Neurology clinical nursing potential safety hazard analysis method and system
By generating low-dimensional feature data and real-time update of risk score models, the problems of data isolation and static rules in traditional methods are solved, and the accurate assessment of safety hazards of patients with neurology and personalized nursing suggestions are achieved, which improves nursing safety and efficiency.
Patent Information
- Application Number
- CN202510015699.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-06
- Publication Date
- 2025-04-18
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional neurology clinical nursing safety hazard analysis methods rely on manual monitoring and static rules, and cannot effectively integrate multi-source data, making it difficult to comprehensively evaluate patient safety hazards, and lack a real-time dynamic update mechanism, so it is impossible to respond to changes in patient conditions in a timely manner.
By collecting and preprocessing patients' multi-source data, a high-quality low-dimensional feature data set is generated using the generative adversarial network, and a safety hazard monitoring model is formed to monitor safety hazards. Update each patient's comprehensive risk score in real time based on the monitoring results, and set alert thresholds based on the comprehensive risk score to provide personalized clinical care recommendations.
It realizes a comprehensive and accurate assessment of patient safety hazards, overcomes the limitations of data isolation processing, improves the sensitivity and accuracy of monitoring, provides personalized nursing suggestions, improves the accuracy of nursing decisions, and reduces the possibility of false alarms and missed alarms.
Smart Images
Figure CN119851899A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of medical information technology, and in particular to a method and system for analyzing safety hazards in clinical nursing of neurology. Background Art
[0002] Traditional methods for analyzing safety hazards in clinical nursing in neurology rely on manual monitoring and empirical judgment, which has many limitations. Multi-source data of patients (such as medical history, medication, physiological signs, etc.) are usually collected and analyzed in isolation, and various types of information cannot be effectively integrated, making it difficult for nursing staff to fully grasp the safety hazards of patients.
[0003] Most existing safety hazard monitoring methods rely on manually set thresholds and static rules, lacking a real-time dynamic update mechanism, resulting in an inability to respond to changes in patient conditions in a timely manner. Secondly, traditional methods have limited capabilities for data processing and feature extraction, making it difficult to mine potential safety hazard features from complex multidimensional data, especially in high-dimensional data analysis and dimensionality reduction. Existing methods often neglect the formulation of personalized care plans and fail to provide personalized clinical recommendations based on the characteristics and real-time status of each patient. Therefore, there is an urgent need for a new intelligent method that can make full use of multi-source data, dynamically update patient risk scores, and provide personalized and accurate care recommendations based on monitoring results to improve the safety and efficiency of neurology care. Summary of the invention
[0004] In view of the above existing problems, the present invention is proposed.
[0005] Therefore, the present invention provides a method for analyzing safety hazards in clinical nursing of neurology to solve the problem that traditional methods lack dynamic risk assessment and personalized nursing suggestions.
[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides a method for analyzing safety hazards in clinical nursing of neurology, which comprises collecting multi-source data of patients and performing preprocessing; Perform feature extraction and dimensionality reduction on the preprocessed multi-source data to obtain low-dimensional safety hazard features; Generate high-quality low-dimensional feature data sets using generative adversarial networks and form a safety hazard monitoring model to monitor safety hazards; Update each patient's comprehensive risk score in real time based on monitoring results; Set an alarm threshold based on the comprehensive risk score and compare it with the alarm threshold to obtain alarm information; Respond to alert messages and provide personalized clinical care recommendations.
[0007] As a preferred solution of the method for analyzing safety hazards in clinical nursing of neurology department of the present invention, wherein: the multi-source data includes the medical history, medication data, physiological sign data, surrounding environment parameters and limb acceleration data of neurology patients; The physiological characteristic data include heart rate, blood pressure, body temperature, muscle tremor amplitude and blood oxygen concentration; The ambient environment parameters include temperature and humidity, noise intensity and light intensity; The limb acceleration data is the movement data of the patient's limbs collected by the wearable device; The preprocessing includes low-pass filtering, band-pass filtering and data normalization.
