Nursing validity analysis method based on big data

Abstract

The invention relates to the technical field of data analysis, in particular to a big data-based nursing effectiveness analysis method, which comprises the following steps of: acquiring historical nursing data of each user category, user use data of each nursing category and user information of each user, and forming a database; establishing an association model, and learning the historical nursing data and the user use data according to the user information; establishing an understanding model, setting a user portrait corresponding to each user, and establishing a sub-database of the corresponding user; determining a corresponding understanding model accuracy rate according to the user nursing result; according to the accuracy rate of the understanding model, iterating the understanding model and the association model according to the accuracy rate; according to the method, the database of the user category and the database of the nursing category are tested, and the accuracy is verified in real time to adjust each model, so that the model failure problem caused by the individual difference of the users is reduced while the adaptability of the models is effectively improved, and the accuracy of model prediction is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and particularly to a method for analyzing the effectiveness of nursing based on big data. Background Art

[0002] Traditional nursing safety management is mostly retrospective and experience-based management, that is, the cause analysis is carried out after the event, lacking data support. According to modern management concepts, safety management should first implement the idea of "prevention first". Only by changing the handling after nursing unsafe events into active prevention before they occur can it be an important strategy for ensuring safety management and a scientific management. Risk pre-control safety management shifts the focus of safety management work forward, changing the previous safety management's attention from the accident itself or hidden dangers to the attention of hazard sources, and changing the passive accident handling or hidden danger investigation into active safety management. Hospitals generate a large amount of information data every day. These information include various types of data such as semantics and time. They truly record and reflect the daily work and nursing activities of nurses, and are the most objective and real-time generated data. If these data are deeply analyzed and the nursing activities reflected by them are objectively presented, and the large amount of basic data stored in the system is fully utilized, the quality of clinical nursing can be objectively and fairly reflected.

[0003] Chinese Patent Authorization Publication No.: CN118626678B discloses a method and system for analyzing nursing data based on artificial intelligence. When obtaining multiple real-time data segments in a real-time data segment combination, the current data segment core point group obtained by bucketing the multiple real-time data segments and the previous data segment core point group indicating the bucketing results in the past are obtained. At the same time, the target data segment core point group is obtained by updating according to the previous data segment core point group and the current data segment core point group, so as to prevent the initial data segment and the real-time data segment in the multiple data segment bucketing in the existing bucketing from being subjected to similarity calculation and bucketing, improving the bucketing speed and the efficiency of data processing, and ensuring the high real-time response requirements during patient care.

[0004] However, the above method has the following problems: it cannot perform forward-looking prediction, and at the same time, when the model fails to achieve the expected effect, it cannot adjust the model according to a single user, resulting in the problem of loss of accuracy in the judgment of nursing results. Summary of the Invention

[0005] Therefore, the present invention provides a method for analyzing the effectiveness of nursing based on big data to overcome the problems in the prior art that forward-looking prediction cannot be performed, and at the same time, when the model fails to achieve the expected effect, the model cannot be adjusted according to a single user, thereby reducing the accuracy rate of the judgment of nursing results.

[0006] To achieve the above object, the present invention provides a method for analyzing the effectiveness of nursing based on big data, including: Collecting historical nursing data of each user category, user usage data of each nursing category, and user information of each user, and forming a database; Establishing an association model, and learning the historical nursing data and user usage data according to the user information; Establishing an understanding model, setting a user portrait corresponding to each user, and establishing a sub-database corresponding to the user; Verifying user data, and generating a number of user nursing results corresponding to the user data; Determining the accuracy rate of the corresponding understanding model according to the user nursing results; According to the accuracy rate of the understanding model, iterating the understanding model and the association model according to the accuracy rate; When the preset accuracy rate is reached, stop iterating, and output the association model and the understanding model.

[0007] Further, for a single user, the step of establishing the association model includes: Determining the user category corresponding to the user; Querying the nursing category corresponding to the user category according to the user category; Fuzzifying the nursing category, and generating a corresponding fuzzy nursing category; Wherein, the fuzzy nursing category includes nursing categories with a frequency of occurrence not less than a preset frequency among several nursing categories applied by several users of the same user category.

