Cerebrovascular center nursing resource dynamic matching and safety early warning management method and system

Abstract

The application discloses a kind of cerebral vascular center nursing resource dynamic matching and safety early warning management method and system, including monitoring center and with its communication connection data acquisition, cerebral vascular disease special multi-dimensional evaluation, nursing resource full-dimensional image, resource dynamic intelligent matching, nursing real-time monitoring and early warning module.The application exports patient dynamic nursing demand image by special multi-dimensional evaluation system, completes nursing staff quantitative evaluation from physiological state and cognitive load dimension and exports real-time working state image, designs double-layer resource matching model of static scheduling and dynamic optimization, is combined with real-time monitoring and forward-looking prediction double-track trigger mechanism, realizes the linkage of risk early warning and resource scheduling depth.This application solves the pain points of the prior art scheduling static, early warning lag, resource supply-demand mismatch, can significantly improve the safety level and resource utilization efficiency of cerebral vascular center nursing.

Description

Technical Field

[0001] This invention relates to the field of clinical nursing management technology, specifically to a method and system for dynamic matching and safety early warning management of nursing resources in cerebrovascular centers. Background Technology

[0002] As the core unit for the treatment of stroke patients, the cerebrovascular center serves patients who generally have rapid changes in their condition, multiple complications, high risks of specialized nursing care, and high nursing intensity. This places extremely high demands on the precise allocation of nursing resources and the forward-looking management of nursing safety.

[0003] Currently, the nursing management model of cerebrovascular centers has the following significant shortcomings: 1. Static allocation of nursing resources and severe mismatch between supply and demand. Current nursing scheduling largely adopts a static scheduling model based on the number of beds, which cannot be dynamically adjusted according to the real-time severity of patients' conditions and changes in nursing needs. Especially in emergency scenarios such as stroke emergency channels and sudden changes in patient conditions, it is difficult to quickly match nursing staff with the corresponding specialist capabilities, easily leading to problems such as insufficient nursing resources, mismatched skill levels, and delayed emergency response. There are also instances of idle and wasted nursing resources in some scenarios.

[0004] 2. Existing nursing safety management technologies are mostly based on generalized risk assessment and threshold alarms, which can only trigger treatment after changes in the patient's condition or the occurrence of adverse events, resulting in serious early warning lag. At the same time, the generalized models do not fit the characteristics of cerebrovascular disease, such as rapid disease progression, highly specific complications, and high time window dependence. The accuracy of risk identification is insufficient, and it is easy to miss or false alarms, which cannot meet the safety prevention and control needs of patients in the acute phase of stroke.

[0005] 3. Risk warning and resource allocation are disconnected, failing to form a closed-loop management system. In existing technologies, nursing risk warning systems and nursing scheduling management systems operate independently. After a warning is triggered, it is impossible to automatically match nursing staff with corresponding capabilities for emergency treatment, relying on manual scheduling, which is inefficient and prone to delays in response. Furthermore, the effects of interventions cannot be automatically fed back to the model, hindering continuous model optimization and closed-loop improvement of nursing quality. Summary of the Invention

[0006] To address the aforementioned technical problems, the present invention aims to provide a dynamic matching and safety early warning management system for cerebrovascular center nursing resources, including a monitoring center. The monitoring center is communicatively connected to a data acquisition module, a cerebrovascular disease-specific multidimensional assessment module, a nursing resource full-dimensional profiling module, a resource dynamic intelligent matching module, and a nursing real-time monitoring and early warning module. The data acquisition module is used to collect comprehensive diagnosis, treatment, and nursing data of patients with cerebrovascular diseases and comprehensive management data of nursing staff, and to mark the collection time and set the collection cycle; The cerebrovascular disease-specific multidimensional assessment module is used to construct a cerebrovascular disease-specific multidimensional assessment system based on the full-dimensional diagnosis, treatment and nursing data of cerebrovascular disease patients. It can quantitatively assess the severity of the patient's condition, the risk of specialized nursing care and the workload of nursing care in real time, and output a dynamic nursing needs profile of the patient. The nursing resource full-dimensional profile module is used to conduct real-time quantitative assessment of cognitive load and physiological status based on the full-dimensional management data of nursing staff, and output a real-time work status profile of nursing staff, which includes a comprehensive work status level and hard constraint rules for resource matching. The resource dynamic intelligent matching module is used to perform static basic scheduling matching and trigger-based dynamic real-time optimization matching of nursing resources based on the dynamic nursing needs profile of patients and the dynamic capability profile of nursing staff, and output the optimal resource matching scheme. The real-time nursing monitoring and early warning module is used to monitor the dynamic nursing needs profiles of all patients and the real-time work status profiles of all nursing staff in real time. Based on the real-time monitoring results, it determines whether to generate a response level. At the same time, based on the comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients, it constructs a cerebrovascular specialty data prediction model and outputs multi-dimensional feature prediction indicators. Based on the multi-dimensional feature prediction indicators, it determines whether to generate an early warning response level. If a response level or an early warning response level is generated, it triggers the dynamic real-time optimization matching of the resource dynamic intelligent matching module.

[0007] Furthermore, the process of real-time quantitative assessment of the severity of a patient's condition includes: Specialty features were extracted and standardized preprocessing was performed on the collected comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients to generate multi-dimensional feature indicators; A severity assessment system for patients' conditions was constructed, comprising primary sub-dimensions, weighting percentages of these sub-dimensions, secondary core assessment indicators, and specialized quantitative rules. Secondary core assessment indicators were extracted from the multi-dimensional feature indicators based on this system and standardized scoring was applied. A weighted average was then calculated for each secondary core assessment indicator based on its weighting percentage to obtain a comprehensive score for the patient's severity condition. Threshold points were selected within the comprehensive score range to divide the patient's condition into sub-intervals representing different severity levels. The severity level corresponding to the patient's comprehensive score was determined based on the sub-interval to which it belonged.

[0008] Furthermore, the process of real-time quantitative assessment of patients' specialist nursing risks and workload includes: A specialized nursing risk dimension assessment system was constructed, which includes specialized quantitative assessment indicators, the weighting of specialized quantitative assessment indicators, and evidence-based scoring rules. Based on the specialized nursing risk dimension assessment system, specialized quantitative assessment indicators corresponding to the multi-dimensional feature indicators were extracted, and standardized scores were assigned to the specialized quantitative assessment indicators based on evidence-based scoring rules. Subsequently, according to the weighting of each specialized quantitative assessment indicator, the scores of each specialized quantitative assessment indicator were weighted and averaged to obtain a comprehensive risk score. Within the comprehensive risk score range, threshold points were selected to divide the range into sub-intervals with different risk levels. Based on the sub-interval to which the comprehensive risk score belongs, the risk level corresponding to the comprehensive risk score was determined. A three-dimensional stroke-specific working hour quantification model was constructed. Multi-dimensional feature indicators, severity level of illness, and risk level were input into the three-dimensional stroke-specific working hour quantification model. Based on the three-dimensional stroke-specific working hour quantification model, the model outputs basic direct nursing hours, indirect nursing hours, and risk-additional hours, as well as the corresponding weight ratios of the three types of hours. Based on the basic direct nursing hours, indirect nursing hours, and risk-additional hours, as well as the corresponding weight ratios of the three types of hours, the total nursing hours were obtained. The nursing load level corresponding to different nursing hour intervals was preset, and the nursing load level corresponding to the working hour interval to which the total nursing hours belong was obtained. A dynamic nursing needs profile of patients is constructed based on their severity of illness, risk level, and nursing workload level.

