AI risk identification model construction method for nursing monitoring

Abstract

The invention provides an AI risk identification model construction method for nursing monitoring, and relates to the technical field of nursing monitoring. Nursing data of a nursed object is collected; identifying the nursing data by using AI, and constructing a risk monitoring data mapping network to obtain risk monitoring data in the key network points; calculating correlation coefficients of various risk monitoring data, forming a positive correlation mapping network by the risk monitoring data positively correlated with the risk items, and forming a negative correlation mapping network by the risk monitoring data negatively correlated with the risk items; and constructing a risk identification model by using the negative correlation mapping network and the positive correlation mapping network, and performing risk identification by using the risk identification model.

Description

Technical Field

[0001] The present invention proposes a method for constructing an AI risk identification model for nursing monitoring, which relates to the technical field of nursing monitoring. Background Art

[0002] In the healthcare sector, the widespread use of electronic health record (EHR) systems has generated massive amounts of data, including basic patient information, vital signs (such as heart rate, blood pressure, and body temperature), laboratory test results, and nursing records.

[0003] Traditional nursing monitoring, due to human factors such as fatigue and inexperience, can lead to misjudgments or omissions of patient risks. Statistics show that a significant portion of medical errors are related to inadequate monitoring of patient conditions. Furthermore, modern healthcare increasingly emphasizes personalization, as each patient's condition, physical condition, lifestyle, and other factors vary. Traditional nursing monitoring methods based on universal standards struggle to meet individual needs.

[0004] The AI risk identification model can perform personalized risk assessments based on each patient's specific data. For example, for patients with diabetes, the model can combine multiple dimensions of data, such as blood sugar fluctuations, dietary records, and exercise habits, to tailor a risk warning plan. It can accurately identify risks such as hypoglycemia and diabetic ketoacidosis, thereby providing care services that better meet the individual patient's needs.

[0005] By building an AI-powered risk identification model, we can provide more objective and accurate risk assessments. Leveraging AI to build risk identification models allows for efficient processing of these massive amounts of data. AI models can quickly scan and analyze various patient data, uncovering valuable information such as potential trends in disease progression and the risk of complications, thereby assisting medical staff in making more timely and accurate care decisions. Summary of the Invention

[0006] In order to solve the above technical problems, the present invention proposes a method for constructing an AI risk identification model for nursing monitoring, comprising the following steps:

[0007] S1. Collect nursing data of the care recipients;

[0008] S2. Using AI to identify the nursing data and construct a risk monitoring data mapping network to obtain risk monitoring data in key locations and risk items in risk locations;

[0009] S3. Calculate the correlation coefficient of various risk monitoring data, form a positive correlation mapping network for the risk monitoring data that is positively correlated with the risk item, and form a negative correlation mapping network for the risk monitoring data that is negatively correlated with the risk item;

[0010] S4. Use the negative correlation mapping network and the positive correlation mapping network to build a risk identification model, and use the risk identification model to identify risks.

[0011] In a preferred embodiment, in step S3, the risk monitoring data is fitted, and the objective function L of the fitting is:

[0012]

[0013] Among them, p is the number of risk monitoring data, n is the number of risk items, and y i is the i-th risk item, x ij is the jth risk monitoring data corresponding to the i-th risk item, β j is the correlation coefficient corresponding to the j-th risk monitoring data, and μ is the correlation parameter;

[0014] Solve the above objective function through the optimization algorithm to obtain the correlation coefficient β j If the correlation coefficient β j is positive, indicating that the risk monitoring data x is ij With the i-th risk item y i There is a positive correlation; if the correlation coefficient β j is negative, indicating that the risk monitoring data x ij With the i-th risk item y i There is a negative correlation.

[0015] In a preferred embodiment, in step S4, for the positive correlation mapping network, the following positive correlation identification model is adopted:

[0016]

[0017] Among them, q(X′) is the risk value predicted by the positive correlation identification model, f(β′ j X′ ij ) is the correlation function, K1 is the number of risk monitoring data that is positively correlated with the risk term, X′ ij is the jth positive correlation risk monitoring data under the i-th risk item, β′ j is the corresponding correlation coefficient, and σ is the fluctuation parameter.

