Intelligent nursing risk assessment and early warning system and method thereof

Abstract

The invention relates to the technical field of medical information, in particular to an intelligent nursing risk assessment and early warning system and method, and the system comprises a data collection module, a data processing module, a risk assessment module, a threshold value adjustment module, an early warning module, an early warning module and an early warning module, the risk assessment module is in communication connection with the risk assessment module and is used for receiving a risk assessment result output by the risk assessment module; dynamically adjusting a risk early warning threshold based on the risk assessment result; the early warning module is in communication connection with the threshold adjustment module and the risk assessment module and is used for receiving a risk assessment result of the risk assessment module and a dynamic threshold of the threshold adjustment module; when the risk assessment result exceeds a dynamic threshold value, generating early warning information; and the early warning information is sent to a medical staff terminal, and compared with a traditional statistical method or a shallow machine learning algorithm, the method provided by the invention has higher mode recognition capability and prediction accuracy.

Description

Technical Field

[0001] The present invention relates to the field of medical information technology, and in particular to an intelligent nursing risk assessment and early warning system and method thereof. Background Art

[0002] With the continuous advancement of medical technology and the aging of the population, the complexity and importance of nursing work have become increasingly prominent. Traditional nursing risk assessment methods mainly rely on the experience and regular examinations of nursing staff. This method has problems such as strong subjectivity, poor real-time performance, and heavy workload. In recent years, with the widespread application of information technology in the medical field, some intelligent nursing risk assessment systems have begun to appear.

[0003] Existing intelligent nursing risk assessment systems usually use fixed assessment scales and preset risk thresholds. These systems can improve the efficiency of nursing risk assessment to a certain extent, but there are still some obvious shortcomings. First, these systems can often only process structured data and it is difficult to make full use of unstructured nursing records and medical text information. Secondly, the preset risk thresholds lack flexibility and cannot be dynamically adjusted according to individual differences and changes in the condition of patients. Furthermore, the risk assessment models of existing systems are usually based on simple statistical methods or shallow machine learning algorithms, which make it difficult to capture complex time series patterns and multi-factor interactions.

[0004] In addition, existing systems also have deficiencies in data visualization and result interpretation. Most systems can only provide simple risk assessment results in numerical or graded form, lacking intuitive visual display, which makes it difficult for medical staff to quickly understand the patient's overall risk status. At the same time, the system often cannot provide specific explanations of risk assessment results and personalized intervention recommendations, which to some extent limits the practical application value of the system.

[0005] Finally, existing systems generally lack the ability to self-learn and continuously optimize. With the continuous updating of medical knowledge and the continuous accumulation of clinical practice, fixed risk assessment models will soon become outdated and unable to adapt to the ever-changing medical environment and patient characteristics.

[0006] In view of the above problems, there is an urgent need for a system and method that can realize intelligent, personalized, and dynamic nursing risk assessment and early warning. The system should be able to process multi-source heterogeneous medical data, use advanced deep learning algorithms to build a risk assessment model, realize dynamic adjustment of risk thresholds, provide intuitive data visualization and explainable assessment results, and have the ability to continuously learn and optimize. Summary of the invention

[0007] The present invention is proposed to solve the above technical problems and provides an intelligent nursing risk assessment and early warning system and method thereof, which realizes the automation of the whole process from data collection, processing, risk assessment to early warning through the collaborative work of multiple modules.

[0008] The present invention proposes an intelligent nursing risk assessment and early warning system, including:

[0009] Data acquisition module for:

[0010] Collect patients' vital signs and nursing operation data in real time through a variety of sensors;

[0011] Obtain patients’ historical health data from hospital data centers;

[0012] A data processing module is connected to the data acquisition module for:

[0013] Receiving real-time data and historical data sent by the data acquisition module;

[0014] Preprocessing the real-time data and historical data, including data cleaning, missing value filling and standardization;

[0015] A risk assessment module is connected in communication with the data processing module and is used to:

[0016] Based on the preprocessed data, a risk assessment model is constructed using a deep learning algorithm;

[0017] Using the risk assessment model to conduct real-time risk assessment of the patient's current status;

[0018] Output risk assessment results, including risk levels and specific risk factors;

[0019] A threshold adjustment module is connected in communication with the risk assessment module and is used to:

[0020] Receiving the risk assessment result output by the risk assessment module;

[0021] Based on the risk assessment results, dynamically adjust the risk warning threshold;

[0022] The early warning module is in communication with the threshold adjustment module and the risk assessment module, and is used to:

[0023] Receiving the risk assessment result of the risk assessment module and the dynamic threshold of the threshold adjustment module;

[0024] When the risk assessment result exceeds the dynamic threshold, an early warning message is generated;

[0025] The warning information is sent to the medical staff terminal.