[0009] Furthermore, the process of quantitatively assessing the physiological state dimension based on the comprehensive management data of nursing staff, as described in the nursing resource profile module, includes: Core physiological indicators, working hours, and shift data are extracted from the comprehensive management data of nursing staff to construct a quantitative assessment system for physiological fatigue. This system includes secondary sub-indicators from the core physiological indicators and working hours / shift data, their corresponding weights, and specialized quantitative scoring rules. Based on these rules, standardized scores are assigned to the secondary sub-indicators. A weighted average is then calculated based on the score and weight of each secondary sub-indicator to obtain a comprehensive physiological fatigue score. Furthermore, different fatigue levels are preset for different comprehensive physiological fatigue score ranges, allowing for the acquisition of the corresponding fatigue level within each range.

[0010] Furthermore, the nursing resource holistic profiling module performs real-time cognitive load quantification assessment based on holistic management data of nursing staff, and the process of outputting a real-time work status profile of nursing staff includes: A cognitive load quantification model for patients is constructed. The dynamic nursing needs profiles of all patients under the care of each nurse are input into the cognitive load quantification model, and the basic patient load score is output based on the cognitive load quantification model. A real-time task cognitive load assessment model is constructed. Real-time operation data is extracted from the full-dimensional management data corresponding to each nurse. The real-time operation data is input into the real-time task cognitive load assessment model. The real-time task cognitive load assessment model performs standardized scoring on sub-dimensions based on the built-in specialized quantitative scoring rules. Then, a weighted average is performed based on the scores of each sub-dimension and the built-in sub-weights to output the real-time task cognitive load score. The cognitive load score of the real-time task and the basic patient load score are weighted and averaged to obtain the comprehensive cognitive load score. Different cognitive load scores are preset to correspond to different cognitive load levels, and the cognitive load level corresponding to the interval to which the comprehensive cognitive load score belongs is obtained. Construct a two-dimensional fusion matrix, which includes the comprehensive work status level corresponding to different fatigue levels and cognitive load levels, as well as the resource matching constraint parameters corresponding to different comprehensive work status levels. Based on the fatigue level and cognitive load level of each nurse and the two-dimensional fusion matrix, the comprehensive work status level and resource matching hard constraint rules of each nurse are obtained. Based on the comprehensive work status level and resource matching hard constraint rules of each nurse, a real-time work status profile of the nurse is output.

[0011] Furthermore, the resource dynamic intelligent matching module, based on the dynamic nursing needs profile of patients and the dynamic capability profile of nursing staff, performs static basic scheduling matching of nursing resources and outputs the optimal resource matching solution. The process includes: Construct a veto hard constraint rule, import the dynamic nursing needs profile of all patients and the real-time work status profile of all nursing staff, and based on the veto hard constraint rule, eliminate invalid matching pairs between patients and nursing staff in advance and generate feasible matching pairs. Several nurse-patient allocation and scheduling schemes are randomly generated based on feasible matching pairs. Three types of objective functions are constructed. A static optimization function is constructed based on the three types of objective functions. For each group of nurse-patient allocation and scheduling schemes, the static optimization function is substituted to obtain the comprehensive fit. The top 10% of nurse-patient allocation and scheduling schemes with the highest overall suitability were selected as the candidate optimal solution set. The candidate optimal solution set was then subjected to a second verification based on clinical practice habits. Candidate optimal solutions that did not conform to clinical practice habits were eliminated. From the candidate optimal solutions that passed the second verification based on clinical practice habits, the candidate optimal solution with the highest overall suitability was selected as the optimal resource matching scheme.

[0012] Furthermore, the process of trigger-based dynamic real-time optimization matching and outputting the optimal resource matching solution includes: Add real-time constraint rules to the veto hard constraint rules to generate dynamic matching hard constraint rules. Based on real-time monitoring results, lock the local optimization domain in the current optimal resource matching scheme. Then, based on the dynamic matching hard constraint rules, remove invalid matching pairs in the local optimization domain to generate feasible matching pairs. Based on the feasible matching pairs, randomly generate several resource matching schemes. Perform chromosome encoding and initialize the population for several resource matching schemes to generate an initial feasible solution population. Based on the three types of objective functions, a perturbation objective function is added to form a four types of objective functions. The weight coefficients of the four types of objective functions are set according to the response level. A dynamic optimization function is constructed based on the four types of objective functions and the weight coefficients. The initial feasible solution population is substituted into the dynamic optimization function to obtain the fitness value. Then, through adaptive selection, crossover, mutation operations, iterative optimization and secondary verification based on clinical practice habits, the optimal resource matching scheme is output. The original optimal resource matching scheme is updated based on the output optimal resource matching scheme.

[0013] Furthermore, the process of real-time monitoring of the dynamic nursing needs profiles of all patients and the real-time work status profiles of all nursing staff, and determining whether to generate a response level based on the real-time monitoring results, includes: A Level I emergency response is triggered when the patient's dynamic nursing needs profile shows an upgrade in the severity / risk level of the patient's condition, or when the comprehensive work status level of the nursing staff in the real-time work status profile drops to Level I (disabled) / Level II (high-risk), or when a new dynamic nursing needs profile of the patient is generated. When the severity / risk level of the patient's condition is downgraded, the overall status level is upgraded / downgraded, or the patient's dynamic nursing needs profile is marked as discharged / transferred, a Level II routine adjustment response is triggered. When a Level I emergency response or a Level II routine adjustment response is triggered, the dynamic real-time optimization matching of the resource dynamic intelligent matching module is activated.