[0018] In a preferred embodiment, in step S4, the following negative correlation identification model is adopted for the negative correlation mapping network:

[0019]

[0020] Where P(X″) is the risk value predicted by the negative correlation identification model, K2 is the number of risk monitoring data that are negatively correlated with the risk item, δ is the balance parameter, and X″ ij is the jth negative correlation risk monitoring data under the i-th risk item, β″ j is the corresponding correlation coefficient.

[0021] In a preferred embodiment, the positive correlation recognition model and the negative correlation recognition model are fused to obtain a fused model:

[0022]

[0023] Among them, M1 and M2 are weights, and W is the comprehensive risk value.

[0024] In a preferred embodiment, in step S2, AI is used to identify the nursing data to achieve mapping between risk monitoring data and key points, and mapping between risk items and risk points. Key points and risk points are connected by point lines representing the relationship between the two. Key points, risk points and the point lines between them form a risk monitoring data mapping network.

[0025] In a preferred embodiment, high and low risk thresholds are set based on the comprehensive risk value W output by the fused model. When the comprehensive risk value W is greater than the high risk threshold, it is judged as high risk; when the comprehensive risk value W is between the high and low risk thresholds, it is judged as medium risk; when the comprehensive risk value W is less than the low risk threshold, it is judged as low risk.

[0026] Compared with the prior art, the present invention has the following beneficial technical effects:

[0027] In the process of building a risk monitoring data mapping network, we can accurately locate risk monitoring data at key locations and risk items at risk locations. This precise location helps medical staff focus on factors that may actually cause risks, rather than blindly searching for clues in massive amounts of data.

[0028] Constructing a risk identification model using negative and positive correlation mapping networks comprehensively considers multiple factors and their interrelationships, resulting in a more accurate and reliable model. Compared to traditional single-factor risk assessment methods, this model provides a more comprehensive risk assessment. When used in real-world risk identification, this risk identification model can quickly and efficiently provide risk warnings to medical staff.

[0029] The entire process is based on the specific nursing data of each patient, providing solid technical support for personalized care. Due to individual differences such as age, underlying medical conditions, and lifestyle, different patients face different risks and the relationships between risk factors. This approach allows for personalized risk identification based on each patient's specific circumstances. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0031] Figure 1 This is a flow chart of the method for constructing an AI risk identification model for nursing monitoring according to the present invention;

[0032] Figure 2 Schematic diagram of various types of nursing data of the present invention;

[0033] Figure 3 This is a partial structural diagram of the risk monitoring data mapping network of the present invention. DETAILED DESCRIPTION

[0034] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0035] In the drawings of the specific embodiments of the present invention, in order to better and more clearly describe the working principles of the various components in the system, the connection relationship of the various parts in the device is shown, which only clearly distinguishes the relative position relationship between the various components, and does not constitute a limitation on the signal transmission direction, connection sequence and size, dimension and shape of the components or structures.

[0036] Example 1

[0037] like Figure 1 FIG. 1 is a flow chart of a method for constructing an AI risk identification model for nursing monitoring according to the present invention. The method comprises the following steps:

[0038] S1. Collect nursing data of the care recipients.

[0039] Textual data such as nursing records, medical records, and nurse rounds contain valuable data sources for enriching information about the patients being cared for. These data are mostly unstructured. For example, a nursing record might be a description of the patient's daily care, written in natural language by the nurse, such as "The patient is in a poor mental state today, complaining of a headache, and has a temperature of 38.5°C in the afternoon. He has been given antipyretics as prescribed." Medical records contain written accounts of the patient's symptoms, medical history, and examination results. Ward rounds are also textual accounts of changes in the patient's condition and adjustments to treatment, as recorded during rounds.