[0026] Preferably, the data acquisition module comprises:

[0027] A vital signs collection unit, which is used to collect vital signs data including blood pressure, heart rate, respiratory rate, blood oxygen saturation and body temperature;

[0028] Nursing operation recording unit, used to record nursing operation data including medication, dressing changes, body position changes and tube operations;

[0029] The historical data acquisition unit is used to obtain the patient's electronic medical records, previous medical records, medication history and other historical health data from the hospital data center.

[0030] Preferably, the data processing module comprises:

[0031] Data cleaning unit, used to remove outliers and duplicate data;

[0032] Missing value processing unit, used to fill missing data using interpolation algorithms or machine learning methods;

[0033] Data standardization unit is used to convert data of different scales into a unified scale range.

[0034] Preferably, the risk assessment module comprises:

[0035] A feature extraction unit, used to extract key features from the preprocessed data;

[0036] Model training unit, used to train risk assessment models using long short-term memory network (LSTM) or convolutional neural network (CNN) deep learning algorithms;

[0037] A real-time assessment unit, which is used to perform real-time risk assessment of the patient's current status using the trained model;

[0038] The risk grading unit is used to classify the patient's risk level into high risk, medium risk and low risk according to the assessment results.

[0039] Preferably, the threshold adjustment module comprises:

[0040] A threshold initialization unit, used to set the initial warning threshold based on historical data and expert knowledge;

[0041] A dynamic adjustment unit, used to dynamically adjust the warning threshold according to individual patient characteristics and real-time risk assessment results;

[0042] The threshold optimization unit is used to optimize the threshold adjustment strategy by minimizing a predefined loss function.

[0043] As a preference, it also includes:

[0044] The data visualization module is in communication with the risk assessment module and the early warning module and is used to:

[0045] Receiving the risk assessment result of the risk assessment module and the warning information of the warning module;

[0046] Generate visual charts including vital signs trend charts, risk level change charts and early warning information summaries;

[0047] The visualization chart is pushed to the medical staff terminal.

[0048] As a preference, it also includes:

[0049] A knowledge base module, in communication with the risk assessment module, is used to:

[0050] Stores professional knowledge including disease knowledge, care practices and risk factors;

[0051] Provide knowledge support for risk assessment models and improve the accuracy and interpretability of risk assessment.

[0052] As a preference, it also includes:

[0053] The intervention suggestion module is in communication with the risk assessment module and the early warning module and is used to:

[0054] Generate personalized nursing intervention recommendations based on risk assessment results and early warning information;

[0055] The nursing intervention suggestion is sent to the medical staff terminal.

[0056] As a preference, it also includes:

[0057] A model updating module, which is in communication with the risk assessment module, is used to:

[0058] Regularly collect new patient data and risk assessment results;

[0059] Use incremental learning methods to update risk assessment models to improve model adaptability and accuracy.

[0060] The intelligent nursing risk assessment and early warning method based on the system includes the following steps:

[0061] S1, collects patients’ vital signs data and nursing operation data in real time through a variety of sensors, and obtains patients’ historical health data from the hospital data center;

[0062] S2, preprocessing the collected real-time data and historical data, including data cleaning, missing value filling and standardization;

[0063] S3, uses deep learning algorithms to build a risk assessment model, which is trained based on the preprocessed data;

[0064] S4, using the trained risk assessment model to conduct real-time risk assessment of the patient’s current status, and output the risk level and specific risk factors;

[0065] S5, dynamically adjust the risk warning threshold based on the risk assessment results;

[0066] S6, when the risk assessment result exceeds the dynamic threshold, an early warning message is generated and sent to the medical staff terminal;

[0067] S7, generating visual charts including vital signs trend charts, risk level change charts and early warning information summary, and pushing them to the medical staff terminal;

[0068] S8, based on risk assessment results and early warning information, generates personalized nursing intervention recommendations;

[0069] S9, regularly collect new patient data and risk assessment results, and use incremental learning methods to update the risk assessment model.

[0070] The beneficial effects of the present invention are mainly reflected in the following aspects:

[0071] First, the system of the present invention realizes all-round and multi-dimensional patient data collection by integrating multiple sensors and data sources. This includes not only real-time vital sign data and nursing operation records, but also covers the patient's historical health data. This comprehensive data collection provides a rich information basis for subsequent risk assessment, greatly improving the accuracy and comprehensiveness of risk assessment.

[0072] Secondly, the present invention uses advanced deep learning algorithms, especially long short-term memory networks (LSTM), to build risk assessment models. This method can effectively process time series data and capture the dynamic changes in patient status and the complex interactions between multiple factors. Compared with traditional statistical methods or shallow machine learning algorithms, the method of the present invention has stronger pattern recognition capabilities and prediction accuracy.

[0073] Furthermore, the present invention realizes the dynamic adjustment of risk warning threshold. By considering the individual characteristics and real-time status of patients, the system can set personalized warning thresholds for each patient and make real-time adjustments according to changes in the condition. This dynamic threshold mechanism significantly improves the accuracy of warnings and reduces the probability of false positives and false negatives.