[0008] Further, for a single user, the step of establishing the understanding model includes: Determining the historical nursing data and user category of the user; Determining the user information corresponding to the user in different nursing categories according to the historical nursing data; Comprehensively screening the nursing categories with the user information to form a sub-database corresponding to the user.

[0009] Further, when collecting the historical nursing data of each user category, it includes: Collecting several nursing information and corresponding nursing results of the user at regular intervals; Generating corresponding user records according to the nursing information and nursing results; Saving the same nursing information and / or nursing results as historical nursing data; Wherein, the period is the period from a single nursing information to the corresponding nursing result generated by the nursing, and this period is only recorded by the appearance of the nursing result.

[0010] Further, the user information at least includes the doctor's order sheet and the nursing record sheet corresponding to the user when completing a single cycle; Among them, for a single nursing category, the corresponding several doctor's order sheets and the nursing record sheets corresponding to each doctor's order sheet should be the same semantically.

[0011] Further, the historical nursing data at least includes nursing electronic medical records, including temperature sheets, doctor's order sheets, and nursing record sheets.

[0012] Further, the steps of verifying the user data include: Select several users and several incomplete cycles as the verification batch; Use the association model and the understanding model to generate predicted user information corresponding to the verification batch; Compare the predicted user information with the corresponding actual user information to generate the accuracy rate of this verification batch.

[0013] Further, when the accuracy rates of both the association model and the understanding model reach the preset accuracy rate, it is determined that the association model and the understanding model have completed iteration.

[0014] Further, there is also a sampling detection strategy, including: During the operation of the association model and the understanding model, perform sampling detection at a preset ratio; When the accuracy rate of the sampling detection is not greater than the preset accuracy rate, it is determined that the association model and the understanding model fail; When the association model and the understanding model fail, train based on the failed association model and understanding model until the preset accuracy rate is reached.

[0015] Further, the preset accuracy rate is not less than 80%, and the preset ratio is not less than 30%.

[0016] Compared with the prior art, the beneficial effects of the present invention are that by setting the association model and the understanding model, the databases of user categories and nursing categories are tested, and the accuracy rate is verified in real time to adjust each model. While effectively improving the adaptability of the model, the problem of model failure caused by individual differences of users is reduced, thereby effectively improving the accuracy rate of model prediction.

[0017] Further, by using the method of fuzzifying the recognition of the association model to expand the recognition range of the association model, while effectively avoiding the problem of the decrease in the accuracy rate of the model caused by differences in recording formats, and by using the method of fuzzifying nursing categories, the accuracy rate of model prediction is effectively improved.

[0018] Furthermore, by establishing an understanding model and periodically adjusting the model, the model can be better adapted to a number of users corresponding to a certain user profile, thereby improving the adaptation degree to a single user and further enhancing the accuracy of model prediction.

[0019] Furthermore, by recording a single nursing and the occurrence of the result corresponding to the single nursing as a cycle, the problem of model abnormality caused by nursing for a short or long time is avoided. While effectively improving the anti-abnormality performance of the model, the adaptation performance of the model is improved, and thus the accuracy of model prediction is effectively enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a flowchart of the method for analyzing the effectiveness of nursing based on big data according to the present invention; Figure 2 is a flowchart of establishing an association model in an embodiment of the present invention; Figure 3 is a flowchart of establishing an understanding model in an embodiment of the present invention; Figure 4 is a schematic structural diagram of data mining in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] In order to make the objectives and advantages of the present invention clearer, the present invention will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0022] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present invention and do not limit the protection scope of the present invention.

[0023] It should be noted that in the description of the present invention, the terms indicating directions or positional relationships such as "upper", "lower", "left", "right", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings. This is only for convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention.

[0024] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installation", "connection", and "connection" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0025] Please refer to Figure 1 shown in the figure, which is a flowchart of the method for analyzing the effectiveness of nursing based on big data according to the present invention, including: Step S1, collect historical nursing data of each user category, user usage data of each nursing category, and user information of each user, and form a database; Step S2r, establish an association model, and learn from the historical nursing data and user usage data according to the user information; Step S2u, establish an understanding model, set a user portrait corresponding to each user, and establish a sub-database for the corresponding user; Step S3, verify the user data, and generate several user nursing results corresponding to the user data; Step S4, determine the accuracy rate of the corresponding understanding model according to the user nursing results; Step S5, according to the accuracy rate of the understanding model, iterate the understanding model and the association model according to the accuracy rate; Step S6, when the preset accuracy rate is reached, stop the iteration, and output the association model and the understanding model.

