows: Obtain the anomaly - free electroencephalogram signal data regularly updated in the database and train the auto - encoder based on the anomaly - free electroencephalogram signal data. The process is as follows: The encoder in the auto - encoder compresses the input anomaly - free electroencephalogram signal data into a low - dimensional electroencephalogram signal feature vector. The encoder consists of 2 - 3 fully - connected layers and uses the ReLU activation function to compress the high - dimensional electroencephalogram signal (such as a 1 - second signal with a sampling rate of 1000Hz, corresponding to 1000 - dimensional data) into a low - dimensional feature vector (such as 32 - dimensional), achieving "feature dimensionality reduction - noise filtering". The decoder in the auto - encoder reconstructs the low - dimensional electroencephalogram signal feature vector back to the original dimension, obtaining the reconstructed anomaly - free electroencephalogram signal data. The decoder is symmetric to the encoder and reconstructs the low - dimensional vector into the original - dimension signal through transposed convolution or fully - connected layers, using the Sigmoid activation function to ensure that the output value is within a reasonable range.

[0028] Based on the calculation of the mean squared error between the anomaly - free electroencephalogram signal data and the reconstructed anomaly - free electroencephalogram signal data, through the backpropagation algorithm, the value of the loss function is backpropagated to each parameter of the encoder and decoder, and the gradient descent method is used to adjust the parameters, making the mean squared error continuously decrease. When the mean squared error is less than the set mean squared error threshold, it is considered that the model has learned the feature distribution of normal electroencephalograms, and the training ends, outputting the trained encoder, where the parameters include weights and biases; Input the electroencephalogram signal data into the trained encoder, and after encoding and decoding operations, obtain the reconstructed electroencephalogram signal data to be verified. Calculate the mean squared error to be verified based on the reconstructed electroencephalogram signal data to be verified and the electroencephalogram signal data; if the mean squared error to be verified is greater than the set error threshold, then output the electroencephalogram assessment result as abnormal assessment result; if the mean squared error to be verified is not greater than the set error threshold, then output the electroencephalogram assessment result as normal assessment result.

[0029] The autoencoder automatically extracts the core features of normal EEG (such as frequency band energy ratio and inter-electrode synchronization) through unsupervised learning, and is sensitive to subtle changes that deviate from the normal pattern (such as a sudden increase of 15% in delta wave energy and an increase in alpha wave phase difference). It can detect early abnormalities that are difficult to detect with traditional manual analysis (such as local EEG slow wave enhancement within 2 hours after surgery), and warn several hours before the onset of clinical symptoms, thus gaining golden time for intervention.

[0030] The training data comes from normal EEG of patients undergoing similar surgeries. The model adaptively learns the normal baseline of a specific population (such as lower alpha wave frequency in elderly patients and more active beta waves in young patients) to avoid "one-size-fits-all" standard misjudgment. The EEG signal risk assessment unit based on the autoencoder solves the problems of "strong manual dependence, poor timeliness, and insufficient individual adaptability" of traditional EEG monitoring through data-driven intelligent analysis, providing accurate and efficient technical support for neurological function monitoring of postoperative neurosurgery patients, significantly improving the ability to identify risks in the early stage, and helping to achieve the clinical goal of "early detection-early intervention-improved prognosis".