[0074] In addition, the data visualization module of the present invention can generate intuitive, multi-dimensional visualization charts, including vital signs trend charts, risk level change charts, etc. These visualization results help medical staff quickly grasp the overall condition and risk trend of patients, improving work efficiency and decision-making quality.

[0075] The present invention also integrates a professional medical knowledge base to provide knowledge support for risk assessment. This not only improves the reliability of risk assessment, but also enhances the interpretability of the system. The system can provide a specific explanation for each risk assessment result and generate personalized nursing intervention recommendations, which greatly enhances the practical value of the system.

[0076] Finally, the model update module of the present invention realizes the continuous optimization of the risk assessment model. By regularly collecting new data and updating the model using incremental learning methods, the system can continuously adapt to new medical knowledge and patient characteristics, maintaining its long-term effectiveness.

[0077] In summary, the intelligent nursing risk assessment and early warning system and method of the present invention realizes the intelligent, personalized and dynamic nursing risk management through the organic combination of multiple innovative points. It can not only improve the efficiency and quality of nursing work, but also significantly improve the safety and prognosis of patients, and has important clinical application value and social significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 It is a logic block diagram of the whole system of the present invention.

[0079] Figure 2 It is a logic block diagram of the data acquisition module of the present invention.

[0080] Figure 3 It is a logic block diagram of the data processing module of the present invention.

[0081] Figure 4 It is a logic block diagram of the risk assessment module of the present invention.

[0082] Figure 5 It is a logic block diagram of the threshold adjustment module of the present invention. DETAILED DESCRIPTION

[0083] The present invention provides an intelligent nursing risk assessment and early warning system and method thereof. The system can monitor the patient's vital signs and nursing operation data in real time, and build a risk assessment model through a deep learning algorithm in combination with the patient's historical health data, so as to achieve accurate nursing risk assessment and timely early warning.

[0084] like Figure 1-5As shown, the intelligent nursing risk assessment and early warning system of the present invention includes a data acquisition module 1, a data processing module 2, a risk assessment module 3, a threshold adjustment module 4 and an early warning module 5. These modules work together to form a complete risk assessment and early warning process.

[0085] The data acquisition module 1 is used to collect the patient's vital signs data and nursing operation data in real time through a variety of sensors, and obtain the patient's historical health data from the hospital data center. Specifically, the data acquisition module 1 can include a variety of sensors, such as a sphygmomanometer, an electrocardiograph, an oximeter, etc., which can continuously collect the patient's vital signs data. For example, blood pressure can be measured every 15 minutes, and heart rate and blood oxygen saturation can be recorded once a minute. In addition, the data acquisition module 1 can also record the nursing staff's operations, such as medication, dressing changes, body position changes, etc., through the nursing information system. These real-time data provide the latest and most direct basis for risk assessment.

[0086] At the same time, the data acquisition module 1 can also obtain the patient's historical health data from the hospital's electronic medical record system, including past medical history, medication records, surgical records, etc. These historical data can help the system understand the patient's health status more comprehensively, thereby improving the accuracy of risk assessment. For example, for a patient with a history of heart disease, the system will pay more attention to changes in his heart rate and blood pressure.

[0087] The data processing module 2 is connected to the data acquisition module 1 for receiving the real-time data and historical data sent by the data acquisition module 1 and preprocessing the data, including data cleaning, missing value filling and standardization. Data preprocessing is a key step to ensure the accuracy of subsequent risk assessment.

[0088] During the data cleaning process, the system will remove obvious outliers and duplicate data. For example, if the blood pressure measurement value is obviously wrong, such as 300 / 200mmHg, the system will automatically remove it. For missing values, the system will select an appropriate filling method based on the characteristics of the data. For continuous vital signs data, interpolation can be used for filling; for discrete nursing operation data, the nearest neighbor method or mode filling can be used.

[0089] Data standardization is to convert data of different scales into a unified scale range, usually using Z-score standardization or Min-Max standardization method. For example, for blood pressure data, the following Z-score standardization formula can be used:

[0090]

[0091] Among them, Z is the standardized value, X is the original value, μ is the mean, and σ is the standard deviation.

[0092] Through such preprocessing, the data processing module 2 can ensure the quality of data input to the risk assessment module, thereby improving the accuracy and reliability of risk assessment.

[0093] The risk assessment module 3 is in communication with the data processing module 2, and is used to construct a risk assessment model based on the preprocessed data using a deep learning algorithm. The present invention preferably uses a long short-term memory network (LSTM) as the core algorithm of the risk assessment model, because LSTM is particularly suitable for processing time series data and can capture the changing trend of the patient's status over time.

[0094] The basic structure of the LSTM model is as follows:

[0095]

[0096] Among them, f t 、i t , o t They are forget gate, input gate and output gate respectively, c t is the cell state, h t is the hidden state, W and b are the weight and bias parameters, σ is the sigmoid activation function, and tanh is the hyperbolic tangent activation function.