[0026] By setting the association model and the understanding model, the databases of user categories and nursing categories are tested, and the accuracy rate is verified in real time to adjust each model. While effectively improving the adaptability of the model, the problem of model failure caused by individual differences of users is reduced, thereby effectively improving the accuracy rate of model prediction.

[0027] Example 1: In the geriatric ward, falls are common nursing safety events that may lead to serious consequences. Through data mining and model iteration, high-risk patients are identified in advance and preventive measures are taken.

[0028] Specific steps and applications of using the above solution: Historical nursing data: Collect the records of fall events of patients in the geriatric ward in the past year, including fall time, location, cause, etc.

[0029] User usage data: Record the implementation of fall prevention measures by nurses, such as whether fall prevention assessment is carried out, whether anti-slip mats are used, etc.

[0030] User information: The age, gender, medical history, and physical condition (such as vision, balance ability, etc.) of the patient.

[0031] Database: Integrate the above data into a database for subsequent analysis.

[0032] Association Model: Use the Apriori algorithm to analyze historical nursing data and user usage data to identify factors related to fall events, such as advanced age, poor eyesight, and non-use of anti-slip mats.

[0033] Learning Process: The model learns that the fall risk of patients over 80 years old with poor eyesight increases significantly.

[0034] Understanding Model: Based on the results of the association model, establish an understanding model to predict the fall risk of patients.

[0035] User Portrait: Generate a user portrait for each patient and label their fall risk level (high, medium, low).

[0036] Sub-database: Establish a sub-database for high-risk patients to record their detailed information and preventive measures.

[0037] Data Verification: Conduct a fall risk assessment for newly admitted elderly patients and generate nursing outcomes (such as recommending anti-slip mats and regular inspections for high-risk patients).

[0038] Accuracy Evaluation: Calculate the accuracy of the understanding model by comparing actual fall events with model prediction results. For example, 80% of the patients predicted to be at high risk actually had a fall.

[0039] Iteration Process: Adjust the parameters of the association model and the understanding model based on the accuracy. For example, add balance ability as a risk factor and retrain the model.

[0040] Stop Iteration: Stop iterating when the accuracy of the understanding model reaches over 90%.

[0041] Output Model: The final output association model and understanding model can be used for fall risk assessment and preventive management in the geriatric ward.

[0042] Example 2: In the neonatal ward, infection is an important hidden danger to nursing safety. Through data mining and model iteration, identify newborns at high risk of infection in advance and take preventive measures.

[0043] Specific Steps and Applications of Using the Above Scheme: Historical Nursing Data: Collect infection event records in the neonatal ward in the past year, including infection type, time, transmission route, etc.

[0044] User Usage Data: Record the implementation of infection prevention measures by nurses, such as hand hygiene compliance rate, isolation measures, etc.

[0045] User Information: Birth weight, gestational age, immune status, etc. of newborns.

[0046] Database: Integrate the above data into a database for subsequent analysis.

[0047] Association model: Use the Apriori algorithm to analyze historical nursing data and user usage data to identify factors related to infection events, such as low birth weight, non - strict implementation of hand hygiene, etc.

[0048] Learning process: The model learns that the infection risk of newborns with a birth weight below 2500 grams and a low hand hygiene implementation rate is significantly increased.

[0049] Understanding model: Based on the results of the association model, establish an understanding model to predict the infection risk of newborns.

[0050] User portrait: Generate a user portrait for each newborn and label their infection risk level (high, medium, low).

[0051] Sub - database: Establish a sub - database for high - risk newborns to record their detailed information and preventive measures.

[0052] Data verification: Conduct an infection risk assessment for newly admitted newborns and generate nursing outcomes (such as recommending enhanced hand hygiene, isolation, etc. for high - risk newborns).