[0031] The physiological indicator risk assessment unit is used to obtain a physiological indicator assessment coefficient based on the physiological indicator data, and compare the physiological indicator assessment coefficient with the physiological indicator assessment threshold stored in the database: if the physiological indicator assessment coefficient is greater than the physiological indicator assessment threshold stored in the database, then the physiological indicator assessment result is output as an abnormal assessment result; if the physiological indicator assessment coefficient is not greater than the physiological indicator assessment threshold stored in the database, then the physiological indicator assessment result is output as a normal assessment result; Basic vital signs parameters include heart rate ,blood pressure , respiratory rate and body temperature ; Neurosurgery-specific parameters include intracranial pressure and blood oxygen saturation The process of obtaining the physiological indicator evaluation coefficient based on the physiological indicator data is as follows: obtain the patient's basic pathological information, including age, gender, weight, height and basic disease information; obtain the basic case information of historical patients who have undergone the same neurosurgery operation stored in the case database (which stores the basic pathological information and postoperative physiological indicator data of patients undergoing the same neurosurgery operation (such as brain tumor resection, cerebral hemorrhage operation) to form a multi-dimensional feature space), analyze the similarity between the patient's basic pathological information and the basic case information of historical patients based on cosine similarity, and determine the basic case information of historical patients that is most similar to the patient's basic pathological information.

[0032] The basic information of historical patient cases includes basic physical sign parameters of historical patients and special parameters of historical patients in neurosurgery. Basic physical sign parameters of historical patients include heart rate ,blood pressure , respiratory rate and body temperature ; The historical patient's neurosurgery-specific parameters include intracranial pressure and blood oxygen saturation .

[0033] Based on the basic information of the most similar historical patient cases, obtain the corresponding postoperative physiological index record data stored in the database, including the postoperative physiological index data for each day after surgery; perform a difference process on the physiological index data of the current day and the corresponding postoperative physiological index data in the patient's postoperative physiological index record data (align the physiological index data of the current patient on the nth day after surgery (such as heart rate 85 beats / min, intracranial pressure 150 mmHg) with the data of the nth day after surgery of the matched historical case (such as historical average heart rate 78 beats / min, intracranial pressure 130 mmHg) in the time dimension), and obtain the physiological index evaluation coefficient.

[0034] Among them, the physiological index evaluation coefficient has the following calculation formula: ; and are both the first transfer functions, is the weight factor of the function value of, is the weight factor of the function value of. Taking the physiological index evaluation scenario as an example, by calculating based on the relative differences from historical / standard values for different physiological parameters, the deviation degree of individual physiological data can be accurately captured; Using the form of Euclidean distance to quantify the differences of multi-dimensional data such as neurosurgery-specific parameters; and then through weighted integration, the risk of physiological indexes can be measured personalized, converting the vague "data abnormality" into a computable and comparable quantitative value, making the risk characteristics of complex physiological / signal data obvious, breaking through the traditional "one-size-fits-all" standard, fitting the individual pathological characteristics, improving the evaluation accuracy, and reducing misjudgment and missed judgment.

[0035] The physiological index risk assessment unit solves the defects of the traditional physiological monitoring of "one-size-fits-all" and "static threshold" through personalized baseline construction and dynamic difference analysis, realizes the accurate characterization of the physiological state of neurosurgical postoperative patients, provides a scientific basis for early detection of key risks such as circulatory disorders and abnormal intracranial pressure, helps medical staff formulate individualized nursing plans, and improves the refinement level of postoperative monitoring.

[0036] The behavioral risk assessment unit is used to obtain a behavioral assessment coefficient based on behavioral data, and compare the behavioral assessment coefficient with the behavioral assessment threshold stored in the database: if the behavioral assessment coefficient is greater than the behavioral assessment threshold stored in the database, the output behavioral assessment result is that the assessment result is abnormal; if the behavioral assessment coefficient is not greater than the behavioral assessment threshold stored in the database, the output behavioral assessment result is that the assessment result is normal.

[0037] The process of obtaining the behavioral assessment coefficient based on behavioral data is as follows: obtain the postoperative behavioral data of patients undergoing the same type of neurosurgery; perform standardized processing on the behavioral data and the postoperative behavioral data, and then perform cosine similarity value analysis to obtain the behavioral assessment coefficient.

[0038] The process of obtaining the assessment coefficient based on behavioral data starts from accurately identifying individual behavioral differences, and plays a role in assisting accurate judgment, supporting personalized care, and improving comprehensive assessment in the risk assessment and intervention of neurosurgical postoperative care, which helps to improve the quality of postoperative care and the rehabilitation effect of patients.