[0097] The risk assessment module 3 uses the trained LSTM model to conduct real-time risk assessment on the patient's current status and outputs the risk assessment results, including risk level and specific risk factors. The risk level can be divided into three levels: high, medium, and low. For example, a risk probability greater than 0.7 can be defined as high risk, between 0.3 and 0.7 as medium risk, and less than 0.3 as low risk. This classification method can help medical staff quickly identify patients who need special attention.

[0098] The threshold adjustment module 4 is in communication with the risk assessment module 3, and is used to receive the risk assessment results output by the risk assessment module 3, and dynamically adjust the risk warning threshold based on these results. The dynamic adjustment of the threshold is an important innovation of the present invention, which can make the system adapt to the individual differences and changes in the condition of different patients.

[0099] For example, for a patient who has just undergone surgery, the system may lower the warning threshold to more sensitively capture potential risks. For a patient who is recovering well, the system may gradually increase the warning threshold to reduce unnecessary alarms. The threshold adjustment can be done using the following formula:

[0100] T new =T old +α(RT old ),

[0101] Among them, T newis the new threshold, T old is the old threshold, R is the current risk assessment result, and α is the learning rate (usually a value between 0.1 and 0.3). This formula describes a method for dynamically adjusting the warning threshold, which allows the system to adaptively adjust the threshold according to the specific condition of the patient. For patients who have just undergone surgery, by setting a lower initial threshold and an appropriate learning rate α, any abnormal changes that may indicate complications can be more sensitively monitored. For patients who have recovered well, gradually increasing the threshold can reduce false positive alarms caused by nonspecific fluctuations, ensure the effective use of medical resources and reduce unnecessary anxiety for patients.

[0102] The early warning module 5 is connected to the threshold adjustment module 4 and the risk assessment module 3 for receiving the risk assessment result of the risk assessment module 3 and the dynamic threshold of the threshold adjustment module 4. When the risk assessment result exceeds the dynamic threshold, the early warning module 5 generates early warning information and sends the information to the medical staff terminal. The early warning information may include the patient's basic information, current risk level, specific risk factors, and recommended intervention measures.

[0103] The intelligent nursing risk assessment and early warning system of the present invention realizes the automation of the whole process from data collection, processing, risk assessment to early warning through this modular design. The core advantage of the system is that it can assess patient risks in real time and continuously, and dynamically adjust the early warning threshold according to the individual characteristics and status changes of the patient, thereby improving the accuracy and timeliness of the early warning. This can not only help medical staff to discover and intervene in potential risks in a timely manner, but also reduce false positives and missed positives, and improve the efficiency and quality of nursing work.

[0104] In practical applications, this system can be deployed in hospital intensive care units, general wards, and even home care environments. By continuously monitoring and analyzing patient data, the system can help prevent common care-related complications such as pressure sores, falls, and infections, thereby significantly improving patient safety and care quality.

[0105] In addition, the present invention has good scalability and adaptability. As new medical data and knowledge accumulate, the system can continuously update and optimize its risk assessment model, enabling it to identify and warn more types of nursing risks. This ability to continuously learn and optimize enables the system to maintain its advancement and practicality in the rapidly developing field of medical care.

[0106] In general, the intelligent nursing risk assessment and early warning system and method provided by the present invention realizes intelligent, personalized and dynamic management of nursing risks through advanced data processing technology, deep learning algorithm and dynamic threshold adjustment mechanism. This can not only improve the efficiency and quality of nursing work, but also significantly improve the safety and prognosis of patients, and has important clinical application value and social significance.

[0107] The risk assessment module 3 of the present invention comprises a feature extraction unit 31, a model training unit 32, a real-time assessment unit 33 and a risk classification unit 34. These units work together to complete the whole process from data feature extraction to final risk assessment.

[0108] The feature extraction unit 31 is used to extract key features from the preprocessed data. In the scenario of nursing risk assessment, feature extraction is a crucial step, which directly affects the accuracy and efficiency of risk assessment. The feature extraction unit 31 of the present invention adopts a variety of advanced feature extraction methods, including but not limited to time series feature extraction, statistical feature extraction and frequency domain feature extraction.

[0109] For example, for electrocardiogram data, the feature extraction unit 31 can extract features such as RR interval, QT interval, ST segment change, etc.; for blood pressure data, it can extract features such as systolic pressure, diastolic pressure, mean arterial pressure, etc.; for nursing operation data, it can extract features such as operation frequency and operation duration, etc. Through this multi-dimensional feature extraction, the system can fully capture all aspects of the patient's status and provide a rich information basis for subsequent risk assessment.

[0110] The model training unit 32 is used to train the risk assessment model using a deep learning algorithm such as a long short-term memory network (LSTM) or a convolutional neural network (CNN). The present invention preferably uses LSTM as the main model training algorithm because LSTM is particularly suitable for processing time series data and can effectively capture the changing trend of the patient's status over time.