[0053] Accuracy evaluation: Calculate the accuracy of the understanding model by comparing actual infection events with model prediction results. For example, among the newborns predicted as high - risk by the model, 75% actually had an infection.

[0054] Iterative process: Adjust the parameters of the association model and the understanding model according to the accuracy. For example, add immune status as a risk factor and retrain the model.

[0055] Stop iteration: When the accuracy of the understanding model reaches over 85%, stop the iteration.

[0056] Output model: The finally output association model and understanding model can be used for infection risk control and preventive management in the newborn ward.

[0057] Example 3: The operating room is a high - risk area in the hospital. Nursing safety events such as surgical instrument omission, infection, etc. may lead to serious consequences. Through data mining and model iteration, identify high - risk surgeries in advance and take preventive measures.

[0058] Specific steps and applications of using the above solutions: Historical nursing data: Collect the records of nursing safety events in the operating room in the past year, including event types, times, surgical types, etc.

[0059] User usage data: Record the implementation of operations such as nurses' inventory and disinfection of surgical instruments.

[0060] User information: The age of the surgical patient, the type of surgery, the surgery time, etc.

[0061] Database: Integrate the above data into a database for subsequent analysis.

[0062] Association model: Use the Apriori algorithm to analyze historical nursing data and user usage data to find factors related to nursing safety events, such as long surgeries and irregular instrument counts.

[0063] Learning process: The model learns that for surgeries with a surgery time exceeding 4 hours and an irregular instrument count, the risk of nursing safety events increases significantly.

[0064] Understanding model: Based on the results of the association model, establish an understanding model to predict the risk of surgical nursing safety events.

[0065] User portrait: Generate a user portrait for each surgical patient and label their risk levels of nursing safety events (high, medium, low).

[0066] Sub-database: Establish a sub-database for high-risk surgeries to record their detailed information and preventive measures.

[0067] Data verification: Conduct a risk assessment of nursing safety events for upcoming surgeries and generate nursing results (such as for high-risk surgeries, it is recommended to strengthen instrument counts and increase circulating nurses).

[0068] Accuracy evaluation: Calculate the accuracy of the understanding model by comparing the actual nursing safety events and the model prediction results. For example, among the surgeries predicted as high-risk by the model, 80% actually had nursing safety events.

[0069] Iterative process: Adjust the parameters of the association model and the understanding model according to the accuracy. For example, add the type of surgery as a risk factor and retrain the model.

[0070] Stop iteration: When the accuracy of the understanding model reaches over 90%, stop the iteration.

[0071] Output models: The finally output association model and understanding model can be used for the prevention and management of nursing safety events in the operating room.

[0072] Please refer to Figure 2 As shown, it is the flowchart for establishing an association model in an embodiment of the present invention, including: Step Sr1, determine the user category corresponding to the user; Step Sr2, query the nursing category corresponding to the user category according to the user category; Step Sr3, fuzzify the nursing category and generate the corresponding fuzzy nursing category; Among them, the fuzzy nursing categories include the nursing categories that appear no less than a preset frequency in several nursing categories applied by several users with the same user category.

[0073] The method of using the recognition of the association model for fuzzification to expand the recognition scope of the association model can effectively avoid the problem of the decline in the model accuracy rate caused by differences in record formats, and at the same time, by using the method of fuzzy nursing categories, the accuracy rate of model prediction is effectively improved.

[0074] For Embodiment 1, the fuzzification processing steps include: User category: The patients in the geriatric ward are divided into two categories: "elderly patients with poor eyesight" and "other patients".

[0075] Nursing category: For "elderly patients with poor eyesight", query their corresponding nursing categories, such as "fall prevention assessment", "using non-slip mats", "regular rounds", etc.

[0076] Fuzzy nursing category: Fuzzify the nursing categories such as "fall prevention assessment", "using non-slip mats", "regular rounds", etc. For example, fuzzify "fall prevention assessment" into "fall risk assessment", "using non-slip mats" into "anti-slip measures", and "regular rounds" into "frequent rounds".

[0077] Fuzzification result: The fuzzified nursing categories include "fall risk assessment", "anti-slip measures", "frequent rounds", etc. These fuzzy nursing categories appear with a relatively high frequency among the high-risk patients in the geriatric ward and can effectively avoid the problem of the decline in the model accuracy rate caused by differences in record formats.