[0039] The inflammation risk assessment unit is used to obtain an inflammation assessment coefficient based on inflammation data, and compare the inflammation assessment coefficient with the inflammation assessment threshold stored in the database: if the inflammation assessment coefficient is greater than the inflammation assessment threshold stored in the database, the output inflammation assessment result is that the assessment result is abnormal; if the inflammation assessment coefficient is not greater than the inflammation assessment threshold stored in the database, the output inflammation assessment result is that the assessment result is normal.

[0040] Blood inflammation parameters include the content of various white blood cells , the content of C-reactive protein and the content of procalcitonin , where i is the type of white blood cell; inflammation mediator parameters include the content of interleukin-6 and the content of tumor necrosis factor-α ; as Figure 2 shown, the process of obtaining the inflammation assessment coefficient based on inflammation data is as follows: obtain the normal human inflammation data, including normal human blood inflammation parameters and normal human inflammation mediator parameters; normal human blood inflammation parameters include the normal content of various white blood cells , the normal content of C-reactive protein and the normal content of procalcitonin ; normal human inflammation mediator parameters include the normal content of interleukin-6 and the normal content of tumor necrosis factor-α .

[0041] The normal human inflammation data is sourced from a large-scale medical examination database of healthy individuals. Through long-term tracking and testing of people of different ages, genders, and constitutions, the normal reference ranges of various indicators such as various white blood cells (such as neutrophils, lymphocytes, etc., corresponding to different i), C-reactive protein, procalcitonin, interleukin-6, tumor necrosis factor-α, etc. are determined through statistical analysis, serving as the basic yardstick for judging whether inflammation is abnormal. For example, the normal content of neutrophils in healthy adults is usually within a certain range, and this data is statistically obtained based on the test results of a large number of physical examination populations, which can reflect the steady state of inflammation indicators when the human body is not stimulated by pathology.

[0042] Obtain the postoperative inflammation data of patients undergoing the same type of neurosurgery, including postoperative blood inflammation parameters and postoperative inflammatory mediator parameters; the postoperative blood inflammation parameters include the postoperative content of various white blood cells, the postoperative content of C-reactive protein, and the postoperative content of procalcitonin; the postoperative inflammatory mediator parameters include the postoperative content of interleukin-6 and the postoperative content of tumor necrosis factor-α.

[0043] The postoperative inflammation data of patients undergoing the same type of neurosurgery is screened from the hospital case system for historical cases that match the surgical type of the target patient (such as the same surgical procedures like brain tumor resection, cerebral hemorrhage clearance, etc.), the degree of surgical trauma (such as the size of the bone window in craniotomy, the impact of the operation path in minimally invasive surgery), and the patient's basic conditions (similar in age, whether there are underlying diseases, etc.), and the blood inflammation parameters and inflammatory mediator parameters of these cases are extracted. These data reflect the changing rules of postoperative inflammation indicators in patients undergoing the same type of surgery and can be used as a reference for evaluating whether the postoperative inflammation of the target patient is abnormal.

[0044] Based on the normal human inflammation data and the inflammation data, obtain the first inflammation assessment factor; based on the postoperative inflammation data of patients undergoing the same type of neurosurgery and the inflammation data, obtain the second inflammation assessment factor; weight and sum the first inflammation assessment factor and the second inflammation assessment factor to obtain the inflammation assessment coefficient.

[0045] Taking the first inflammation assessment factor as an example, its calculation formula is: ; and are both the second transfer functions, is the weight factor of the function value of is the weight factor of the function value of

[0046] The formula eliminates the influence of dimensions in the form of relative differences and unifies the calculation logic of different types of data; at the same time, it covers a variety of indicators (from conventional physiology to special postoperative indicators) to adapt to the complex clinical data environment.

[0047] The form of the second inflammation assessment factor is the same as that of the first inflammation assessment factor.