[0111] During the model training process, the model training unit 32 uses a large amount of historical data for training. These data include the patient's vital signs data, nursing operation records, and the final nursing results (such as whether an adverse event occurred). The training process can be expressed as the following optimization problem:

[0112]

[0113] Among them, θ represents the model parameters, N is the number of training samples, and x i is the input feature, y i is the true label, f θ is the model function, and L is the loss function. Preferably, the present invention adopts the cross entropy loss function:

[0114] L=-∑ i y i log(p i ),

[0115] Among them, y i is the true label, p i This description explains how the model training unit uses a large amount of historical data (including patients’ vital signs, nursing operation records, and final nursing results) to adjust the model parameters θ so that the model predicts the output f θ (x i ) is as close to the actual result y as possible i By minimizing the average loss That is, the difference between each predicted value and the true value can effectively train a model that can accurately predict the patient's risk assessment results. Using cross entropy as the loss function L is particularly suitable for classification tasks because it measures the predicted probability distribution p i and the true label distribution y i The difference between.

[0116] The real-time assessment unit 33 is used to perform real-time risk assessment on the patient's current status using the trained model. This unit will continuously receive the patient's latest data, and quickly infer through the trained model to output the risk assessment results. Preferably, the frequency of real-time assessment can be dynamically adjusted according to the patient's condition. For high-risk patients, the assessment frequency can be increased, for example, once every 5 minutes; for low-risk patients, the frequency can be appropriately reduced, for example, once every 30 minutes.

[0117] The risk grading unit 34 is used to classify the patient's risk level into high risk, medium risk and low risk according to the evaluation results. Preferably, the present invention adopts the following risk grading standards:

[0118] High risk: risk score greater than 0.7;

[0119] Medium risk: risk score between 0.3 and 0.7;

[0120] Low risk: risk score less than 0.3;

[0121] This risk grading method can help medical staff quickly identify patients who need special attention and allocate medical resources rationally.

[0122] The threshold adjustment module 4 of the present invention comprises a threshold initialization unit 41, a dynamic adjustment unit 42 and a threshold optimization unit 43. These units together realize the intelligent and personalized adjustment of the warning threshold.

[0123] The threshold initialization unit 41 is used to set the initial warning threshold based on historical data and expert knowledge. When the system is initialized or a new patient is admitted to the hospital, a suitable initial threshold needs to be set. The setting of this initial threshold will take into account multiple factors, including the patient's basic information (such as age, gender, main diagnosis, etc.), historical data statistics, and the advice of medical experts. For example, for a heart disease patient over 65 years old, the initial warning threshold may be set lower than that of a young and healthy patient.

[0124] The dynamic adjustment unit 42 is used to dynamically adjust the warning threshold according to the individual characteristics of the patient and the real-time risk assessment results. This is an important innovation of the present invention, which enables the system to adapt to the dynamic changes in the patient's status. The dynamic adjustment process can be expressed by the following formula:

[0125] T new =T old +α(RT old )+β(PT old ),

[0126] Among them, T new is the new threshold, T old is the old threshold, R is the current risk assessment result, P is the individual characteristic factor of the patient, and α and β are adjustment coefficients (usually between 0 and 1).

[0127] The threshold optimization unit 43 is used to optimize the threshold adjustment strategy by minimizing a predefined loss function. The goal of this unit is to reduce false positives while ensuring that real high-risk situations are not missed. The optimization process can be expressed as the following problem:

[0128] min θ L(FPR,FNR),

[0129] Wherein, θ represents the parameter of the threshold adjustment strategy, FPR is the false positive rate, FNR is the false negative rate, and L is the loss function. Preferably, the following weighted loss function can be used:

[0130] L=w 1 ×FPR+w 2 × FNR,

[0131] Among them, w 1 and w 2 is the weight coefficient, which can be adjusted according to the specific application scenario. For example, in the intensive care unit, more attention may be paid to reducing the false negative rate, so w 2This description emphasizes an important innovation of the present invention, the dynamic adjustment unit 42, which allows the system to dynamically adjust the warning threshold according to the individual characteristics of the patient and the real-time risk assessment results to adapt to the changes in the patient's status. The threshold optimization unit 43 is dedicated to optimizing this adjustment strategy to ensure that the system's warning mechanism is both sensitive and reliable, thereby effectively supporting medical decision-making.

[0132] The present invention further includes a data visualization module 6, which is in communication with the risk assessment module 3 and the early warning module 5. The main function of the data visualization module 6 is to convert complex data and analysis results into an intuitive and easy-to-understand visual representation to help medical staff quickly understand the patient's condition and risk trend.

[0133] The data visualization module 6 can generate various types of visualization charts, including but not limited to:

[0134] 1. Vital signs trend chart: shows the changing trend of various vital signs of the patient (such as heart rate, blood pressure, body temperature, etc.) over time. Preferably, a line chart can be used, with the x-axis representing time and the y-axis representing the corresponding vital sign index. Different vital signs can be represented by lines of different colors for easy distinction.