[0078] For Embodiment 2, the fuzzification processing steps include: User category: Newborns are divided into two categories: "newborns with low birth weight and poor immune status" and "other newborns".

[0079] Nursing category: For "newborns with low birth weight and poor immune status", query their corresponding nursing categories, such as "hand hygiene", "isolation measures", "immune support", etc.

[0080] Fuzzy nursing category: Fuzzify "hand hygiene" into "infection prevention measures", "isolation measures" into "isolation protection", and "immune support" into "immune enhancement".

[0081] Fuzzification result: The fuzzified nursing categories include "infection prevention measures", "isolation protection", "immune enhancement", etc. These fuzzy nursing categories appear with a relatively high frequency among the high-risk newborns in the neonatal ward and can effectively avoid the problem of the decline in the model accuracy rate caused by differences in record formats.

[0082] For Example 3, the fuzzification process steps include: User category: The surgeries are divided into two categories: "long surgeries with irregular instrument counting" and "other surgeries".

[0083] Nursing category: For "long surgeries with irregular instrument counting", query its corresponding nursing categories, such as "instrument counting", "disinfection measures", "circulating nurse supervision", etc.

[0084] Fuzzy nursing category: Fuzzify "instrument counting" into "instrument management", "disinfection measures" into "infection control", and "circulating nurse supervision" into "surgery supervision".

[0085] Fuzzification result: The fuzzified nursing categories include "instrument management", "infection control", "surgery supervision", etc. These fuzzy nursing categories appear frequently in high-risk surgeries in the operating room and can effectively avoid the problem of decreased model accuracy caused by differences in recording formats.

[0086] Please refer to Figure 3 As shown, it is a flowchart for establishing an understanding model in an embodiment of the present invention. For a single user, the steps for establishing the understanding model include: Step Su1, determine the historical nursing data and user category of the user; Step Su2, determine the user information corresponding to the user under different nursing categories according to the historical nursing data; Step Su3, comprehensively screen the nursing categories based on the user information to form a sub-database corresponding to the user.

[0087] By using the established understanding model and periodically adjusting the model, the model can be made more adaptable to several users corresponding to a certain user portrait, thereby improving the adaptability to a single user and further enhancing the accuracy of model prediction.

[0088] Specifically, when collecting the historical nursing data of each user category, it includes: Collect several nursing information of the user and the corresponding nursing results at regular intervals; Generate corresponding user records according to the nursing information and nursing results; Save the same nursing information and / or nursing results as historical nursing data; Among them, the cycle is the cycle from a single nursing information to the corresponding nursing result generated by this nursing, and this cycle is only recorded based on the appearance of the nursing result.

[0089] By recording a single care and the corresponding result occurrence as a cycle, the problem of model anomalies caused by care for shorter or longer periods is avoided. While effectively improving the model's anti-anomaly performance, the model's adaptation performance is enhanced, and thus the accuracy of model prediction is effectively improved.

[0090] For Example 1, the preservation of its corresponding historical care data is as follows: Cycle definition: Starting from a single care operation (such as a fall prevention assessment), to the cycle of the corresponding care result (such as the patient did not fall or fell) generated by this care operation.

[0091] Collection of care information: Each time a fall prevention assessment is performed on a patient, care information such as the patient's age, vision condition, whether a non-slip mat is used, and whether regular inspections are carried out is recorded.

[0092] Recording of care results: At the end of the care cycle, record whether the patient fell and the specific circumstances of the fall (such as the time, location, and degree of injury of the fall).

[0093] Generate user records User records: Integrate the collected care information and the corresponding care results into user records. For example: Patient A, 82 years old, poor vision, uses a non-slip mat, regular inspections, did not fall during the care cycle.

[0094] Patient B, 85 years old, normal vision, did not use a non-slip mat, irregular inspections, fell during the care cycle, slightly injured.

[0095] Save historical care data Historical care data: Save the same care information and / or care results as historical care data. For example, save the records of all patients who use non-slip mats and have regular inspections as a group of historical care data, and save the records of all patients who do not use non-slip mats and have irregular inspections as another group of historical care data.