[0048] A comprehensive risk assessment module, which is used to fuse various index data to obtain a trained comprehensive risk assessment model when the results of all risk assessments are normal assessment results, and process the fused various index data based on the trained comprehensive risk assessment model to obtain a comprehensive risk assessment result, including an abnormal comprehensive assessment result and a normal comprehensive assessment result; The comprehensive risk assessment model is a gradient boosting decision tree model, a gradient boosting decision tree (GBDT) model composed of multiple decision trees (such as 100 trees, 200 trees). When each decision tree is trained, based on the comprehensive index vector of historical patients and the corresponding risk label (normal / abnormal), it learns the mapping relationship between features and risks. When a new comprehensive index vector is input, each decision tree judges the vector according to the internal splitting rule (such as selecting a feature splitting point based on the Gini coefficient, mean squared error, etc.), and outputs a prediction value (this value can be understood as the probability contribution of the decision tree that the input vector belongs to the "normal" category). For example, a certain decision tree judges that there is a certain abnormal risk based on the excessive proportion of δ-wave energy in the electroencephalogram feature, and outputs a prediction value of 0.3 (representing that the probability of belonging to the normal for this tree is 0.3).

[0049] The process of obtaining the comprehensive risk assessment result based on the comprehensive risk assessment model is as follows: perform dimensionless processing (normalization or standardization processing) on various index data and splice them into a comprehensive index vector; the splicing logic is to sort according to the index category or importance, and sequentially splice the dimensionless processed index data. For example, first splice the feature vector after processing the electroencephalogram signal, and then splice the physiological index vector, behavior index vector, and inflammation index vector to form a comprehensive index vector with a unified dimension.

[0050] Input the comprehensive index vector into the comprehensive risk assessment model to obtain the prediction values output by each decision tree, accumulate the prediction values according to the set weights to obtain the probability that the comprehensive index vector belongs to the normal; if the probability is greater than the set normal probability threshold, output the comprehensive risk assessment result as a normal comprehensive assessment result; if the probability is not greater than the set normal probability threshold, output the comprehensive risk assessment result as an abnormal comprehensive assessment result.

[0051] The risks after neurosurgery are synergistically affected by multiple factors such as electroencephalogram, physiology, behavior, and inflammation. The normalcy of a single indicator does not mean that the overall risk is controllable. The comprehensive risk assessment module integrates multi-dimensional indicators and uses the GBDT model to explore the complex correlations between indicators (such as the synergistic risk of abnormal brain waves and elevated inflammation indicators). For example, the electroencephalogram assessment is normal and the physiological indicators are normal, but the behavior indicators show that the patient's activity endurance continues to decline and the inflammation indicators have a potential upward trend. Through the comprehensive model, this "hidden" risk can be identified, avoiding the neglect of the overall risk due to the normalcy of a single indicator and improving the comprehensiveness of risk assessment. Through the application of multi-dimensional data fusion and the gradient boosting decision tree model, the problem of one-sidedness in single-indicator assessment after neurosurgery is solved, achieving the accurate quantitative assessment of comprehensive risks, providing strong support for medical staff to formulate personalized and precise postoperative care and intervention plans, and helping to improve the quality of patients' postoperative recovery and the effect of risk control.

[0052] An early warning module is used to obtain the results of various risk assessments and the comprehensive risk assessment results, and issue early warnings based on the results of various risk assessments and the comprehensive risk assessment results.

[0053] The process of issuing early warnings based on the results of various risk assessments and the comprehensive risk assessment results is as follows: Count the number of abnormal assessment results obtained from the sub-item risk assessment. If the number is no more than two, a first-level early warning is issued; if the number is more than two, a second-level early warning is issued; if an abnormal comprehensive assessment result is received, a second-level early warning is directly issued.