[0040] In this step, the lexical analysis tool in AI natural language processing is used to segment and tag the words in the text. For example, for the sentence "The person being cared for has symptoms of coughing and sputum, accompanied by mild wheezing", it will be segmented into words such as "person being cared for", "appearance", "cough", "sputum", "symptoms", "accompanied by", "mild", and "wheezing", and their respective parts of speech (noun, verb, adjective, etc.) will be marked; then the named risk monitoring data recognition technology will be used to identify risk monitoring data with specific meanings in the text, such as names of people, names of diseases, names of symptoms, names of drugs, etc. After the above analysis, it is possible to accurately identify risk monitoring data representing symptoms such as "coughing", "sputum", and "wheezing", as well as possible implicit risk monitoring data of the person being cared for (although the text may use a general reference such as "person being cared for", it can be associated with the actual individual person being cared for in the specific context).

[0041] By developing information extraction methods based on grammatical rules, semantic rules, and machine learning models, we can extract risk monitoring data fragments from text. For example, we can set rules to extract content related to "symptom descriptions." Whenever a text contains a guide word like "appears" or "accompanied by" followed by a symptom noun, we extract it as symptom description information. For diagnostic results, we can extract disease names following keywords like "diagnosed as" or "confirmed." For treatment processes, we focus on information such as treatment methods and medications following words like "give" or "adopt."

[0042] The extracted nursing data is organized according to a pre-defined structured framework. For example, a table structure is constructed with columns such as the nursing patient number, date, symptom description, diagnosis, and treatment process. The corresponding information extracted from different texts is sequentially filled into the corresponding cells. This transforms the originally scattered and disorganized unstructured text information into an organized structured data format that is easier for computers to analyze and process.

[0043] like Figure 2As shown in the figure, it is a schematic diagram of various types of nursing data extracted, where the frequency is the number of times counted per month, and the percentage is the proportion of the frequency of each type of nursing data to the frequency of all nursing data, so as to facilitate the statistics of the nursing data with the highest frequency.

[0044] S2. Use AI to identify the nursing data and build a risk monitoring data mapping network to obtain risk monitoring data in key outlets and risk items in risk outlets.

[0045] AI is used to extract various nursing data from the nursing records of the care recipients, including diseases (such as coronary heart disease and diabetes), symptoms (such as chest pain, polydipsia, etc.), nursing methods (such as drug monitoring, nursing records, etc.), and risk factors (such as hypertension, smoking, etc.). At the same time, the relationships between these nursing data are determined. For example, there is a "symptom manifestation" relationship between "coronary heart disease" and "chest pain", a "treatment method" relationship between "coronary heart disease" and "drug treatment", and a "risk association" relationship between "hypertension" and "coronary heart disease".

[0046] The AI database stores the extracted connected nursing data and their connections. Key nodes represent risk monitoring data, and lines connecting nodes represent the relationships between them. This constructs a medical knowledge network, also known as the risk monitoring data mapping network. Nursing data without connections is deleted or filtered and not reflected in the risk monitoring data mapping network.

[0047] like Figure 3 As shown in the figure, in the risk monitoring data mapping network, the key node of "chest pain" is connected to the risk node of "coronary heart disease" through the "trigger" node connection line, the key node of "arrhythmia" is connected to the risk node of "coronary heart disease" through the "cause" node connection line, and the key node of "hypertension" is connected to the risk node of "coronary heart disease" through the "inducing risk" node connection line, which intuitively presents the relationship between various knowledge elements.

[0048] Specifically include:

[0049] Extract nursing information of the cared persons, realize the mapping between risk monitoring data and key outlets, and the mapping between risk items and risk outlets, and build a risk monitoring data mapping network.

[0050] Let W represent the risk monitoring data mapping network, where Z is the set of key nodes (containing risk monitoring data, such as symptoms, indicators, age, status, and other related health information), and L is the set of node connections (representing various relationships between risk monitoring data, such as "cause," "care," and "association." Each item of risk monitoring data is mapped and matched to the corresponding key node in the risk monitoring data mapping network to ensure that the corresponding key node representing this information can be accurately found in the risk monitoring data mapping network. For example, if the person being cared for is diagnosed with diabetes, the key node "diabetes" is found in the risk monitoring data mapping network; if the person has foot ulcer symptoms, the symptom key node "foot ulcer" is found, and so on.