[0008] As a preferred solution of the method for analyzing safety hazards in clinical nursing of neurology department of the present invention, the following steps are performed: feature extraction is performed on the pre-processed multi-source data to obtain safety hazard features, and dimension reduction is performed to obtain low-dimensional safety hazard features. Use principal component analysis to extract features from preprocessed multi-source data to obtain streamlined features that reduce data redundancy; The simplified features are reduced in dimension through t-SNE, expressed as: ; in, For point Selecting points in low-dimensional space The probability of being its neighbor, For point The coordinates in low-dimensional space, For point The coordinates in low-dimensional space, For the point-to-preceding point, For point-to-point postposition, is the index variable that refers to the specific point. is the index variable that refers to a specific point. is the index variable used to traverse all points to the previous point, is the index variable used to traverse all the post-points of the point pair; The low-dimensional safety hazard features after dimensionality reduction are saved in time series format.
[0009] As a preferred solution of the method for analyzing safety hazards in clinical nursing of neurology of the present invention, a generative adversarial network is used to generate a high-quality low-dimensional feature data set, and a safety hazard monitoring model is formed to monitor safety hazards. The specific steps are as follows: Program according to the format of low-dimensional feature data to build the generator and discriminator of the generative adversarial network; Obtain a random noise vector through programming, input the random noise vector into a generator, and generate a realistic low-dimensional safety hazard feature; The cross entropy loss function is used as the optimization target of the discriminator, and the expression is: ; in, is the cross entropy loss value, is the distribution of low-dimensional safety hazard characteristics, is a low-dimensional security risk feature, is the distribution of realistic low-dimensional safety hazard features, is a random noise vector, is the interpolation data, is the interpolation sample, is the index variable that refers to the specific point. is the index variable that refers to the specific point later. is the expectation of the distribution of low-dimensional safety hazard features, is the probability of the discriminator judging the low-dimensional safety hazard features, is the expected value of the distribution of realistic low-dimensional safety hazard features, The realistic low-dimensional security risk features generated by the generator, is the gradient penalty weight, is the expected value of the distribution of the interpolated data, is the gradient of the interpolated sample, is the probability of the interpolation sample, is the contribution weight, For point Selecting points in low-dimensional space The probability of being its neighbor, Neighborhood relationship for Gaussian distribution; The simulated low-dimensional safety hazard features are mixed with the low-dimensional safety hazard features to obtain a mixed feature data set; According to Bayes' theorem, the posterior probability of mixed feature data is calculated, and the expression is: ; in, is the posterior probability value, is the abnormal status value, is mixed feature data, is the likelihood function, is the prior probability value, is the marginal likelihood value; The joint likelihood method is used to evaluate all mixed features comprehensively, and the ranking results are statistically consistent with the assumption of uniform distribution. The SBC method is verified and the verified mixed feature data set is obtained. The Bayesian anomaly detection model is trained using the verified mixed feature dataset and monitored to obtain anomaly scores and potential risk labels.
[0010] As a preferred solution of the method for analyzing safety hazards in clinical nursing of neurology of the present invention, the comprehensive risk score of each patient is updated in real time according to the monitored safety hazards. The specific steps are as follows: A baseline composite risk score was set for each patient using the patient's initial physiological sign data; Receive the anomaly score and hazard label output by the safety hazard monitoring model, and combine them with real-time physiological feature data as input for updating the comprehensive risk score; The baseline comprehensive risk score and long short-term memory network are used to perform time series modeling on the patient's comprehensive risk score, and the patient's comprehensive risk score is updated in real time.
[0011] As a preferred solution of the method for analyzing safety hazards in clinical nursing of neurology of the present invention, wherein: an alarm threshold is set for the comprehensive risk score according to the physiological sign data in the multi-source data, and the comprehensive risk score is compared with the alarm threshold to obtain alarm information. The specific steps are as follows: Using the individual characteristics of the patient to set a personalized initial alarm threshold range; Use historical data to calculate dynamic baseline thresholds, set normal ranges for heart rate and blood pressure indicators through statistical analysis, and dynamically adjust alarm thresholds based on real-time monitored physiological sign data; The composite risk scores were compared using dynamic baseline thresholds to obtain comparative results; An alarm is issued based on the comparison result, and alarm information is obtained.
[0012] As a preferred solution of the method for analyzing safety hazards in clinical nursing of neurology of the present invention, the specific steps of responding to the alarm information and providing clinical nursing suggestions are as follows: Receive and analyze alert information to identify specific sources of risk; utilizes medical history and medications to generate initial care recommendations; Optimize care recommendations based on individual patient characteristics and real-time feedback; The nursing recommendations are presented to nursing staff in a visual form, and implementation guidance is provided, while the implementation results of the nursing recommendations are recorded.