[0054] The first-level early warning indicates that there are local and relatively independent risk points after the patient's surgery, and multi-dimensional synergistic risks have not yet formed. For example, only the electroencephalogram signal is abnormal (possibly transient brain function fluctuations), or only the physiological indicators are abnormal (such as a single heart rate fluctuation). Although such situations need attention, the risk diffusivity and urgency are relatively low. When the system triggers a first-level early warning, it can remind medical staff to conduct targeted inspections through methods such as the yellow flashing of the ward warning light and the pop-up prompt on the nurse station computer (marking the specific abnormal sub-items).

[0055] The second-level early warning includes two triggering situations. One is that the number of abnormal sub-item risk assessments > 2, indicating that the patient has risks in multiple dimensions such as electroencephalogram, physiology, behavior, and inflammation after surgery, with strong risk synergy and possibly triggering serious postoperative complications (such as the superposition of multiple organ function abnormalities); the other is that the comprehensive risk assessment result is abnormal. At this time, even if the number of abnormal sub-items is small, the model judges the overall risk to be high through multi-dimensional data fusion (such as the abnormal signs of each indicator are not obvious when viewed separately, but potential serious problems are indicated under the synergistic effect). When the second-level early warning is triggered, strong reminder methods such as the red constant lighting of the ward warning light, the sound and light alarm at the nurse station, and the push of an emergency message to the duty mobile phone are used to urge medical staff to respond immediately and carry out multi-disciplinary collaborative assessment and intervention.

[0056] An electronic device, comprising: a processor; and a memory in which computer program instructions are stored, and when the computer program instructions are run by the processor, the processor is caused to execute the postoperative care risk warning system for neurosurgical patients as described above.

[0057] A computer-readable storage medium for storing a program which, when executed by a processor, implements the postoperative care risk warning system for neurosurgical patients as described above.

[0058] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program code.

[0059] The present invention is described with reference to the flowcharts and / or block diagrams of systems, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0060] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0061] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are performed on the computer or other programmable device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0062] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made by those skilled in the art once they learn the basic creative concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications that fall within the scope of the present invention.

[0063] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A postoperative nursing risk warning system for neurosurgical patients, characterized in that, It includes a data acquisition module, a sub-item risk assessment module, a comprehensive risk assessment module, and a warning module, where: The data acquisition module is used to acquire various index data of neurosurgical patients after surgery, including electroencephalogram signal data, physiological index data, behavior data, and inflammation data; The sub-item risk assessment module is used to conduct sub-item risk assessment on various index data to obtain various risk assessment results, including electroencephalogram assessment results, physiological index assessment results, behavior assessment results, and inflammation assessment results. Each risk assessment result includes abnormal assessment result and normal assessment result; The comprehensive risk assessment module is used to fuse various index data when all risk assessment results are normal assessment results, obtain a trained comprehensive risk assessment model, and process the fused various index data based on the trained comprehensive risk assessment model to obtain a comprehensive risk assessment result, including abnormal comprehensive assessment result and normal comprehensive assessment result; The warning module is used to obtain various risk assessment results and comprehensive risk assessment results, and issue a warning according to various risk assessment results and comprehensive risk assessment results.

2. The postoperative care risk early warning system for neurosurgical patients according to claim 1, characterized in that, The data acquisition module includes an electroencephalogram signal acquisition unit, a physiological index acquisition unit, a behavior data acquisition unit, and an inflammation data acquisition unit, where: The electroencephalogram signal acquisition unit is used to connect with a high-density electroencephalogram device, acquire initial electroencephalogram signals at multiple electrode sites of the whole brain, and process the initial electroencephalogram signals based on independent component analysis and band-pass filtering to obtain electroencephalogram signal data; The physiological index acquisition unit is used to acquire basic physiological indexes of neurosurgical patients after surgery, preprocess the basic physiological indexes and then perform standardization processing to obtain physiological index data; The behavior data acquisition unit is used to acquire patient action data based on a camera, identify the patient's behavior based on an object detection algorithm, determine the patient's behavior, and at the same time acquire behavior data based on an inertial measurement sensor; The inflammation data acquisition unit is used to acquire the postoperative clinical examination results of the patient, extract initial inflammation data, and perform standardization processing on the initial inflammation data to obtain inflammation data.