[0135] 2. Risk level change chart: Displays the dynamic changes of the patient's risk level. This can be in the form of a step chart, with the x-axis representing time and the y-axis representing the risk level (high, medium, low). This visualization method can help medical staff quickly identify the change points of the risk level.

[0136] 3. Early warning information summary: Display historical early warning information in the form of a table or timeline, including the time of the warning, the reason for the warning, the intervention measures taken, etc. This summary view can help medical staff understand the patient's risk history and make better care decisions.

[0137] 4. Risk factor weight chart: Use pie charts or bar charts to display the various factors that lead to high risk and their weights. This can help medical staff understand which factors have the greatest impact on the patient's risk status, so as to take targeted intervention measures.

[0138] 5. Nursing operation time distribution map: Use a heat map to show the time distribution of nursing operations, with the x-axis representing the time of the day and the y-axis representing different types of nursing operations. This visualization can help nursing managers optimize the time schedule of nursing work.

[0139] The output of the data visualization module 6 will be pushed to the medical staff terminal, which can be a mobile device (such as a tablet computer, a smart phone) or a fixed workstation. Preferably, the system of the present invention can also customize different visualization views according to the needs of different roles (such as attending physicians, head nurses, general nurses, etc.) to meet the information needs of different positions.

[0140] Through this intuitive data visualization, the system of the present invention can greatly improve the work efficiency of medical staff, helping them to grasp the overall condition and risk trend of patients more quickly and accurately, so as to make better nursing decisions. At the same time, these visualization results can also be used to communicate with patients and their families, helping them to better understand the condition and treatment process.

[0141] The present invention further comprises a knowledge base module 7, which is in communication connection with the risk assessment module 3. The knowledge base module 7 is an important component of the present system, which provides professional knowledge support for risk assessment and helps to improve the accuracy and interpretability of risk assessment.

[0142] The knowledge base module 7 is mainly used to store professional knowledge including disease knowledge, nursing norms and risk factors. This knowledge comes from medical literature, clinical guidelines, expert consensus and practical experience. Preferably, the knowledge base of the present invention is organized in an ontology manner, which can effectively represent complex medical concepts and the relationships between them.

[0143] For example, for a common nursing risk, pressure injuries (pressure ulcers), the knowledge base might include the following information:

[0144] 1. Risk factors: including long-term bed rest, malnutrition, moist skin, etc.

[0145] 2. Evaluation methods: such as the use and scoring criteria of the Braden scoring scale.

[0146] 3. Preventive measures: such as turning over regularly, using a pressure-relieving mattress, keeping the skin clean and dry, etc.

[0147] 4. Intervention plan: treatment methods for different degrees of pressure injuries.

[0148] The collaborative work of the knowledge base module 7 and the risk assessment module 3 can significantly improve the quality of risk assessment. For example, when conducting a pressure injury risk assessment, the risk assessment module 3 will not only consider the patient's real-time data (such as mobility, skin moisture, etc.), but also combine the professional knowledge in the knowledge base (such as the Braden scoring standard) to make a more accurate assessment.

[0149] In addition, the knowledge base module 7 can also provide explanations for the risk assessment results. When the system gives a high risk assessment result, the specific reasons leading to the high risk can be explained based on the information in the knowledge base, which helps medical staff better understand the risk assessment results and formulate targeted intervention measures.

[0150] The present invention also includes an intervention suggestion module 8, which is in communication with the risk assessment module 3 and the early warning module 5. The main function of the intervention suggestion module 8 is to generate personalized nursing intervention suggestions based on the risk assessment results and early warning information, and send these suggestions to the medical staff terminal.

[0151] The workflow of intervention recommendation module 8 is as follows:

[0152] 1. Receive risk assessment results: Receive the patient's risk assessment results from the risk assessment module 3, including the risk level and specific risk factors.

[0153] 2. Analyze the warning information: Receive the warning information from the warning module 5 to understand the specific reasons for triggering the warning.

[0154] 3. Query the knowledge base: Based on the risk assessment results and warning information, query the relevant knowledge in the knowledge base module 7.

[0155] 4. Generate intervention recommendations: Based on the above information, generate personalized nursing intervention recommendations.

[0156] 5. Send suggestions: Send the generated intervention suggestions to the medical staff terminal.

[0157] For example, if the system detects that a patient is at high risk for pressure injury, the intervention recommendation module 8 may generate the following recommendation:

[0158] Turn the patient over every 2 hours;

[0159] Use a pressure-relieving mattress;

[0160] Keep your skin clean and dry, and check the skin condition every time you turn over;

[0161] Ensure that the patient is adequately nourished and supplement with protein if necessary;

[0162] Encourage patients to perform appropriate activities in bed;

[0163] These recommendations are personalized to the patient’s specific circumstances. For example, for a patient who is nutritionally sound but has limited mobility, the system might emphasize turning over and using assistive devices.