[0013] In a second aspect, the present invention provides a neurology clinical nursing safety hazard analysis system, including a data acquisition module, a low-dimensional feature module, a hazard monitoring module, a risk scoring module, an alarm module, and an alarm response module; The data acquisition module is used to collect multi-source data of patients and perform pre-processing; The low-dimensional feature module is used to extract features and reduce the dimension of pre-processed multi-source data to obtain low-dimensional safety hazard features; The hidden danger monitoring module is used to generate a high-quality low-dimensional feature data set using a generative adversarial network and form a safety hidden danger monitoring model to monitor safety hidden dangers; The risk scoring module is used to update the comprehensive risk score of each patient in real time according to the monitoring results; The alarm module is used to set an alarm threshold according to the comprehensive risk score and compare it with the alarm threshold to obtain alarm information; The alarm response module is used to respond to the alarm information and provide personalized clinical care suggestions.
[0014] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the method for analyzing safety hazards in clinical nursing in neurology as described in the first aspect of the present invention is implemented.
[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the method for analyzing safety hazards in clinical nursing in neurology as described in the first aspect of the present invention is implemented.
[0016] The beneficial effects of the present invention are as follows: through the fusion and intelligent analysis of multi-source data, the patient's safety hazards can be comprehensively and accurately evaluated, overcoming the limitations of isolated data processing in traditional methods. Generative adversarial networks are used to generate high-quality low-dimensional feature data, and combined with a dynamically updated risk scoring model, safety hazard monitoring can reflect patient changes in real time, improving the sensitivity and accuracy of monitoring. Based on real-time data and individual characteristics, this method can provide personalized nursing advice for each patient, effectively avoiding a "one-size-fits-all" nursing plan and improving the accuracy of nursing decisions. The use of dynamic baseline thresholds and statistical analysis methods to set alarm thresholds makes risk assessment more scientific and reasonable, reducing the possibility of false alarms and missed alarms. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use 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 ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.
[0018] Figure 1 This is a flow chart of the method for analyzing safety hazards in clinical nursing of neurology in Example 1.
[0019] Figure 2 This is a module diagram of the neurology clinical nursing safety hazard analysis system in Example 1. DETAILED DESCRIPTION
[0020] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.
[0021] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.
[0023] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a method for analyzing safety hazards in clinical nursing of neurology, comprising the following steps: S1. Collect multi-source data of patients and perform preprocessing, including the following steps: The multi-source data collected from patients include the medical history, medication data, physiological sign data, ambient environment parameters, and limb acceleration data of neurology patients. Specifically, medical history and medication data are obtained through electronic health records (EHR). Physiological sign data are collected through electronic medical devices. Surrounding environment parameters are collected through sensors. Accelerometers or other sensors are used to measure the acceleration data of patients' limbs.
[0024] Physiological characteristic data include heart rate, blood pressure, body temperature, muscle tremor amplitude and blood oxygen concentration. Ambient parameters include temperature and humidity, noise intensity and light intensity. Limb acceleration data is the movement data of the patient's limbs collected by wearable devices.
[0025] Preprocessing includes low-pass filtering, band-pass filtering and data normalization.
[0026] Specifically, low-pass filtering is used to remove high-frequency noise or fluctuations to ensure that changes in the data mainly reflect the basic trend of the signal rather than sudden short-term interference. For example, heart rate data may experience short-term fluctuations caused by factors such as exercise and emotions. Low-pass filtering can help eliminate these interferences and retain relatively stable trend data. Bandpass filtering retains signals within a specific frequency range and removes unnecessary low-frequency and high-frequency noise. For example, in the monitoring of muscle tremor amplitude, bandpass filtering can eliminate irrelevant signals caused by external interference (such as equipment vibration or non-physiological fluctuations) and retain the main frequency information of the patient's body tremor. Data standardization may cause some data to dominate the model training according to the different dimensions of different types of data (such as heart rate, blood pressure, temperature, etc.). Data standardization can unify the range of all data into a standard interval (for example, between 0 and 1), ensuring that different types of features have equal weight in analysis and modeling. Common standardization methods include Z-score standardization (by subtracting the mean and dividing by the standard deviation) and minimum-maximum standardization (linearly mapping the data to a range of 0 to 1). This can effectively reduce the deviation caused by differences in data scale.