[0014] Furthermore, based on comprehensive diagnosis, treatment, and nursing data of cerebrovascular disease patients, a cerebrovascular specialty data prediction model is constructed, outputting multi-dimensional feature prediction indicators. The process of determining whether to generate an early warning response level based on these multi-dimensional feature prediction indicators includes: Multidimensional feature indicators are extracted from the comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients in the patient's recent historical collection period and the current collection period. The multidimensional feature indicators are input into the cerebrovascular specialty data prediction model. Based on the cerebrovascular specialty data prediction model, the multidimensional feature prediction indicators for the next period are output. The multidimensional feature prediction indicators are fed back to the cerebrovascular disease-specific multidimensional assessment module. Based on the cerebrovascular disease-specific multidimensional assessment module, the dynamic nursing needs prediction profile of the patient is output. When the dynamic nursing needs prediction profile of a patient shows an escalation in the severity of their condition / risk level, a Level I emergency warning level is generated for the patient. When a patient’s dynamic nursing needs profile shows a downgrade in the severity of their condition / risk level, a Level II routine adjustment warning level is generated for the patient. When a patient's Level I emergency warning level or Level II routine adjustment warning level is generated, the dynamic real-time optimization matching of the resource dynamic intelligent matching module is activated in advance.

[0015] The method for dynamic matching and safety early warning management of nursing resources in cerebrovascular centers includes the following steps: Step s1: Collect comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients and comprehensive management data of nursing staff, mark the collection time, and set the collection cycle; Step s2: Based on the comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients, construct a multi-dimensional assessment system specifically for cerebrovascular disease, and conduct real-time quantitative assessment of the severity of the patient's condition, the risk of specialized nursing care, and the workload of nursing care, and output a dynamic nursing needs profile of the patient; Step s3: Based on the full-dimensional management data of nursing staff, conduct real-time quantitative assessment of cognitive load and physiological state dimensions, and output a real-time work status profile of nursing staff. The real-time work status profile of nursing staff includes comprehensive work status level and resource matching hard constraint rules. Step s4: Based on the dynamic nursing needs profile of patients and the dynamic capability profile of nursing staff, perform static basic scheduling matching and trigger-based dynamic real-time optimization matching of nursing resources, and output the optimal resource matching scheme; Step s5: Monitor the dynamic nursing needs profiles of all patients and the real-time work status profiles of all nursing staff in real time. Determine whether to generate a response level based on the real-time monitoring results. At the same time, construct a cerebrovascular specialty data prediction model based on the comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients, output multi-dimensional feature prediction indicators, and determine whether to generate an early warning response level based on the multi-dimensional feature prediction indicators. If a response level or early warning response level is generated, trigger dynamic real-time optimization matching.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention constructs a full-chain safety prevention and control system specifically for the characteristics of cerebrovascular diseases. On the one hand, through a dedicated multi-dimensional assessment system, it accurately quantifies the severity of the patient's condition, the risks of specialized nursing care, and the nursing needs, fully aligning with the pathophysiological characteristics and clinical nursing standards of stroke patients, thereby improving the accuracy of risk identification and the suitability of specialized care from the source. On the other hand, through a specialized data prediction model, it enables forward prediction of the patient's future condition characteristics, combined with real-time dynamic monitoring to form a dual-track triggering mechanism. This covers both immediate response to changes in the patient's condition and proactive identification and intervention of potential nursing risks, completely breaking away from the passive post-event handling mode of existing technologies. It constructs a closed-loop safety prevention and control mechanism of "risk prediction - graded early warning - precise intervention - effect feedback," fundamentally reducing the risk of adverse nursing events and providing systematic protection for the life safety and rehabilitation of stroke patients throughout their entire life cycle.

[0017] 2. This invention constructs a two-layer resource matching model combining static basic scheduling and trigger-based dynamic optimization. Static basic scheduling is based on patients' baseline nursing needs, ensuring the continuity, stability, and team collaboration of clinical nursing work. Trigger-based dynamic optimization is driven by real-time changes in patients' nursing needs and dynamic updates to nurses' work status, achieving real-time and accurate matching between patients' nursing needs and nurses' capabilities. Simultaneously, this invention, for the first time, incorporates nurses' real-time cognitive load and physiological fatigue status into the core hard constraint system of resource matching. During the matching process, it not only focuses on patients' nursing needs but also considers nurses' real-time work status and workload capacity. This not only eliminates the safety risks of unqualified or poorly performing nurses undertaking high-risk nursing tasks but also achieves a balanced distribution of workload among nurses on the same shift. In emergency scenarios, it enables rapid and accurate scheduling of nursing resources, completely solving the industry pain points of mismatched supply and demand and delayed response in the traditional model, maximizing the utilization efficiency and rational allocation of nursing resources.

[0018] 3. This invention deeply integrates a tiered response mechanism with a dynamic intelligent resource matching module. Whether it's a response level triggered by real-time monitoring of changes in patient condition or a response level triggered by forward-looking prediction of potential risks, it can directly activate the dynamic optimization and matching of nursing resources. This achieves a seamless connection between "risk warning triggering - matching of corresponding capable nursing personnel - implementation of standardized intervention plans," enabling pre-emptive scheduling and precise matching of emergency nursing resources without manual intervention. It completely breaks down the link between risk warning and clinical treatment, ensuring that when high-risk events occur, nursing personnel with corresponding treatment capabilities can respond immediately. This significantly improves the response speed and standardization of emergency treatment, and guarantees the closed-loop treatment of nursing risks throughout the entire process from a mechanism perspective. Attached Figure Description

[0019] Figure 1 This is a flowchart of the dynamic matching and safety early warning management system for cerebrovascular center nursing resources, which is an embodiment of this application.

[0020] Figure 2 This is a flowchart illustrating the dynamic matching and safety early warning management method for cerebrovascular center nursing resources, as described in this application. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] like Figure 1 As shown, the dynamic matching and safety early warning management system for cerebrovascular center nursing resources includes a monitoring center. The monitoring center is connected to a data acquisition module, a cerebrovascular disease-specific multidimensional assessment module, a nursing resource full-dimensional profiling module, a resource dynamic intelligent matching module, and a nursing real-time monitoring and early warning module. The data acquisition module is used to collect comprehensive diagnosis, treatment, and nursing data of patients with cerebrovascular diseases and comprehensive management data of nursing staff, and to mark the collection time and set the collection cycle; The cerebrovascular disease-specific multidimensional assessment module is used to construct a cerebrovascular disease-specific multidimensional assessment system based on the full-dimensional diagnosis, treatment and nursing data of cerebrovascular disease patients. It can quantitatively assess the severity of the patient's condition, the risk of specialized nursing care and the workload of nursing care in real time, and output a dynamic nursing needs profile of the patient. The nursing resource full-dimensional profile module is used to conduct real-time quantitative assessment of cognitive load and physiological status based on the full-dimensional management data of nursing staff, and output a real-time work status profile of nursing staff, which includes a comprehensive work status level and hard constraint rules for resource matching. The resource dynamic intelligent matching module is used to perform static basic scheduling matching and trigger-based dynamic real-time optimization matching of nursing resources based on the dynamic nursing needs profile of patients and the dynamic capability profile of nursing staff, and output the optimal resource matching scheme. The real-time nursing monitoring and early warning module is used to monitor the dynamic nursing needs profiles of all patients and the real-time work status profiles of all nursing staff in real time. Based on the real-time monitoring results, it determines whether to generate a response level. At the same time, based on the comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients, it constructs a cerebrovascular specialty data prediction model and outputs multi-dimensional feature prediction indicators. Based on the multi-dimensional feature prediction indicators, it determines whether to generate an early warning response level. If a response level or an early warning response level is generated, it triggers the dynamic real-time optimization matching of the resource dynamic intelligent matching module.