[0096] Data storage format: { "Care information": { "Age": 82, "Vision condition": "Poor", "Use non-slip mat": "Yes", "Regular inspections": "Yes" }, "Care results": { "Whether fell": "No", "Fall time": null, "Falling location": null, "Degree of injury": null } } For Example 2, the preservation of its corresponding historical nursing data is as follows: Cycle definition: Starting from a single nursing operation (such as hand hygiene execution), to the cycle corresponding to the nursing result generated by this nursing operation (such as the newborn not getting infected or getting infected).

[0097] Nursing information collection: Record the hand hygiene execution situation each time, including the hand hygiene execution rate of the operating nurse, whether isolation measures are used, the birth weight of the newborn, immune status and other nursing information.

[0098] Nursing result record: At the end of the nursing cycle, record whether the newborn gets infected and the specific situation of the infection (such as the type of infection, infection time, etc.).

[0099] Generate user records User records: Integrate the collected nursing information and the corresponding nursing results into user records. For example: Newborn A, with a birth weight of 2400 grams, poor immune status, a hand hygiene execution rate of 100%, using isolation measures, and not getting infected during the nursing cycle.

[0100] Newborn B, with a birth weight of 2600 grams, normal immune status, a hand hygiene execution rate of 80%, not using isolation measures, getting infected during the nursing cycle, and the type of infection is respiratory tract infection.

[0101] Save historical nursing data Historical nursing data: Save the same nursing information and / or nursing results as historical nursing data. For example, save all the records of newborns with a hand hygiene execution rate of 100% and using isolation measures as a group of historical nursing data, and save all the records of newborns with a hand hygiene execution rate below 90% and not using isolation measures as another group of historical nursing data.

[0102] Data storage format: { "Nursing information": { "Birth weight": 2400, "Immune status": "poor", "Hand hygiene execution rate": 100, "Using isolation measures": "yes" }, "Nursing result": { "Whether getting infected": "no", "Type of infection": null, "Infection time": null } } For Example 3, the preservation of its corresponding historical nursing data is as follows: Cycle definition: Starting from a single nursing operation (such as instrument inventory), to the cycle corresponding to the nursing result generated by this nursing operation (such as no nursing safety event occurred during the operation or a nursing safety event occurred).

[0103] Nursing information collection: Record the inventory situation of surgical instruments each time, including nursing information such as the number of instrument inventories, operation time, operation type, and whether to add a circulating nurse.

[0104] Nursing result record: At the end of the nursing cycle, record whether a nursing safety event occurred during the operation and the specific situation of the event (such as event type, occurrence time, etc.).

[0105] Generate user records User records: Integrate the collected nursing information and the corresponding nursing results into user records. For example: Surgery A, operation time 5 hours, instrument inventory 3 times, operation type is orthopedic surgery, add a circulating nurse, no nursing safety event occurred during the nursing cycle.

[0106] Surgery B, operation time 4 hours, instrument inventory 1 time, operation type is general surgery, no circulating nurse added, a nursing safety event occurred during the nursing cycle, event type is instrument omission.

[0107] Save historical nursing data Historical nursing data: Save the same nursing information and / or nursing results as historical nursing data. For example, save all surgical records with 3 instrument inventories and a circulating nurse added as a group of historical nursing data, and save all surgical records with 1 instrument inventory and no circulating nurse added as another group of historical nursing data.

[0108] Data saving format: { "Nursing information": { "Operation time": 5, "Number of instrument inventories": 3, "Operation type": "Orthopedic surgery", "Add circulating nurse": "Yes" }, "Nursing result": { "Whether a nursing safety event occurred": "No", "Event type": null, "Occurrence Time": null } } Specifically, the user information at least includes the doctor's order sheet and the nursing record sheet corresponding to the user when completing a single cycle; Among them, for a single nursing category, the corresponding several doctor's order sheets and the nursing record sheets corresponding to each doctor's order sheet should be the same respectively in semantics.

[0109] Specifically, the historical nursing data at least includes the nursing electronic medical record, including the temperature sheet, the doctor's order sheet, and the nursing record sheet.