[0027] It should be noted that through reasonable multi-source data collection and preprocessing, a solid data foundation can be provided for subsequent health monitoring, disease warning, and treatment optimization, effectively supporting personalized medical services in the field of neurology.
[0028] S2. Extract features and reduce dimension of preprocessed multi-source data to obtain low-dimensional safety hazard features, including the following steps: Principal component analysis is used to extract features from preprocessed multi-source data to obtain streamlined features that reduce data redundancy. Specifically, the covariance matrix is calculated for the standardized data to describe the correlation between the features. The eigenvalue decomposition of the covariance matrix is performed to obtain the eigenvectors and corresponding eigenvalues. The eigenvector represents the direction of the principal component, and the eigenvalue represents the variance of each principal component. The first few principal components are selected according to the size of the eigenvalue. Usually, the principal components that explain more than 90% of the total variance are selected as streamlined features. By projecting the original data onto the selected principal components, low-dimensional streamlined features are obtained.
[0029] The simplified features are reduced in dimension through t-SNE, expressed as: ; in, For point Selecting points in low-dimensional space The probability of being its neighbor, For point The coordinates in low-dimensional space, For point The coordinates in low-dimensional space, For the point-to-preceding point, For point-to-point postposition, is the index variable that refers to the specific point. is the index variable that refers to a specific point. is the index variable used to traverse all points to the previous point, It is the index variable used to traverse all the subsequent points of the point pair.
[0030] The low-dimensional safety hazard features after dimensionality reduction are saved in time series format. Specifically, the low-dimensional safety hazard features (extracted by PCA and t-SNE) at each moment are saved in chronological order to form time series data. Time series data can be sampled at time granularities such as minutes and hours, depending on the frequency of data updates and the sophistication of patient monitoring. Standardized formats (such as CSV, HDF5, etc.) can be used for saving to facilitate subsequent analysis, visualization, and modeling. The feature vector after dimensionality reduction (such as the low-dimensional coordinates obtained by t-SNE) is stored for each time point, as well as the corresponding metadata such as timestamp and patient ID.
[0031] It should be noted that saving the reduced-dimensional features in a time series format can help analyze the dynamic risk changes of patients, monitor the real-time evolution of safety hazards, and support the construction of prediction and early warning systems.
[0032] S3. Generate a high-quality low-dimensional feature data set using a generative adversarial network and form a safety hazard monitoring model to monitor safety hazards, including the following steps: According to the format of low-dimensional feature data, the generator and discriminator of the generative adversarial network are programmed. Specifically, the generative adversarial network (GAN) consists of two main components, including the generator and the discriminator. They continuously optimize their respective performance through adversarial training.
[0033] A random noise vector is obtained through programming, and the random noise vector is input into the generator to generate a realistic low-dimensional safety hazard feature. Specifically, random sampling is performed from a uniform distribution or a Gaussian distribution. The generator receives this noise vector as input and converts it into a low-dimensional safety hazard feature. The output result is a realistic low-dimensional safety hazard feature.
[0034] The cross entropy loss function is used as the optimization target of the discriminator, and the expression is: ; in, is the cross entropy loss value, is the distribution of low-dimensional safety hazard characteristics, is a low-dimensional security risk feature, is the distribution of realistic low-dimensional safety hazard features, is a random noise vector, is the interpolation data, is the interpolation sample, is the index variable that refers to the specific point. is the index variable that refers to the specific point later. is the expectation of the distribution of low-dimensional safety hazard features, is the probability of the discriminator judging the low-dimensional safety hazard features, is the expected value of the distribution of realistic low-dimensional safety hazard features, The realistic low-dimensional security risk features generated by the generator, is the gradient penalty weight, is the expected value of the distribution of the interpolated data, is the gradient of the interpolated sample, is the probability of the interpolation sample, is the contribution weight, For point Selecting points in low-dimensional space The probability of being its neighbor, is the neighborhood relationship of Gaussian distribution.
[0035] The simulated low-dimensional safety hazard features are mixed with the low-dimensional safety hazard features to obtain a mixed feature dataset. Specifically, by combining the generated samples with the real data, the diversity and representativeness of the mixed feature dataset are enhanced, so that the safety hazard monitoring model can better adapt to potential abnormal patterns, thereby improving the monitoring effect.