[0023] It should be further explained that, in the specific implementation process, the data acquisition module's acquisition process includes: A nursing blockchain network is constructed using blockchain technology, assigning a unique blockchain node to each nursing terminal within a preset range. Each nursing terminal communicates with the assigned unique blockchain node through a communication terminal, and each blockchain node is interconnected with the others. The nursing terminal's communication terminal preprocesses and encrypts the collected comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients and comprehensive management data of nursing staff. This encryption process involves pre-setting a public key (stored on the blockchain node) and a private key (stored on the communication terminal) for each communication terminal using an asymmetric encryption algorithm. When the nursing terminal uploads data to the blockchain node via the communication terminal, it uses the private key to digitally sign the data. The blockchain node verifies the validity of the digital signature based on the public key. Once the validity verification is successful, the PBFT (Practical Byzantine Fault Tolerance) consensus mechanism is adopted, with all blockchain nodes within the preset range participating in the consensus to ensure the consistency and immutability of on-chain data across multiple nodes. The data is then uploaded to the assigned unique blockchain node.

[0024] Among them, nursing terminals include core clinical nursing operation terminals (mobile nursing PDA / handheld smart terminals), nursing management and decision-making terminals (ward nursing management workstations (PC / tablet terminals)), and multimodal diagnosis and treatment and Internet of Things sensing terminals (bedside vital signs monitoring terminals, cerebrovascular disease monitoring equipment terminals). Comprehensive diagnostic and nursing data for cerebrovascular disease patients includes: Patient baseline identity and treatment data (including precise classification of cerebrovascular diseases: ischemic stroke / hemorrhagic stroke / subarachnoid hemorrhage / transient ischemic attack, etc.); time-series data of the entire process of cerebrovascular disease diagnosis and treatment (pre-hospital onset time, call for help time, pre-hospital emergency arrival time, arrival time, completion time of head CT / MRI, thrombolysis DNT time, thrombectomy puncture time, vascular recanalization time, postoperative transfer to ward time, and other data on the golden time nodes of stroke treatment); core execution data of full-cycle specialized nursing (full record of basic nursing execution: vital sign measurement, oral care, airway care, etc.; cerebrovascular disease specialized nursing operation record: full-process nursing cooperation for thrombolysis, intracranial pressure monitoring special nursing, stroke rehabilitation nursing, etc.); real-time monitoring and equipment acquisition data (heart rate, systolic / diastolic blood pressure, intracranial pressure, cerebral perfusion pressure, cerebrospinal fluid drainage volume and characteristics, etc.).

[0025] Comprehensive management data for nursing staff includes: Basic identity and qualification certification data of nursing staff, special skills assessment data (assessment records, pass / fail status, and validity period of special skills such as thrombolysis nursing cooperation, intracranial pressure monitoring nursing, cardiopulmonary resuscitation / defibrillation and other emergency skills, and dysphagia screening), real-time on-duty status data (sign-in / sign-out time, on-duty / off-duty / standby / leave status, real-time physical location, ward / position, and shift type), real-time workload data (continuous working hours, number of patients currently under care, and corresponding patient criticality level), full-chain data of operational behavior (full process record of initiation, execution, and confirmation of all nursing operations), core nursing quality indicator data (nursing documentation qualification rate, medical order execution accuracy rate, nursing operation standardization rate, and adverse event incidence rate), and rewards, punishments, and event records (reporting and handling records of adverse nursing events, patient praise / complaint records, outstanding deeds, and hospital / hospital-level commendation records).

[0026] It should be further explained that, in the specific implementation process, the real-time quantitative assessment of the patient's condition severity and the risks of specialized nursing care includes: The cerebrovascular disease-specific multidimensional assessment module is designed around the pathophysiological characteristics, nursing risk features, and clinical nursing needs of cerebrovascular disease patients. It constructs a dedicated assessment system based on three core dimensions: severity of illness, specialist nursing risk, and nursing workload. Each dimension is based on the "Chinese Guidelines for Stroke Nursing (2025 Edition)," the AHA / ASA Stroke Nursing Guidelines, and JBI evidence-based nursing evidence. The weight allocation of each indicator is determined through the Analytic Hierarchy Process (AHP) combined with the Delphi method of more than 20 national stroke nursing and neurology clinical experts, ensuring the scientific nature and specialist suitability of the assessment. All assessment indicators, weights, and scoring rules are supported by clear industry guidelines and evidence-based medicine, and have been reviewed by the clinical expert committee, fully complying with the requirements of the "Medical Quality Management Measures" and the "Stroke Nursing Guidelines."

[0027] Specialty features were extracted and standardized preprocessing was performed on the collected comprehensive diagnosis, treatment, and nursing data of cerebrovascular disease patients to generate multi-dimensional feature indicators, including: Continuous characteristic indicators, such as vital signs, ICP, NIHSS scores, and test indicators, are converted into standardized scores in the range of 0-100 using the min-max standardization method to eliminate dimensional differences. Classification characteristic indicators, such as stroke classification, tubing type, and dysphagia grading, are converted into standardized scores for the corresponding grades based on evidence-based methods using a combination of unique coding and graded scoring. Time-series characteristic indicators, such as blood pressure fluctuations and trends in changes in consciousness, are extracted using a sliding window algorithm to extract core features such as rate of change, duration, and number of times thresholds are exceeded, and then converted into trend quantification scores. A severity assessment system for patients' conditions was constructed, comprising primary sub-dimensions, weighting percentages for each sub-dimension, secondary core assessment indicators, and specialized quantitative rules. Based on this system, secondary core assessment indicators were extracted from the multi-dimensional feature indicators and standardized to a score of 0-100. After standardizing the scores for each specialized quantitative assessment indicator, a weighted average was calculated based on the weighting percentage of each secondary core indicator to obtain a comprehensive severity score (out of 100, with higher scores indicating more severe conditions). This score was then used in conjunction with expert experience. Within the comprehensive score range of a patient's condition severity, threshold points are selected to divide the patient's condition into sub-intervals for different severity levels. Based on the sub-interval to which the patient's comprehensive score belongs, the severity level of the patient's condition is determined. For a severity level of I (extremely critical), the corresponding sub-interval is 85-100 points; for a severity level of II (critical), the corresponding sub-interval is 70-84 points; for a severity level of III (severe), the corresponding sub-interval is 55-69 points; for a severity level of IV (moderate), the corresponding sub-interval is 30-54 points; and for a severity level of V (recovery period), the corresponding sub-interval is 0-29 points. The standardized scoring of each secondary core assessment indicator from 0 to 100 points is based on guidelines such as the "Chinese Guidelines for the Diagnosis and Treatment of Acute Ischemic Stroke 2024" and the "Chinese Guidelines for Intravenous Thrombolysis in Acute Ischemic Stroke 2024". The scoring formula for core specialty indicators (accounting for over 90%) with clear clinical warning / critical values ​​is as follows: .