[0051] Construct risk points. For example, for the risk item "diabetes", find the key points connected to it through the "easy to cause" relationship, and map the risk item "diabetes" to the risk point. Traverse the point connection lines connected to each key point in the key point set Z of the care object in the risk monitoring data mapping network and the corresponding risk monitoring data, and filter according to the pre-defined risk association relationship type (such as "easy to cause", "increase the risk of disease" and other relationship types).

[0052] All risk points are finally included in the risk point set R = {r1, r2, ...r m}, the same operation is performed on other elements in the key network set Z of the care object, and finally they are all connected to the risk network set R, where m represents the number of risk networks.

[0053] S3. Calculate the correlation coefficients of various risk monitoring data, form a positive correlation mapping network for the risk monitoring data that is positively correlated with the risk items, and form a negative correlation mapping network for the risk monitoring data that is negatively correlated with the risk items.

[0054] The risk monitoring data of various risk monitoring data are fitted, and the fitting objective function L is:

[0055]

[0056] Among them, p is the number of risk monitoring data, n is the number of risk items, and y i is the i-th risk item, x ij is the jth risk monitoring data corresponding to the i-th risk item, β j is the correlation coefficient corresponding to the j-th risk monitoring data, and μ is the correlation parameter.

[0057] Solve the above objective function through the optimization algorithm to obtain the correlation coefficient β j If the correlation coefficient β j is positive, indicating that the risk monitoring data x isij With the i-th risk item y i There is a positive correlation. That is, as the risk monitoring data x ij The increase of the i-th risk item y i For example, if the correlation coefficient for age is positive, it means that the older you are, the higher the risk.

[0058] If the correlation coefficient β j is negative, indicating that the risk monitoring data x ij With the i-th risk item y i There is a negative correlation. That is, as the risk monitoring data x ij For example, if the correlation coefficient for a nursing indicator is negative, it may mean that increasing the nursing indicator will reduce the risk.

[0059] Based on the control relationship, the risk monitoring data mapping network is separated. The risk monitoring data that is positively correlated with the risk item forms a positive correlation mapping network, and the risk monitoring data that is negatively correlated with the risk item forms a negative correlation mapping network.

[0060] S4. Use the negative correlation mapping network and the positive correlation mapping network to build a risk identification model, and use the risk identification model to identify risks.

[0061] S41. For the positive correlation mapping network, construct a positive correlation identification model.

[0062] For the positive correlation mapping network, the following positive correlation identification model is adopted:

[0063]

[0064] Among them, q(X′) is the risk value predicted by the positive correlation identification model, f(β′ j X′ ij ) is the correlation function, K1 is the number of risk monitoring data that is positively correlated with the risk term, X′ ij is the jth positive correlation risk monitoring data under the i-th risk item, β′ j is the corresponding correlation coefficient; σ is the fluctuation parameter, which reflects the degree of fluctuation of the risk factor. For example, if the nursing risk is affected by multiple unstable factors, σ 2 The value may be larger.

[0065] For example, when p=3, The age (age), systolic blood pressure (sbp) and blood glucose level (glu) of the care recipient are the risk monitoring data association coefficients, and α is the error parameter.

[0066] It should be noted that in actual nursing care, positively correlated risk monitoring data may include the patient's age (the older the patient, the more pronounced the decline in physical function and the higher the health risk), the number of underlying diseases (for example, those with multiple conditions such as hypertension, diabetes, and heart disease have an increased risk of complications), and the frequency of recent condition fluctuations (frequent blood pressure instability and abnormal blood sugar levels). These factors are positively correlated with the health risks faced by patients. The AI risk identification model, based on a large amount of historical nursing data, learns the relationship between these factors and the probability of adverse health events, constructing a model that reflects the trend of increasing positive risk.