[0036] According to Bayes' theorem, the posterior probability of mixed feature data is calculated, and the expression is: ; in, is the posterior probability value, is the abnormal status value, is mixed feature data, is the likelihood function, is the prior probability value, is the marginal likelihood value.
[0037] Using the joint likelihood method, all mixed features are evaluated together to obtain the assumption that the ranking results are statistically consistent with the uniform distribution, complete the verification of the SBC method (Simulation-Based Calibration), and obtain the verified mixed feature data set. Specifically, the Bayesian theorem is applied to combine the prior probability and the likelihood function to calculate the posterior probability of the abnormal state given the data. Using Bayesian inference methods (such as Markov chain Monte Carlo method or variational Bayes method), multiple groups of parameters are sampled from the posterior distribution to form posterior samples. For possible repeated or discrete distributions of parameter values, small random perturbations are added to ensure the uniqueness of each sample and the accuracy of the ranking.
[0038] The Bayesian anomaly detection model is trained using the verified mixed feature data set, and monitoring is performed to obtain anomaly scores and hidden danger labels. Specifically, the Bayesian anomaly detection model is used to analyze the mixed feature data monitored in real time and calculate the anomaly score for each data point. According to the set anomaly score threshold, the anomaly score is converted into a specific hidden danger label to identify potential safety hazards and provide targeted suggestions and intervention measures for clinical care.
[0039] It should be noted that Bayesian reasoning is an unsupervised learning method that does not require a large amount of labeled data, which is especially important for safety hazard monitoring because labeled data may be very scarce.
[0040] S4. Update the comprehensive risk score of each patient in real time according to the monitoring results, including the following steps: The patient's initial physiological sign data is used to set a baseline comprehensive risk score for each patient. Specifically, a preliminary comprehensive risk score is calculated by analyzing the patient's initial physiological sign data, such as heart rate, blood pressure, body temperature, blood oxygen concentration, etc. This baseline comprehensive risk score can be determined in a variety of ways, such as based on a standard health risk assessment model, or through a combination of historical data and expert knowledge. The baseline comprehensive risk score usually takes into account the patient's personal background and medical history (such as age, gender, history of chronic diseases, etc.) to provide a reference point for subsequent risk assessments.
[0041] Receive the anomaly score and hazard label output by the safety hazard monitoring model, and combine them with real-time physiological characteristic data as input for comprehensive risk score update. Specifically, the safety hazard monitoring model (such as isolation forest, generative adversarial network, etc.) compares the real-time monitoring data with the predetermined safety standards and outputs anomaly scores and hazard labels. The anomaly score indicates the degree of abnormality of the current physiological data, while the hazard label identifies possible health risks (such as heart abnormalities, decreased blood oxygen, etc.). On this basis, the patient's real-time physiological characteristic data (such as current heart rate, blood pressure, etc.) is combined with the output of the safety hazard monitoring model and input into the risk score update mechanism to provide input for the next scoring cycle.
[0042] The patient's comprehensive risk score is modeled in time series using the baseline comprehensive risk score and the long short-term memory network, and the patient's comprehensive risk score is updated in real time. Specifically, the patient's risk score is modeled in time series using the long short-term memory network (LSTM). LSTM is a neural network structure that can capture long-term dependencies and is particularly suitable for processing and predicting time series data. In this step, LSTM is able to analyze the changing trend of the patient's comprehensive risk score, taking into account the impact of historical data, and updating the comprehensive risk score in real time. Whenever new monitoring data (such as new physiological signs, abnormal scores, or hidden danger labels) are input, the LSTM network dynamically adjusts the patient's comprehensive risk score based on these inputs. The updated comprehensive risk score can reflect the latest changes in the patient's health status and help the medical team adjust the intervention strategy in a timely manner.
[0043] It should be noted that the baseline comprehensive risk score provides a stable risk benchmark for patients, and through the continuous updating of real-time data, the comprehensive risk score can reflect the dynamic changes in the patient's health status. The long short-term memory network (LSTM) plays an important role here. The long short-term memory network can process the time series of patient health data and capture the potential trends of physiological changes to ensure the continuity and accuracy of risk assessment. In addition, the real-time updated comprehensive risk score will directly affect the subsequent alarm mechanism, nursing decisions and intervention measures, which will help improve patient safety and optimize treatment effects.