[0028] in, : Actual measured value of the indicator; The guidelines specify the minimum normal value for the indicators. The guidelines clearly define three core segmentation points: normal upper limit, warning threshold, and critical value. The slope of the assigned score for each interval follows the principle that the higher the risk, the steeper the slope. This ensures that the scores in the high-risk range increase more rapidly, aligning with clinical risk characteristics; The maximum score for the previous interval, such as the maximum score for a normal interval. Points, warning range capped. The high-risk zone is capped at 100 points.

[0029] For general continuous indicators where risk increases linearly and there are no clearly defined segmented warning thresholds, such as age and the number of underlying diseases, the scoring formula is as follows: ; in, These are measured values. The minimum normal value for the indicators specified in the guidelines. The maximum value specified in the guidelines.

[0030] The assessment system for the severity of illness is shown in Table 1 below: Table 1 Core indicators of neurological deficit 40% 1. NIHSS (National Institutes of Health Stroke Scale) score 2. Glasgow Coma Scale (GCS) score 3. Four-hour Coma Scale (FOUR) score (for intubated patients) 1. An increase of ≥4 points in the NIHSS score within 2 hours is weighted twice (indicating disease progression). Cerebrovascular disease specific indicators 30% 1. Intracranial pressure (ICP) / cerebral perfusion pressure (CPP) monitoring data; 2. Stroke lesion type and location; 3. Onset time window and reperfusion treatment status; 4. Hunt-Hess / Fisher classification (specifically for subarachnoid hemorrhage). 1. ICP > 25 mmHg with abnormal waveform, weight doubled; 2. Brainstem infarction, large-area cerebral infarction, massive cerebral hemorrhage, and ruptured aneurysm, given the highest score; 3. Hunt-Hess classification ≥ IV, directly included in the critically ill category. Overall physiological condition and critical value indicators 20% 1. APACHE II Acute Physiology Score 2. Stability of vital signs (blood pressure, heart rate, respiratory rhythm) 3. Critical laboratory values ​​(coagulation function, blood glucose, electrolytes, blood gas analysis) <![CDATA[1. After thrombolysis, fibrinogen < 1.5 g / L and platelet count < 100×10 9 / L directly trigger a grade upgrade (one level higher than the original critical illness grade); 2. Blood gas analysis indicating respiratory failure doubles the weight]]> Indicators of complications and comorbidities 10% 1. Precursor cerebral herniation, respiratory failure, heart failure, and other serious complications; 2. Multiple organ dysfunction; 3. Complications such as severe infection and gastrointestinal bleeding. 1. The presence of Cushing's reaction, unequal pupil size, or other precursors to brain herniation will result in immediate classification as extremely critical; 2. For each additional serious complication, the score will increase by 20%. Complete patient scoring calculation example: Patient basic information: 58-year-old male, acute brainstem infarction, 6 hours after intravenous thrombolysis, NIHSS score 19, GCS score 10, ICP 23 mmHg, systolic blood pressure fluctuation of 28 mmHg within 2 hours, fibrinogen 1.4 g / L, Kubota water swallowing test grade 4, RASS score +2.

[0031] 1. Scoring of each secondary indicator: NIHSS score: 88; GCS score: 60; ICP value: 54; Systolic blood pressure fluctuation: 100; Fibrinogen: 100; Kubota drinking water test: 80; RASS score: 60; 2. Weighted calculation of the severity of the illness: Overall score = 88 40%+54 30%+100 20%+80 10% = 35.2 + 16.2 + 20 + 8 = 79.4 points, corresponding to Level II critical condition.