[0164] By providing such personalized intervention suggestions, the system of the present invention can not only detect risks in a timely manner, but also provide medical staff with specific action guidance, thereby more effectively preventing and reducing the occurrence of adverse events.

[0165] The present invention also includes a model updating module 9, which is in communication with the risk assessment module 3. The main function of the model updating module 9 is to regularly collect new patient data and risk assessment results, update the risk assessment model using an incremental learning method, and improve the adaptability and accuracy of the model.

[0166] Model updates are key to maintaining the long-term effectiveness of the system. Over time, the characteristics of the patient population may change, and new medical knowledge and nursing practices may emerge, all of which require the risk assessment model to be able to adapt and update in a timely manner.

[0167] The workflow of the model updating module 9 is as follows:

[0168] 1. Data collection: Regularly collect new patient data and risk assessment results. These data include patients’ vital signs, nursing operation records, actual adverse events, etc.

[0169] 2. Data preprocessing: Perform preprocessing operations such as cleaning and standardization on the newly collected data to make it meet the requirements of model training.

[0170] 3. Incremental learning: Use incremental learning methods to update existing risk assessment models. Incremental learning allows the model to learn new knowledge without losing the knowledge it has already learned.

[0171] 4. Model evaluation: Evaluate the performance of the updated model to ensure that the new model is at least as good as the old one.

[0172] 5. Model deployment: If the new model performance meets the requirements, it is deployed to the production environment.

[0173] Preferably, the present invention adopts the following incremental learning strategy:

[0174]

[0175] Among them, w t and w t +1 represents the model parameters before and after the update, η is the learning rate, L is the loss function, f wt is the current model, (x new ,y new ) is a new training sample.

[0176] This formula describes an incremental learning strategy based on gradient descent, which allows the model to gradually adjust its parameters as new data is received without re-training using all historical data. This approach is particularly useful for real-time or online learning because it can effectively incorporate new information into existing models while maintaining computational efficiency. By appropriately selecting the learning rate η, the pace of updates can be controlled to ensure that the model can adapt to new data patterns while avoiding overfitting to the latest samples. This incremental learning method is a preferred embodiment of the present invention and helps to improve the speed and accuracy of the model's response to changes in patient status.

[0177] In this way, the model can continually learn new data patterns while retaining useful information learned previously. For example, if the system finds that the use of a new type of medical device is associated with a specific type of care risk, the model can quickly learn this new association and factor it into future risk assessments.

[0178] The frequency of model updates can be adjusted according to actual conditions. In the early stage of system deployment, more frequent updates may be required, such as once a week; as the system runs stably, the update frequency can be gradually reduced, such as once a month or once a quarter.

[0179] Through regular model updates, the system of the present invention can maintain the advancement and accuracy of its risk assessment capabilities, adapt to the ever-changing medical environment and patient characteristics, and provide patients with continuous high-quality nursing services.

[0180] Finally, the present invention also provides an intelligent nursing risk assessment and early warning method based on the above system. The method comprises the following steps:

[0181] S1, collects patients’ vital signs data and nursing operation data in real time through a variety of sensors, and obtains patients’ historical health data from the hospital data center;

[0182] S2, preprocessing the collected real-time data and historical data, including data cleaning, missing value filling and standardization;

[0183] S3, uses deep learning algorithms to build a risk assessment model, which is trained based on the preprocessed data;

[0184] S4, using the trained risk assessment model to conduct real-time risk assessment of the patient’s current status, and output the risk level and specific risk factors;

[0185] S5, dynamically adjust the risk warning threshold based on the risk assessment results;

[0186] S6, when the risk assessment result exceeds the dynamic threshold, an early warning message is generated and sent to the medical staff terminal;

[0187] S7, generating visual charts including vital signs trend charts, risk level change charts and early warning information summary, and pushing them to the medical staff terminal;

[0188] S8, based on risk assessment results and early warning information, generates personalized nursing intervention recommendations;

[0189] S9, regularly collect new patient data and risk assessment results, and use incremental learning methods to update the risk assessment model.

[0190] This method covers the entire process from data collection to risk assessment, early warning, visualization, intervention suggestion generation, and model updating. It embodies the core workflow of the system of the present invention, and through the organic combination of these steps, intelligent, personalized, and dynamic nursing risk management is achieved.

[0191] For example, in step S1, the system may collect the patient's heart rate, blood pressure, respiratory rate and other vital signs data once a minute, and each nursing operation (such as medication, turning over) will be recorded. At the same time, the system will also extract the patient's medical history, medication records and other information from the hospital's electronic medical record system.

[0192] In step S3, the system may use a large-scale dataset containing millions of historical records to train a deep learning model. This training process may take hours or even days, but once completed, the model can perform risk assessment on new patient data in milliseconds.

[0193] In step S5, the system may adjust the warning threshold according to the individual characteristics and current status of the patient. For example, for a patient who has just undergone a major operation, the system may lower the warning threshold to more sensitively capture potential risks.