[0044] S5. Setting an alarm threshold according to the comprehensive risk score and comparing it with the alarm threshold to obtain alarm information includes the following steps: The individual characteristics of the patient are used to set a personalized initial alarm threshold range. Specifically, the patient's basic information (such as age, gender, medical history, weight, basic physiological data, etc.) is used to set a personalized initial alarm threshold based on medical knowledge and clinical experience. The initial threshold range is usually set based on the patient's physiological signs (such as heart rate, blood pressure, body temperature, blood oxygen concentration, etc.) to ensure early detection of the patient's health risks.
[0045] Use historical data to calculate the dynamic baseline threshold, set the normal range of heart rate and blood pressure indicators through statistical analysis, and dynamically adjust the alarm threshold according to the real-time monitored physiological sign data. Specifically, based on the patient's historical health data, use statistical analysis methods (such as mean and standard deviation calculation) to calculate the individual's dynamic baseline threshold. Determine the personalized normal range based on the patient's long-term monitoring of physiological data such as heart rate, blood pressure, and body temperature. For example, for heart rate, the upper and lower limits of the heart rate alarm can be set based on the patient's historical heart rate fluctuations. Dynamically adjust the alarm threshold based on real-time monitored physiological sign data (such as data collected by wearable devices or other sensors). If the patient's health status changes, the alarm threshold should be adaptively adjusted based on the current data to more accurately reflect the patient's current health status.
[0046] The comprehensive risk score is compared with the dynamic baseline threshold to obtain a comparison result. Specifically, the real-time comprehensive risk score is compared with the dynamically adjusted alarm threshold to calculate whether the score exceeds the normal range. If it exceeds, an alarm needs to be triggered. The comparison result may indicate the potential risk of the patient's health status.
[0047] An alarm is issued based on the comparison result, and the alarm information is obtained. Specifically, if the comprehensive risk score exceeds the set threshold range, an alarm will be issued to remind medical staff of the patient's health risk. The alarm information may include the specific risk type (such as heart abnormalities, low blood oxygen concentration, etc.), as well as the severity of the alarm (such as high, medium, or low level).
[0048] It should be noted that the setting of alarm thresholds is based on individual differences of patients, not just on universal health standards. Dynamic adjustment of thresholds can make the alarm mechanism more flexible, avoid underestimation or overestimation of patient risks, and thus achieve more accurate early warning.
[0049] S6. Respond to the alarm information and provide personalized clinical care suggestions, including the following steps: Receive and analyze alarm information to identify specific sources of risk. Specifically, once an alarm is triggered, the alarm information will be received and analyzed to identify specific sources of risk that the patient may face (such as abnormal heart rate, increased blood pressure, or decreased blood oxygen concentration). The purpose of this step is to clarify the health issues behind the alarm and avoid false positives or false negatives.
[0050] Generate preliminary care recommendations using medical history and medication. Specifically, generate personalized preliminary care recommendations based on the patient's medical history, medication records, allergy history, and other information. For example, if the patient has a history of heart disease and an abnormal heart rate, care recommendations may include accelerating heart monitoring and contacting relevant experts for intervention.
[0051] Optimize nursing recommendations based on individual patient characteristics and real-time feedback. Specifically, optimize and adjust initial nursing recommendations based on real-time feedback data from patients (such as self-reports from patients, changes in monitoring data, etc.). Nursing recommendations are customized based on the patient's actual health status and clinical needs to improve personalized treatment outcomes.
[0052] Nursing recommendations are presented to nursing staff in a visual form, and implementation guidance is provided, while the implementation results of nursing recommendations are recorded. Specifically, nursing recommendations will be presented to nursing staff in an intuitive visual form, such as through electronic health records (EHR) or dedicated nursing platforms, to ensure that nursing staff can quickly understand and implement corresponding measures. In addition, the implementation status of nursing staff (such as whether the recommendations are followed) will be recorded for subsequent evaluation and optimization.
[0053] It should be noted that the care recommendations provided should be based on the patient's real-time data and historical medical history to ensure that they can address the patient's current health problems in a targeted manner. Continuous optimization and feedback mechanisms help ensure that care recommendations always adapt to the patient's dynamic health status, maximizing the effectiveness of treatment and patient safety.