3. The postoperative nursing risk warning system for neurosurgical patients according to claim 2, wherein: The electroencephalogram signal data includes electroencephalogram frequency band parameters, electroencephalogram time domain parameters, and electroencephalogram spatial parameters; The physiological index data includes basic physical sign parameters and special neurosurgery parameters; The behavior data includes the acceleration during movement and the angular velocity during movement corresponding to each patient's behavior; The inflammation data includes blood inflammation parameters and inflammatory mediator parameters.

4. The postoperative nursing risk warning system for neurosurgical patients according to claim 3, characterized in that: The sub-item risk assessment module includes an electroencephalogram signal risk assessment unit, a physiological index risk assessment unit, a behavior risk assessment unit, and an inflammation risk assessment unit, where: The electroencephalogram signal risk assessment unit is used to identify abnormalities in electroencephalogram signal data based on an autoencoder and output an electroencephalogram assessment result; The physiological index risk assessment unit is used to obtain a physiological index assessment coefficient based on physiological index data and compare the physiological index assessment coefficient with the physiological index assessment threshold stored in the database: If the physiological index assessment coefficient is greater than the physiological index assessment threshold stored in the database, the output physiological index assessment result is an abnormal assessment result; If the physiological index evaluation coefficient is not greater than the physiological index evaluation threshold stored in the database, the output physiological index evaluation result is that the evaluation result is normal; The behavior risk assessment unit is used to obtain a behavior evaluation coefficient based on the behavior data and compare the behavior evaluation coefficient with the behavior evaluation threshold stored in the database: If the behavior evaluation coefficient is greater than the behavior evaluation threshold stored in the database, the output behavior evaluation result is that the evaluation result is abnormal; If the behavior evaluation coefficient is not greater than the behavior evaluation threshold stored in the database, the output behavior evaluation result is that the evaluation result is normal; The inflammation risk assessment unit is used to obtain an inflammation evaluation coefficient based on the inflammation data and compare the inflammation evaluation coefficient with the inflammation evaluation threshold stored in the database: If the inflammation evaluation coefficient is greater than the inflammation evaluation threshold stored in the database, the output inflammation evaluation result is that the evaluation result is abnormal; If the inflammation evaluation coefficient is not greater than the inflammation evaluation threshold stored in the database, the output inflammation evaluation result is that the evaluation result is normal.

5. The postoperative nursing risk early warning system for neurosurgical patients according to claim 4, characterized in that, The process of the autoencoder for abnormal recognition of electroencephalogram signal data is as follows: Obtain the electroencephalogram signal data without abnormalities regularly updated in the database; Train the autoencoder based on the electroencephalogram signal data without abnormalities. The process is as follows: The encoder in the autoencoder compresses the input electroencephalogram signal data without abnormalities into a low-dimensional electroencephalogram signal feature vector, and the decoder in the autoencoder reconstructs the low-dimensional electroencephalogram signal feature vector back to the original dimension to obtain the reconstructed electroencephalogram signal data without abnormalities; Based on the calculation of the mean square error between the electroencephalogram signal data without abnormalities and the reconstructed electroencephalogram signal data without abnormalities, through the backpropagation algorithm, the value of the loss function is backpropagated to each parameter of the encoder and decoder, and the gradient descent method is used to adjust the parameters to continuously reduce the mean square error. When the mean square error is less than the set mean square error threshold, the training ends, and the trained encoder is output, where the parameters include weights and biases; Input the electroencephalogram signal data into the trained encoder, and after encoding and decoding operations, obtain the reconstructed electroencephalogram signal data to be verified. Calculate the mean square error to be verified based on the reconstructed electroencephalogram signal data to be verified and the electroencephalogram signal data; If the mean square error to be verified is greater than the set error threshold, the output electroencephalogram evaluation result is that the evaluation result is abnormal; If the mean square error to be verified is not greater than the set error threshold, the output electroencephalogram evaluation result is that the evaluation result is normal.