[0110] Specifically, the steps for verifying user data include: Select several users and several incomplete cycles as the verification batch; Use the association model and the understanding model to generate the predicted user information corresponding to the verification batch; Compare the predicted user information with the corresponding actual user information to generate the accuracy rate of this verification batch.

[0111] Specifically, when the accuracy rates of both the association model and the understanding model reach the preset accuracy rate, it is determined that the association model and the understanding model have completed iteration.

[0112] Specifically, the nursing effectiveness analysis method based on big data also has a sampling detection strategy, including: During the operation of the association model and the understanding model, conduct sampling detection at a preset ratio; When the accuracy rate of the sampling detection is not greater than the preset accuracy rate, it is determined that the association model and the understanding model fail; When the association model and the understanding model fail, train based on the failed association model and understanding model until the preset accuracy rate is reached.

[0113] Specifically, the preset accuracy rate is not less than 80%, and the preset ratio is not less than 30%.

[0114] This application includes six steps: business understanding, data understanding, data preparation, model establishment, model evaluation, and result deployment as Figure 4 shown, which is the connection schematic diagram of data mining in the embodiment of the present invention. Among them, The business understanding phase mainly analyzes business problems to determine the goals of data mining; the data understanding phase mainly collects original business data according to business goals and fully understands the data in combination with business knowledge; the data preparation phase mainly screens, cleans, and standardizes the original data to make the data meet the requirements of mining methods; the model building phase is the core phase of the entire data mining. In this phase, the model type, implementation method, and model construction need to be determined; the model evaluation phase mainly conducts business understanding and evaluation on the results obtained from the data mining model. If the mining results do not meet the requirements, it is necessary to return to the business understanding phase and conduct business analysis again; finally, it is the result deployment phase, where the mined results are applied to actual management work to realize the value of data mining.

[0115] In this study, the data sources for data mining are various nursing electronic medical records, including temperature sheets, doctor's order sheets, nursing record sheets, etc. The data mining program is designed according to the steps in the CRISP-DM model, and the previously formulated safety indicators are embedded in it. The following three functions in data mining are planned to be implemented and developed in this project: concept description, association knowledge mining, and predictive knowledge mining. The specific descriptions and implementation examples of each function are as follows.

[0116] Concept description: It is a summary of the data in the database to achieve an overall grasp of the data. Concept description is usually realized by using methods of mathematical statistics, such as calculating the sum, mean, maximum value, minimum value, variance, etc. of each data item, or by using online analytical processing to achieve multi-dimensional query and operation of data. For example, the daily regular statistics of the use of antibiotics in the department (administration time, administration speed, etc.) are used to judge whether the application of antibiotics is standardized.

[0117] Association knowledge mining: It reflects the dependence or interaction relationship between an object and other objects in the database. Usually, the associations among a large amount of data in the database are implicit and cannot be directly shown. Through association rule mining, the temporal relationship, quantitative relationship, or causal relationship, etc. existing between different objects can be found. Such association rules usually have great practical value and can judge the future development trend of associated objects based on the state of known objects. The algorithm used is the Apriori algorithm. For example, when the postoperative nursing record of a patient undergoing hepatobiliary surgery is entered as "the patient is restless due to the influence of sedative drugs", through association, the system automatically evaluates the risk factors related to the patient's unplanned extubation and fall / fall from bed in the data and prompts the nurse to pay key attention to and prevent the occurrence of these two types of nursing safety (adverse) events for this patient.

[0118] Predictive knowledge mining: It constructs a predictive model based on past and current data values in the database. Through the predictive model, the future development trend of an object can be determined. The method used is the regression analysis method in mathematical statistics and its various variants such as the log-linear model. For example, for the indicators proposed clinically, through historical data analysis, the situation in the past 3 years and the incidence rate of medical safety events are examined, and the mean value is used as the baseline. Combining clinical experience, management requirements, and health economics references, the early warning trigger conditions are determined. Another example is to embed some existing or newly developed risk early warning models in the system, such as the improved early warning score system and the delirium prediction model. The system automatically captures the data corresponding to the model for the patient and warns of the possibility of the patient developing a certain disease / complication according to the established rules, prompting medical staff to intervene early.