[0054] This embodiment also provides a neurology clinical nursing safety hazard analysis system, including: a data acquisition module, a low-dimensional feature module, a hazard monitoring module, a risk scoring module, an alarm module and an alarm response module; The data acquisition module is used to collect multi-source data of patients and perform preprocessing; the low-dimensional feature module is used to extract features and reduce the dimension of the preprocessed multi-source data to obtain low-dimensional safety hazard features; the hazard monitoring module is used to use the generative adversarial network to generate a high-quality low-dimensional feature data set, and form a safety hazard monitoring model to monitor safety hazards; the risk scoring module is used to update the comprehensive risk score of each patient in real time according to the monitoring results; the alarm module is used to set the alarm threshold according to the comprehensive risk score, and compare it with the alarm threshold to obtain alarm information; the alarm response module is used to respond to the alarm information and provide personalized clinical care recommendations.
[0055] This embodiment also provides a computer device, which is suitable for the method of analyzing safety hazards in clinical nursing in neurology, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the method of analyzing safety hazards in clinical nursing in neurology as proposed in the above embodiment.
[0056] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.
[0057] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for analyzing safety hazards in clinical nursing of neurology department proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.
[0058] In summary, the present invention can comprehensively and accurately evaluate the patient's safety hazards through the fusion and intelligent analysis of multi-source data, overcoming the limitations of isolated data processing in traditional methods. Generative adversarial networks are used to generate high-quality low-dimensional feature data, and combined with a dynamically updated risk scoring model, safety hazard monitoring can reflect patient changes in real time, improving the sensitivity and accuracy of monitoring. Based on real-time data and individual characteristics, this method can provide personalized nursing advice for each patient, effectively avoiding a "one-size-fits-all" nursing plan and improving the accuracy of nursing decisions. The use of dynamic baseline thresholds and statistical analysis methods to set alarm thresholds makes risk assessment more scientific and reasonable, reducing the possibility of false alarms and missed alarms.
[0059] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A method for analyzing safety hazards in clinical nursing of neurology, characterized by: include, Collect multi-source data of patients and perform pre-processing; Perform feature extraction and dimensionality reduction on the preprocessed multi-source data to obtain low-dimensional safety hazard features; Generate high-quality low-dimensional feature data sets using generative adversarial networks and form a safety hazard monitoring model to monitor safety hazards; Update each patient's comprehensive risk score in real time based on monitoring results; Set an alarm threshold based on the comprehensive risk score and compare it with the alarm threshold to obtain alarm information; Respond to alert messages and provide personalized clinical care recommendations.
2. The method for analyzing safety hazards in clinical nursing of neurology according to claim 1, characterized in that: The multi-source data includes the medical history, medication data, physiological sign data, surrounding environment parameters and limb acceleration data of neurology patients; The physiological characteristic data include heart rate, blood pressure, body temperature, muscle tremor amplitude and blood oxygen concentration; The ambient environment parameters include temperature and humidity, noise intensity and light intensity; The limb acceleration data is the movement data of the patient's limbs collected by the wearable device; The preprocessing includes low-pass filtering, band-pass filtering and data normalization.
3. The method for analyzing safety hazards in clinical nursing of neurology according to claim 2, characterized in that: Perform feature extraction on the preprocessed multi-source data to obtain safety hazard features, and perform dimensionality reduction to obtain low-dimensional safety hazard features. The specific steps are as follows: Use principal component analysis to extract features from preprocessed multi-source data to obtain streamlined features that reduce data redundancy; The simplified features are reduced in dimension through t-SNE, expressed as: ； in, For point Selecting points in low-dimensional space The probability of being its neighbor, For point The coordinates in low-dimensional space, For point The coordinates in low-dimensional space, For point-to-preceding point, For point-to-point postposition, is the index variable that refers to the specific point. is the index variable that refers to a specific point. is the index variable used to traverse all points to the previous point, is the index variable used to traverse all the post-points of the point pair; The low-dimensional safety hazard features after dimensionality reduction are saved in time series format.