[0032] It should be further explained that, in the specific implementation process, the real-time quantitative assessment of the patient's specialist nursing risks and workload includes: A specialized nursing risk dimension assessment system was constructed, which includes specialized quantitative assessment indicators, their weighting, and evidence-based scoring rules. Based on this system, specialized quantitative assessment indicators were extracted from the multi-dimensional feature indicators. Standardized scoring (0-100 points) was then applied to each indicator according to the evidence-based scoring rules (each indicator was initially assigned a raw score of 0-10, then normalized and mapped to the 0-100 range; for example, 6 points were mapped to 60 points). After standardizing the scoring for each indicator, the system was then used to assess the risk of complications arising from the specific circumstances. The weighted average of the scores for each specialty quantitative assessment indicator is used to obtain a comprehensive risk score (out of 100, with higher scores indicating higher risk). Based on expert experience, threshold points are selected within the comprehensive risk score range to divide it into sub-intervals for different risk levels. The risk level corresponding to the comprehensive risk score is determined according to the sub-interval to which it belongs. When the severity level of the illness is extremely high, the corresponding sub-interval is 85-100 points; when the severity level of the illness is high, the corresponding sub-interval is 70-84 points; when the severity level of the illness is medium, the corresponding sub-interval is 40-69 points; and when the severity level of the illness is low, the corresponding sub-interval is 0-39 points. The risk assessment system for specialized nursing is shown in Table 2 below: Table 2 Risk of early signs of brain herniation 25% Progressive decline in consciousness, pupillary changes, persistently elevated ICP, Cushing's reflex, altered respiratory rhythm, vomiting / restlessness 1. A GCS score decreasing by ≥2 points per hour, accompanied by unequal pupils and loss of light reflex, is awarded a perfect score of 10 points; 2. ICP consistently >20 mmHg, with an increase of 2 points for every 5 mmHg increase; 3. The presence of Cushing's reaction is awarded a perfect score of 10 points. Risk of post-thrombolysis / thrombectomy hemorrhage transformation 20% Time from onset to reperfusion, preoperative NIHSS score, blood pressure fluctuation range, postoperative coagulation function, platelet count, history of cerebral small vessel disease, and blood glucose level 1. Systolic blood pressure fluctuation >20 mmHg within 24 hours postoperatively: score increases by 3 points; 2. Fibrinogen <1.5 g / L: score increases by 5 points; 3. Preoperative NIHSS score ≥21: baseline score increases by 5 points. Risk of cerebral vasospasm 15% Hunt-Hess classification, Fisher classification, TCD blood flow velocity, changes in level of consciousness, worsening of neurological deficits, headache / meningeal irritation signs. 1. TCD indicates a blood flow velocity in the middle cerebral artery >200cm / s, awarding full marks directly; 2. Fisher classification ≥3, base score 4 points. Epilepsy seizure risk 10% Stroke type, lesion location (cortical involvement), onset time, history of epilepsy, altered consciousness, electrolyte imbalance 1. Cerebral hemorrhage / subarachnoid hemorrhage with cortical involvement: base score 3 points; 2. Electrolyte disturbance within 72 hours of onset: score increased by 3 points. Risk of stress ulcers / gastrointestinal bleeding 10% GCS score, duration of mechanical ventilation, use of anticoagulants / antiplatelet drugs, gastric juice / fecal occult blood test results, and history of ulcers. 1. Mechanical ventilation >48h, GCS ≤ 8 points, baseline score 4 points; 2. Positive occult blood test, direct score 8 points. Post-interventional procedure puncture site bleeding / hematoma risk 10% Puncture path, postoperative anticoagulation regimen, coagulation function, blood pressure control, and limb immobilization. 1. Femoral artery puncture and subsequent dual antiplatelet therapy: base score 3 points; 2. Systolic blood pressure >160 mmHg: score increased by 3 points. Post-stroke agitation / risk of unplanned extubation 10% RASS sedation score, altered consciousness, cognitive impairment, anxiety and depression score, number of tubing. 1. RASS score ≥ +2 points (agitation), directly awarded 8 points; 2. Confusion, 3 or more indwelling tubes, score increased by 3 points. A three-dimensional stroke-specific work hour quantification model was constructed. Multi-dimensional feature indicators, severity level of the patient's condition, and risk level were input into the model. Based on this model, the model outputs basic direct nursing hours, indirect nursing hours, and risk-related additional hours, along with their corresponding weights. The total nursing hours were then calculated based on these weights. ; in, , , These are the weighted percentages for basic direct nursing hours, indirect nursing hours, and risk-additional nursing hours, respectively. An additional coefficient is added to the risk level. Standard working hours for basic procedures as required by the doctor's orders. For operation frequency, Basic direct care hours, The indirect working hours corresponding to the severity of the illness are defined, and the nursing load levels corresponding to different nursing working hour intervals are preset. A single shift working hour interval > 180 minutes corresponds to an ultra-high nursing load level, a single shift working hour interval 120-180 minutes corresponds to a high nursing load level, a single shift working hour interval 60-120 minutes corresponds to a medium nursing load level, and a single shift working hour interval < 60 minutes corresponds to a low nursing load level. The nursing load level corresponding to the working hour interval to which the total nursing working hours belong is obtained. A dynamic nursing needs profile of patients is constructed based on their severity of illness, risk level, and nursing workload level.

[0033] The dynamic nursing needs profile of patients includes a basic information layer, a core level layer, a quantitative indicator layer, and a dynamic update trajectory layer; Basic information layer: Basic identification information such as patient basic information, stroke classification and diagnosis, treatment plan, admission time, attending medical staff, and bed number to achieve unique patient identification; Core grading level: Core grading results across three dimensions: severity of illness, risk level, and nursing workload level; Quantitative Indicator Layer: Comprehensive score of three dimensions, detailed score of each sub-item, trend of core indicator changes, and weighting percentage; Dynamically updated trajectory layer: Records the time, triggering reasons, score and grade changes, risk evolution trajectory, and demand change trends of all assessments since the patient's admission, forming a full-cycle time series curve to provide data support for disease prediction and model optimization.

[0034] The quantification rules for the three-dimensional stroke-specific work-hour quantification model are shown in Table 3 below: Table 3 Basic direct care hours 60% Based on the "National Nursing Service Price Item Standard" and the results of time measurement for cerebrovascular specialty nursing operations, a stroke-specific nursing operation library containing 126 operations was constructed. These operations are categorized into four main types, and cumulative working hours are precisely calculated according to doctor's orders and operation frequency: 1. Basic nursing operations: morning and evening care, oral care, turning and percussion, skin care, etc., with clearly defined standard working hours per operation; 2. Specialty nursing operations: stroke-specific operations such as swallowing function training, proper limb positioning, intracranial pressure monitoring, lumbar drainage, and post-thrombolysis care, with standard working hours per operation exceeding those of general operations; 3. Therapeutic operations: intravenous infusion, sputum suction, nasogastric feeding, nebulization, specimen collection, etc., with cumulative working hours calculated based on frequency; 4. Monitoring operations: vital sign monitoring, consciousness and pupil monitoring, blood glucose monitoring, etc., with precise calculation based on monitoring frequency, such as blood pressure monitoring every 15 minutes, with cumulative working hours per shift added up based on the number of monitoring sessions. Indirect nursing hours 25% Based on the severity of patients' conditions, differentiated indirect work hours per shift are set, with automatic matching by the system, eliminating the need for manual statistics: 1. Level I extremely critical patients: 40 minutes of indirect work hours per shift (including document writing, shift handover, ward rounds, doctor-patient communication, etc.); 2. Level II critical patients: 30 minutes of indirect work hours per shift; 3. Level III severely ill patients: 20 minutes of indirect work hours per shift; 4. Level IV-V patients: 10-15 minutes of indirect work hours per shift. Risk-related additional working hours 15% The core innovation of this module is to set additional working hours coefficients for stroke patients with high risk levels, accurately reflecting the actual clinical nursing time and solving the deviation of traditional working hour calculation that "only counts the operation and not the risk": Risk level additional coefficients: extremely high risk patients ×1.8, high risk ×1.5, medium risk ×1.2, low risk ×1.0; It should be further explained that, in the specific implementation process, the process of the nursing resource full-dimensional profiling module to conduct quantitative assessment of the physiological state dimension based on the full-dimensional management data of nursing staff includes: Current resource allocation technologies only consider nurses' static qualifications, workload, and historical performance, completely ignoring their real-time physiological fatigue and cognitive load. Cerebrovascular center nurses are chronically under high stress and high workload, and the continuous handling of critically ill patients and emergency situations leads to cognitive overload and physiological fatigue, which are core contributing factors to nursing errors, delayed responses, and adverse events. Current technologies lack relevant quantitative assessment mechanisms and corresponding error-mitigation measures, making it highly susceptible to the safety hazard of "qualified but poorly performing nurses handling high-risk patients."

[0035] This assessment adheres to five core principles throughout: non-invasiveness, clinical compliance, real-time dynamism, specialty suitability, and interpretability. All data collection does not interfere with normal clinical work or increase the burden on nursing staff. The assessment indicators, weights, and thresholds were all validated using the Analytic Hierarchy Process (AHP) combined with the Delphi method by 20 stroke nursing experts, nursing management experts, and human factors engineering experts nationwide. It is 100% suitable for the occupational characteristics of cerebrovascular center nurses, which include "high stress, high-risk operation intensity, frequent emergency events, and irreversible consequences of operational errors."