[0067] S42. For the negative correlation mapping network, a negative correlation identification model is constructed.

[0068] For the negative correlation mapping network, the following negative correlation identification model is adopted:

[0069]

[0070] Where P(X″) is the risk value predicted by the negative correlation identification model, K2 is the number of risk monitoring data that are negatively correlated with the risk item, δ is the balance parameter, and X″ ij is the jth negative correlation risk monitoring data under the i-th risk item, β″ j is the corresponding correlation coefficient, where K1+K2=p.

[0071] It's important to note that in actual nursing care, negatively correlated risk monitoring data focuses on factors such as patient adherence to medical advice (patients who strictly follow doctor's orders, attend regular follow-up appointments, and adhere to dietary and exercise recommendations tend to have more stable conditions), timeliness of nursing interventions (for example, if a patient experiences minor discomfort, prompt response from the nurse and appropriate measures can effectively prevent worsening of the condition), and family support (family members actively participating in care and assisting in monitoring the patient's lifestyle can reduce risk). These factors are negatively correlated with health risks, and a model based on historical data has been constructed to quantify their impact on risk reduction. Statistics show that patients with high adherence to medical advice have a 30% lower complication rate than those who do not; and responding to nursing interventions within one hour reduces the likelihood of worsening of the condition by 25%.

[0072] S43. Fusion of positive correlation identification model and negative correlation identification model:

[0073]

[0074] Where W is the comprehensive risk value, M1 and M2 are weights, which can be determined through experiments or based on domain knowledge. Preferably, M1 = 0.6 and M2 = 0.4.

[0075] It's important to note that in complex risk identification systems, risk predictions are often influenced by the interaction of multiple factors. A positive correlation identification model can capture monitoring data that is positively correlated with increased risk, while a negative correlation identification model focuses on monitoring data that reduces risk. By integrating these two models, we can comprehensively consider various factors that contribute to both increased and decreased risk, avoiding forecast bias caused by focusing on only one factor in each direction.

[0076] It's worth noting that a single positive or negative correlation model can only provide partial risk information. A fused model can examine risk from multiple perspectives. For example, a positive correlation model might focus on factors that increase risk, such as a patient's age or the presence of multiple underlying conditions; whereas a negative correlation model can consider factors that reduce risk, such as a patient's good lifestyle habits and timely nursing interventions. This fusion generates a comprehensive assessment encompassing multidimensional risk information, providing caregivers with a more complete picture. Based on the fused model, caregivers can more accurately determine the patient's condition development risk and develop more appropriate care plans.

[0077] Risk identification and assessment:

[0078] Based on the comprehensive risk value W output by the fused model, high and low risk thresholds are set. When the comprehensive risk value W is greater than the high-risk threshold, the risk is determined to be high; when the comprehensive risk value W is between the high and low risk thresholds, the risk is determined to be medium; and when the comprehensive risk value W is less than the low-risk threshold, the risk is determined to be low. By inputting new nursing monitoring data into the trained fused model, risk identification and assessment can be achieved. Simultaneously, the fused model needs to be continuously updated and optimized, adjusting network parameters and fusion weights based on emerging data and actual risk conditions.

[0079] Example 2

[0080] Table 1 below shows the risk monitoring data mapping network construction and AI-assisted risk identification effects and practical application value for different cases.

[0081] Table 1

[0082]

[0083]

[0084] The system monitors falls and other common risk factors, including visual impairment, poor balance, and the use of sedatives. This AI-based risk identification model helps prevent falls for elderly care recipients by integrating various risk factors, such as "visual impairment," "poor balance," and "taking sedatives," into the risk monitoring data. This model provides supervision for elderly care recipients, including those with multiple chronic diseases, taking multiple walking aids, and the ability to fall. For example, if a care recipient has poor vision, is taking medication that may cause dizziness, and is taking medication, this information is mapped to key points in the risk monitoring data mapping network, corresponding to "visual impairment" and "taking sedatives." Risk monitoring data is used to map the risk factors.