4. The method for analyzing safety hazards in clinical nursing of neurology according to claim 3, characterized in that: Generate a high-quality low-dimensional feature data set using a generative adversarial network and form a safety hazard monitoring model to monitor safety hazards. The specific steps are as follows: Program according to the format of low-dimensional feature data to build the generator and discriminator of the generative adversarial network; Obtain a random noise vector through programming, input the random noise vector into a generator, and generate a realistic low-dimensional safety hazard feature; The cross entropy loss function is used as the optimization target of the discriminator, and the expression is: ； in, is the cross entropy loss value, is the distribution of low-dimensional safety hazard characteristics, is a low-dimensional security risk feature, is the distribution of realistic low-dimensional safety hazard features, is a random noise vector, is the interpolation data, is the interpolation sample, is the index variable that refers to the specific point. is the index variable that refers to the specific point later. is the expectation of the distribution of low-dimensional safety hazard features, is the probability of the discriminator judging the low-dimensional safety hazard features, is the expected value of the distribution of realistic low-dimensional safety hazard features, The realistic low-dimensional security risk features generated by the generator, is the gradient penalty weight, is the expected value of the distribution of the interpolated data, is the gradient of the interpolated sample, is the probability of the interpolation sample, is the contribution weight, For point Selecting points in low-dimensional space The probability of being its neighbor, Neighborhood relationship of Gaussian distribution; The simulated low-dimensional safety hazard features are mixed with the low-dimensional safety hazard features to obtain a mixed feature data set; According to Bayes' theorem, the posterior probability of mixed feature data is calculated, and the expression is: ； in, is the posterior probability value, is the abnormal status value, is mixed feature data, is the likelihood function, is the prior probability value, is the marginal likelihood value; The joint likelihood method is used to evaluate all mixed features comprehensively, and the ranking results are obtained to statistically conform to the assumption of uniform distribution. The SBC method is verified and the verified mixed feature data set is obtained. The Bayesian anomaly detection model is trained using the verified mixed feature dataset and monitored to obtain anomaly scores and potential risk labels.
5. The method for analyzing safety hazards in clinical nursing of neurology according to claim 4, characterized in that: Update the comprehensive risk score of each patient in real time based on the monitored safety hazards. The specific steps are as follows: A baseline composite risk score was set for each patient using the patient's initial physiological sign data; Receive the anomaly score and hazard label output by the safety hazard monitoring model, and combine them with real-time physiological feature data as input for updating the comprehensive risk score; The baseline comprehensive risk score and long short-term memory network are used to perform time series modeling on the patient's comprehensive risk score, and the patient's comprehensive risk score is updated in real time.
6. The method for analyzing safety hazards in clinical nursing of neurology according to claim 5, characterized in that: According to the physiological sign data in the multi-source data, an alarm threshold is set for the comprehensive risk score, and the comprehensive risk score is compared with the alarm threshold to obtain the alarm information. The specific steps are as follows: Using the individual characteristics of the patient to set a personalized initial alarm threshold range; Use historical data to calculate dynamic baseline thresholds, set normal ranges for heart rate and blood pressure indicators through statistical analysis, and dynamically adjust alarm thresholds based on real-time monitored physiological sign data; The composite risk scores were compared using dynamic baseline thresholds to obtain comparative results; An alarm is issued based on the comparison result, and alarm information is obtained.
7. The method for analyzing safety hazards in clinical nursing of neurology according to claim 6, characterized in that: Respond to alarm messages and provide clinical care recommendations. The specific steps are as follows: Receive and analyze alert information to identify specific sources of risk; utilizes medical history and medications to generate initial care recommendations; Optimize care recommendations based on individual patient characteristics and real-time feedback; The nursing recommendations are presented to nursing staff in a visual form, and implementation guidance is provided, while the implementation results of the nursing recommendations are recorded.
8. A neurology clinical nursing safety hazard analysis system, based on the neurology clinical nursing safety hazard analysis method according to any one of claims 1 to 7, characterized in that: Including data acquisition module, low-dimensional feature module, hidden danger monitoring module, risk scoring module, alarm module and alarm response module; The data acquisition module is used to collect multi-source data of patients and perform pre-processing; The low-dimensional feature module is used to extract features and reduce the dimension of pre-processed multi-source data to obtain low-dimensional safety hazard features; The hidden danger monitoring module is used to generate a high-quality low-dimensional feature data set using a generative adversarial network and form a safety hidden danger monitoring model to monitor safety hidden dangers; The risk scoring module is used to update the comprehensive risk score of each patient in real time according to the monitoring results; The alarm module is used to set an alarm threshold according to the comprehensive risk score and compare it with the alarm threshold to obtain alarm information; The alarm response module is used to respond to the alarm information and provide personalized clinical care suggestions.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for analyzing safety hazards in clinical nursing of neurology described in any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for analyzing safety hazards in clinical nursing of neurology described in any one of claims 1 to 7 are implemented.