[0036] The core physiological indicators, working hours, and shift schedule data extracted from the comprehensive management data of nursing staff are shown in Table 4 below: Table 4 Core physiological indicators Medical-grade wearable smart bracelet 1. Heart Rate Variability (HRV): Time-domain metrics (SDNN, RMSSD, PNN50), frequency-domain metrics (LF / HF sympathetic / parasympathetic ratio) 2. Real-time heart rate, resting heart rate 3. Physical activity data (steps, standing time, resting time) 1Hz HRV is the internationally recognized gold standard for assessing autonomic nervous system function and fatigue. It is the core indicator for this dimension and can accurately identify early signs of physiological fatigue. Working hours and shift data Nursing scheduling system, medical staff positioning system, mobile nursing terminal (PDA) 1. Continuous working hours 2. Shift type (day shift / night shift / long night shift) 3. Number of consecutive night shifts 4. Total mileage walked during the shift 1 minute / time Quantify the cumulative fatigue caused by continuous work and adjust the impact of physiological rhythms on fatigue levels. A quantitative assessment system for physiological fatigue was constructed, comprising core physiological indicators and secondary sub-indicators from work hours and scheduling data, the weighting of each sub-indicator, and specialized quantitative scoring rules. Based on these rules, the core physiological indicators and secondary sub-indicators from work hours and scheduling data were standardized and scored from 0 to 100. A weighted average was then calculated based on the score and weighting of each sub-indicator to obtain a comprehensive physiological fatigue score (out of 100; a higher score indicates greater fatigue). (The more severe the fatigue); and preset the fatigue level corresponding to different comprehensive physiological fatigue score ranges. The comprehensive physiological fatigue score range of a single shift is 85-100 points, which corresponds to the fatigue level of severe fatigue; the comprehensive physiological fatigue score range of a single shift is 70-84 points, which corresponds to the fatigue level of moderate fatigue; the comprehensive physiological fatigue score range of a single shift is 40-69 points, which corresponds to the fatigue level of mild fatigue; and the comprehensive physiological fatigue score range of a single shift is 0-39 points, which corresponds to the fatigue level of no fatigue. Obtain the fatigue level corresponding to the comprehensive physiological fatigue score range. The quantitative assessment system for physiological fatigue is shown in Table 5 below: Table 5 Core physiological indicators SDNN (Overall Heart Rate Variability) 25% 30% below personal baseline → 80 points; 50% below baseline → full score of 100 points; within the baseline range → 0-20 points. RMSSD (parasympathetic nervous system activity) 20% 25% below personal baseline → 70 points; 40% below baseline → full score of 100 points LF / HF ratio (sympathetic / parasympathetic balance) 15% A score more than twice the baseline is assigned 70 points; a score more than three times the baseline is assigned a perfect score of 100 points (sympathetic over-excitation, indicating severe fatigue). Working hours and shift data Continuous on-the-job working hours 20% Working continuously for less than 4 hours → 0 points; for each hour exceeding 4 hours, the score increases by 10 points; working continuously for ≥10 hours → full score of 100 points. Physical activity expenditure 15% ≥15,000 steps during the shift → 60 points; ≥20,000 steps → 80 points The influence of shift rhythm 5% 20 points are awarded on the basis of the night shift; 10 points are awarded for each additional consecutive night shift; 4 consecutive night shifts → full score of 100 points. It should be further explained that, in the specific implementation process, the nursing resource full-dimensional profiling module performs real-time cognitive load quantitative assessment based on the full-dimensional management data of nursing staff, and outputs a real-time work status profile of nursing staff. The process includes: A cognitive load quantification model for patients is constructed. The dynamic nursing needs profiles of all patients under the care of each nurse are input into the cognitive load quantification model, and the basic patient load score is output based on the cognitive load quantification model. The calculation formula for the patient cognitive load quantification model is as follows: ; Where de represents the baseline patient workload score. This indicates the severity score of patient i's condition. This represents the risk score of patient i, and m represents the number of patients. This represents the correction factor for the number of patients; The quantification rules for the severity score, risk score, and patient number correction coefficient are as follows: Patient severity level weighting: Grade I extremely critical = 30 points / person, Grade II critical = 20 points / person, Grade III severe = 10 points / person, Grade IV-V = 5 points / person; Patient risk level weighting: Very high risk = 25 points / person, High risk = 15 points / person, Yellow intermediate risk = 5 points / person, Blue low risk = 2 points / person; Patient number correction factor: When there are less than 8 patients, the patient number correction factor = 1. When there are more than 8 patients, the patient number correction factor is multiplied by 1.1 for each additional patient.