[0085] Fall. Relational reasoning of shooting net, "visual impairment" and "service

[0086] The key points of "using sedatives" and "falls" are

[0087] There is an "increased risk" relationship between outlets.

[0088] The system identifies that the person being cared for has a high risk of falling

[0089] risk.

[0090] In one embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0091] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the steps in the above-mentioned method embodiments when executed by a processor.

[0092] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps of each of the above-described method embodiments.

[0093] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. The database involved in the embodiments provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited to this. The processor involved in the embodiments provided in this application may be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, etc., but is not limited to this.

[0094] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0095] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for constructing an AI risk identification model for nursing monitoring, characterized in that: The steps include: S1. Collect nursing data of the care recipients; S2. Using AI to identify the nursing data and construct a risk monitoring data mapping network to obtain risk monitoring data in key locations and risk items in risk locations; S3. Calculate the correlation coefficients of various risk monitoring data, form a positive correlation mapping network for the risk monitoring data that is positively correlated with the risk item, and form a negative correlation mapping network for the risk monitoring data that is negatively correlated with the risk item; S4. Use the negative correlation mapping network and the positive correlation mapping network to build a risk identification model, and use the risk identification model to identify the risks of the monitored objects.

2. The method for constructing an AI risk identification model for nursing monitoring according to claim 1, characterized in that: In step S3, the risk monitoring data is fitted, and the objective function L of the fitting is: Among them, p is the number of risk monitoring data, n is the number of risk items, and y i is the i-th risk item, x ij is the jth risk monitoring data corresponding to the i-th risk item, β j is the correlation coefficient corresponding to the j-th risk monitoring data, and μ is the correlation parameter; Solve the above objective function through the optimization algorithm to obtain the correlation coefficient β j If the estimated value of the correlation coefficient β j is positive, indicating that the risk monitoring data x is ij With the i-th risk item y i There is a positive correlation; if the correlation coefficient β j is negative, indicating that the risk monitoring data x ij With the i-th risk item y i There is a negative correlation.

3. The method for constructing an AI risk identification model for nursing monitoring according to claim 2, characterized in that: In step S4, for the positive correlation mapping network, the following positive correlation identification model is adopted: Among them, q(X′) is the risk value predicted by the positive correlation identification model, f(β′ j X′ ij ) is the correlation function, K1 is the number of risk monitoring data that is positively correlated with the risk term, X′ ij is the jth positive correlation risk monitoring data under the i-th risk item, β′ j is the corresponding correlation coefficient, and σ is the fluctuation parameter.

4. The method for constructing an AI risk identification model for nursing monitoring according to claim 3, characterized in that: In step S4, for the negative correlation mapping network, the following negative correlation identification model is adopted: Where P(X″) is the risk value predicted by the negative correlation identification model, K2 is the number of risk monitoring data that are negatively correlated with the risk item, δ is the balance parameter, and X″ ij is the jth negative correlation risk monitoring data under the i-th risk item, β″ j is the corresponding correlation coefficient.

5. The method for constructing an AI risk identification model for nursing monitoring according to claim 4, characterized in that: The positive correlation recognition model and the negative correlation recognition model are integrated to obtain the fused model: Among them, M1 and M2 are weights, and W is the comprehensive risk value.

6. The method for constructing an AI risk identification model for nursing monitoring according to claim 1, characterized in that: In step S2, AI is used to identify the nursing data to achieve mapping between risk monitoring data and key points, and mapping between risk items and risk points. Key points and risk points are connected by point lines representing the relationship between the two. Key points, risk points and the point lines between them form a risk monitoring data mapping network.

7. The method for constructing an AI risk identification model for nursing monitoring according to claim 5, characterized in that: According to the comprehensive risk value W output by the fused model, high and low risk thresholds are set. When the comprehensive risk value W is greater than the high risk threshold, it is judged as high risk; when the comprehensive risk value W is between the high and low risk thresholds, it is judged as medium risk; when the comprehensive risk value W is less than the low risk threshold, it is judged as low risk.

Images