6. The postoperative nursing risk early warning system for neurosurgical patients according to claim 4, characterized in that, The basic physical sign parameters include heart rate, blood pressure, respiratory rate, and body temperature; the special parameters for neurosurgery include intracranial pressure and blood oxygen saturation; The process of obtaining the physiological index evaluation coefficient based on the physiological index data is as follows: Obtain the patient's pathological basic information, including age, gender, weight, height, and basic disease information; Obtain the historical patient case basic information of patients who have undergone the same neurosurgery stored in the case database, and based on the cosine similarity, analyze the similarity between the patient's pathological basic information and the historical patient case basic information to determine the historical patient case basic information most similar to the patient's pathological basic information; Obtain the corresponding postoperative physiological index record data stored in the database based on the most similar historical patient case basic information, including the postoperative physiological index data for each day after surgery; Perform a difference operation on the physiological index data of the current day and the corresponding postoperative physiological index data in the postoperative physiological index record data of the patient to obtain a physiological index evaluation coefficient.

7. The postoperative nursing risk early warning system for neurosurgical patients according to claim 4, characterized in that, The process of obtaining a behavior evaluation coefficient based on behavior data is as follows: Obtain the postoperative behavior data of patients undergoing the same type of neurosurgery; Perform a standardization process on the behavior data and the postoperative behavior data, and then conduct a cosine similarity value analysis to obtain a behavior evaluation coefficient.

8. The postoperative care risk early warning system for neurosurgical patients according to claim 4, wherein Blood inflammation parameters include the contents of various white blood cells, C-reactive protein, and procalcitonin; inflammation mediator parameters include the contents of interleukin-6 and tumor necrosis factor-α; The process of obtaining an inflammation evaluation coefficient based on inflammation data is as follows: Obtain normal human inflammation data, including normal human blood inflammation parameters and normal human inflammation mediator parameters; Obtain the postoperative inflammation data of patients undergoing the same type of neurosurgery, including postoperative blood inflammation parameters and postoperative inflammation mediator parameters; Obtain a first inflammation evaluation factor based on normal human inflammation data and inflammation data; Obtain a second inflammation evaluation factor based on the postoperative inflammation data of patients undergoing the same type of neurosurgery and inflammation data; Perform a weighted sum of the first inflammation evaluation factor and the second inflammation evaluation factor to obtain an inflammation evaluation coefficient.

9. The postoperative nursing risk early warning system for neurosurgical patients according to claim 2, wherein The comprehensive risk assessment model is a gradient boosting decision tree model. The process of obtaining a comprehensive risk assessment result based on the comprehensive risk assessment model is as follows: Perform a dimensionless process on each index data and splice it into a comprehensive index vector; Input the comprehensive index vector into the comprehensive risk assessment model to obtain the predicted values output by each decision tree. Accumulate the predicted values according to the set weights to obtain the probability that the comprehensive index vector belongs to the normal state; If the probability is greater than the set normal probability threshold, output the comprehensive risk assessment result as the comprehensive assessment result is normal; If the probability is not greater than the set normal probability threshold, output the comprehensive risk assessment result as the comprehensive assessment result is abnormal.

10. The postoperative nursing risk early warning system for neurosurgical patients according to claim 1, characterized in that, The process of giving an early warning based on each risk assessment result and the comprehensive risk assessment result is as follows: Count the number of abnormal evaluation results obtained from the sub-item risk assessment. If the number is not greater than two, give a first-level early warning. If the number is greater than two, give a second-level early warning; If the comprehensive assessment result of abnormality is received, directly give a second-level early warning.

Images