[0119] Taking safe blood transfusion as an example, existing clinical information systems and nursing information systems can record each time point of the entire blood transfusion process. There are also clear regulations for each link of blood transfusion in the relevant infusion operation industry guidelines. The original blood transfusion safety management is evaluated by asking nurses. The "safety" module can automatically compare the compliance rate between the two. If there is a gap, first, a warning prompt is given to the nurse, and second, the quality and safety manager is notified to come to the scene for supervision, reflecting timeliness and scientificity.

[0120] So far, the technical solution of the present invention has been described in combination with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the protection scope of the present invention.

[0121] The above are only the preferred embodiments of the present invention and are not used to limit the present invention; for those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for analyzing the effectiveness of nursing based on big data, characterized in that, It includes: Collect the historical nursing data of each user category, the usage data of users in each nursing category, and the user information of each user, and form a database; Establish an association model, and learn from the historical nursing data and user usage data according to the user information; Establish an understanding model, set the user portraits corresponding to each user, and establish a sub-database for the corresponding user; Verify the user data, and generate a number of user nursing results corresponding to the user data; Determine the accuracy rate of the corresponding understanding model according to the user nursing results; According to the accuracy rate of the understanding model, iterate the understanding model and the association model according to the accuracy rate; When the preset accuracy rate is reached, stop the iteration, and output the association model and the understanding model.

2. The method for analyzing the effectiveness of nursing based on big data according to claim 1, characterized in that For a single user, the steps of establishing the association model include: Determine the user category corresponding to the user; Query the nursing category corresponding to the user category according to the user category; Fuzzify the nursing category, and generate the corresponding fuzzy nursing category; Among them, the fuzzy nursing category includes the nursing categories with a frequency of occurrence not less than the preset frequency among the several nursing categories used by several users with the same user category as the user.

3. The method for analyzing the effectiveness of nursing based on big data according to claim 1, wherein For a single user, the steps of establishing the understanding model include: Determine the historical nursing data and user category of the user; Determine the user information of the corresponding user under different nursing categories according to the historical nursing data; Comprehensively screen the nursing categories according to the user information to form a sub-database corresponding to the user.

4. The method for analyzing the effectiveness of nursing based on big data according to claim 2 or 3, characterized in that When collecting the historical nursing data of each user category, it includes: Collect several nursing information of the user and the corresponding nursing results at regular intervals; Generate the corresponding user records according to the nursing information and nursing results; Save the same nursing information and / or nursing results as historical nursing data; Among them, the cycle is the cycle from a single nursing information to the corresponding nursing result generated by the nursing, and this cycle is only recorded when the nursing result appears.

5. The method for analyzing the effectiveness of nursing based on big data according to claim 4, characterized in that, The user information at least includes the doctor's order form and the nursing record form corresponding to the user when completing a single cycle; Among them, for a single nursing category, the corresponding several doctor's order forms and the nursing record forms corresponding to each doctor's order form should be the same in semantics.

6. The method for analyzing the effectiveness of nursing based on big data according to claim 5, wherein The historical nursing data at least includes nursing electronic medical records, including temperature sheets, doctor's order forms, and nursing record forms.

7. The method for analyzing the effectiveness of care based on big data according to claim 6, wherein, The steps of verifying the user data include: Select several users and several incomplete cycles as the verification batches; Use the association model and the understanding model to generate the predicted user information corresponding to the verification batches; Compare the predicted user information with the corresponding actual user information to generate the accuracy rate of this verification batch.

8. The method for analyzing the effectiveness of nursing based on big data according to claim 7, characterized in that When the accuracy rates of the association model and the understanding model both reach the preset accuracy rate, it is determined that the association model and the understanding model have completed the iteration.

9. The method for analyzing the effectiveness of nursing based on big data according to claim 8, wherein There is also a sampling detection strategy, including: During the operation of the association model and the understanding model, perform sampling detection at a preset ratio; When the accuracy rate of the sampling detection is not greater than the preset accuracy rate, it is determined that the association model and the understanding model fail; When the association model and the understanding model fail, train based on the failed association model and understanding model until the preset accuracy rate is reached.

10. The method for analyzing the effectiveness of nursing based on big data according to claim 9, wherein The preset accuracy rate is not less than 80%, and the preset ratio is not less than 30%.

Images