[0194] Through this method, the system of the present invention can provide each patient with round-the-clock, personalized risk monitoring and early warning services, greatly improving the efficiency and quality of nursing work and providing a strong guarantee for patient safety.

[0195] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. Intelligent nursing risk assessment and early warning system, characterized by: include: Data acquisition module for: Collect patients' vital signs and nursing operation data in real time through a variety of sensors; Obtain patients’ historical health data from hospital data centers; A data processing module is connected to the data acquisition module for: Receiving real-time data and historical data sent by the data acquisition module; Preprocessing the real-time data and historical data, including data cleaning, missing value filling and standardization; A risk assessment module is connected in communication with the data processing module and is used to: Based on the preprocessed data, a risk assessment model is constructed using a deep learning algorithm; Using the risk assessment model to conduct real-time risk assessment of the patient's current status; Output risk assessment results, including risk levels and specific risk factors; A threshold adjustment module is connected in communication with the risk assessment module and is used to: Receiving the risk assessment result output by the risk assessment module; Based on the risk assessment results, dynamically adjust the risk warning threshold; The early warning module is in communication with the threshold adjustment module and the risk assessment module and is used to: Receiving the risk assessment result of the risk assessment module and the dynamic threshold of the threshold adjustment module; When the risk assessment result exceeds the dynamic threshold, an early warning message is generated; The warning information is sent to the medical staff terminal.

2. The system according to claim 1, characterized in that The data acquisition module comprises: A vital signs collection unit, which is used to collect vital signs data including blood pressure, heart rate, respiratory rate, blood oxygen saturation and body temperature; Nursing operation recording unit, used to record nursing operation data including medication, dressing changes, body position changes and tube operations; The historical data acquisition unit is used to obtain the patient's electronic medical records, previous medical records, medication history and other historical health data from the hospital data center.

3. The system according to claim 1, characterized in that The data processing module comprises: Data cleaning unit, used to remove outliers and duplicate data; Missing value processing unit, used to fill missing data using interpolation algorithms or machine learning methods; Data standardization unit is used to convert data of different scales into a unified scale range.

4. The system according to claim 1, characterized in that The risk assessment module includes: A feature extraction unit, used to extract key features from the preprocessed data; Model training unit, used to train risk assessment models using long short-term memory network (LSTM) or convolutional neural network (CNN) deep learning algorithms; A real-time assessment unit, which is used to perform real-time risk assessment of the patient's current status using the trained model; The risk grading unit is used to classify the patient's risk level into high risk, medium risk and low risk according to the assessment results.

5. The system according to claim 1, characterized in that The threshold adjustment module comprises: A threshold initialization unit, used to set the initial warning threshold based on historical data and expert knowledge; A dynamic adjustment unit, used to dynamically adjust the warning threshold according to individual patient characteristics and real-time risk assessment results; The threshold optimization unit is used to optimize the threshold adjustment strategy by minimizing a predefined loss function.

6. The system according to claim 1, characterized in that Also includes: The data visualization module is in communication with the risk assessment module and the early warning module and is used to: Receiving the risk assessment result of the risk assessment module and the warning information of the warning module; Generate visual charts including vital signs trend charts, risk level change charts and early warning information summaries; The visualization chart is pushed to the medical staff terminal.

7. The system according to claim 1, characterized in that Also includes: A knowledge base module, in communication with the risk assessment module, is used to: Stores professional knowledge including disease knowledge, care practices and risk factors; Provide knowledge support for risk assessment models and improve the accuracy and interpretability of risk assessment.

8. The system according to claim 1, characterized in that Also includes: The intervention suggestion module is in communication with the risk assessment module and the early warning module and is used to: Generate personalized nursing intervention recommendations based on risk assessment results and early warning information; The nursing intervention suggestion is sent to the medical staff terminal.

9. The system according to claim 1, characterized in that Also includes: A model updating module, which is in communication with the risk assessment module, is used to: Regularly collect new patient data and risk assessment results; Use incremental learning methods to update risk assessment models to improve model adaptability and accuracy.

10. An intelligent nursing risk assessment and early warning method based on the system according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1, collects patients’ vital signs data and nursing operation data in real time through a variety of sensors, and obtains patients’ historical health data from the hospital data center; S2, preprocessing the collected real-time data and historical data, including data cleaning, missing value filling and standardization; S3, uses deep learning algorithms to build a risk assessment model, which is trained based on the preprocessed data; S4, using the trained risk assessment model to conduct real-time risk assessment of the patient’s current status, and output the risk level and specific risk factors; S5, dynamically adjust the risk warning threshold based on the risk assessment results; S6, when the risk assessment result exceeds the dynamic threshold, an early warning message is generated and sent to the medical staff terminal; S7, generating visual charts including vital signs trend charts, risk level change charts and early warning information summary, and pushing them to the medical staff terminal; S8, based on risk assessment results and early warning information, generates personalized nursing intervention recommendations; S9, regularly collect new patient data and risk assessment results, and use incremental learning methods to update the risk assessment model.

Images