[0037] A real-time task cognitive load assessment model is constructed. Real-time operation data is extracted from the full-dimensional management data corresponding to each nurse. The real-time operation data is input into the real-time task cognitive load assessment model. The real-time task cognitive load assessment model is standardized and scored from 0 to 100 points for each sub-dimension based on the built-in specialized quantitative scoring rules. Then, a weighted average is performed based on the score of each sub-dimension and the built-in sub-weights to output the real-time task cognitive load score. The real-time task cognitive load assessment model is calculated based on real-time operation data within a 15-minute acquisition period. It is divided into four sub-dimensions, each with a corresponding specialized quantitative scoring rule. The built-in rules of the real-time task cognitive load assessment model are shown in Table 6 below: Table 6 Task complexity and risk 35% Scoring based on the level of operational risk: • P0 Level Rigid High-Risk Operations (Thrombolytic Administration, Emergency Treatment of Brain Herniation, Intracranial Pressure Tube Care, Blood Transfusion, Critical Value Management): 50 points / operation • P1 Level Semi-Rigid Operations (Routine Administration, Suctioning, Specialized Nursing): 20 points / operation • P2 Level Flexible Operations (Health Education, Document Entry): 5 points / operation The more frequent the high-risk operations within a unit of time, the higher the score. Multitasking and Task Switching 25% • Handling ≥3 tasks simultaneously within 15 minutes: 30 points; ≥4 tasks: 50 points. • Switching between tasks ≥5 times within 15 minutes: 20 points; ≥10 times: 40 points. The more parallel tasks and the more frequent the switching, the higher the cognitive load. Operation response and error rate 20% • High-risk operation response timeout: 20 points / time • Operation error / correction / cancellation: 25 points / time • Document entry error: 10 points / time The slower the response and the more errors, the higher the score, indicating insufficient cognitive resources. Emergency response load 20% • Red Extremely High Risk Warning / Emergency Incident Response: 40 points / incident • Orange High Risk Warning Response: 20 points / incident • Emergency Green Channel Patient Reception: 30 points / incident Emergency events cause a sharp increase in cognitive load and have a high weighting. The real-time task cognitive load score (weighted by a coefficient of 0.6) and the baseline patient load score (weighted by a coefficient of 0.4) are weighted and averaged to obtain the comprehensive cognitive load score. Different cognitive load scores correspond to different cognitive load levels. A comprehensive cognitive load score range of 85-100 points corresponds to cognitive overload, a comprehensive cognitive load score range of 70-84 points corresponds to high cognitive load, a comprehensive cognitive load score range of 40-69 points corresponds to medium cognitive load, and a comprehensive cognitive load score range of 0-39 points corresponds to low cognitive load. The cognitive load level corresponding to the comprehensive cognitive load score range is obtained. A two-dimensional fusion matrix of physiological fatigue and cognitive load was constructed. The two-dimensional fusion matrix includes the comprehensive work status level corresponding to different fatigue levels and cognitive load levels, as well as the resource matching constraint parameters corresponding to different comprehensive work status levels. (The resource matching constraint parameters are fully aligned with the national nurse hierarchical management guidelines and the clinical rules for cerebrovascular center specialties. For example, when the comprehensive work status level is Level I, which is prohibited, the maximum number of new patients that can be added is 0 (new patients are prohibited, and all existing patients must be diverted), the highest critical level of patients that can be accepted is zero (no patients can be accepted), and the highest risk level of operations that can be performed is zero (no operations can be performed independently)). The comprehensive work status levels corresponding to different fatigue levels and cognitive load levels are shown in Table 7 below. Table 7 Level I Disabled Status Severe fatigue + cognitive overload; severe fatigue + high cognitive load Level II High-Risk Status Severe fatigue + low / medium cognitive load; moderate fatigue + cognitive overload; moderate fatigue + high cognitive load Level III warning status Moderate fatigue + moderate cognitive load; mild fatigue + cognitive overload; mild fatigue + high cognitive load Level IV Normal State Mild fatigue + moderate / low cognitive load; No fatigue + moderate cognitive load Level V optimal state No fatigue + low cognitive load Based on the fatigue level and cognitive load level of each nurse and the two-dimensional fusion matrix, the comprehensive work status level and resource matching hard constraint rules of each nurse are obtained. Based on the comprehensive work status level and resource matching hard constraint rules of each nurse, a real-time work status profile of the nurse is output.

[0038] The real-time work status profile of nursing staff includes the following content: 1. Core Results: Comprehensive work status level (Levels I-V), physiological fatigue score and level, and cognitive load score and level; 2. Resource matching constraint parameters: maximum number of patients that can be added, maximum critical illness level of patients that can be admitted, maximum risk level of operations that can be performed, and remaining threshold for available nursing hours; 3. Professional competency tags: These include five core dimensions: specialist nursing certification, cerebrovascular disease specialist nursing competency, emergency response competency, risk management competency, nurse-patient communication competency, and teaching competency. Each dimension is further subdivided into competency items, and each competency item uses a 1-5 level quantitative scoring system. The scoring is dynamically updated based on departmental assessment results, daily operation records, emergency response performance, and adverse event handling. 4. On-duty status tag: including on-duty and off-duty status, number of patients currently under care, and real-time physical location; 5. Score change curve over the past 4 hours.

[0039] It should be further explained that, in the specific implementation process, the resource dynamic intelligent matching module performs static basic scheduling matching of nursing resources based on the dynamic nursing needs profile of patients and the dynamic capability profile of nursing staff, and outputs the optimal resource matching solution. The process includes: Construct a veto hard constraint rule, import the dynamic nursing needs profile of all patients and the real-time work status profile of all nurses, and based on the veto hard constraint rule, eliminate invalid matching pairs between patients and nurses in advance (such as Level I critically ill patients, nurses whose comprehensive work status level is Level I prohibited cannot accept them, and directly mark them as infeasible), and generate feasible matching pairs (fully satisfying the veto hard constraint rule). Several nurse-patient allocation and scheduling schemes are randomly generated based on feasible matching pairs (more than 1000 nurse-patient allocation and scheduling schemes are generated, and the content of the nurse-patient allocation and scheduling schemes includes the following: shift overview: shift type, date, total number of nurses on duty, total number of patients under care; nurse-patient matching list: details of patients under care for each nurse, patient core level, and single shift work hours; matching degree details: needs-competency matching degree score of each nurse and the patients under their care, core matching basis; emergency standby arrangement: list of nurses in the emergency team on duty, and available access permissions). Three types of objective functions are constructed, and a static optimization function is constructed based on the three types of objective functions. For each nurse-patient allocation and scheduling scheme, the static optimization function is substituted to obtain the comprehensive fit degree. The top 10% of nurse-patient allocation and scheduling schemes with the highest overall suitability were selected as the candidate optimal solution set. The candidate optimal solution set was then subjected to a secondary verification based on clinical practice habits (e.g., at least one nurse in Level V status must be reserved for emergency backup in each shift). Candidate optimal solutions that did not conform to clinical practice habits were eliminated. From the candidate optimal solutions that passed the secondary verification based on clinical practice habits, the candidate optimal solution with the highest overall suitability was selected as the optimal resource matching scheme. Subsequently, a visual scheduling table was automatically generated, and an interface for head nurses to manually review and modify the schedule was opened. All modifications were fully recorded, and the system automatically re-verified the hard constraints after modification. After the head nurse reviewed and confirmed the schedule, it was officially released and synchronized to mobile nursing terminals and nurse station screens.

[0040] Among them, the hard constraint rule of veto includes: 1. Criticality level matching constraint: The criticality level of patient i is less than or equal to the highest criticality level that nurse j can handle; 2. Risk level matching constraint: The risk level of patient i ≤ the highest feasible operation risk level of nurse j; 3. Patient number constraint: The total number of patients assigned to nursing staff j is less than or equal to the maximum number of new patients that can be added; 4. Working hours capacity constraint: The sum of the total nursing hours allocated to all patients by nursing staff j is less than or equal to the remaining available nursing hours threshold; 5. Status and permission constraints: Level I nurses in the disabled state are forcibly excluded from the matching range; Level II nurses in the high-risk state are prohibited from assigning patients beyond their permissions.

[0041] The three types of objective functions include nurse competency matching degree. Nurses' workload balance Compatibility with nurses' condition The function formulas are as follows: ; Note: M = total number of patients, N = total number of nurses; 0-1 decision variables (( =1) represents patient i being assigned to caregiver j, otherwise 0); Match score (0-100 points), calculation rules: ,in, The highest level of critical illness that nurse j can handle. The severity level of patient i's condition. The highest possible risk level for the procedure performed by nurse j. The highest possible risk level for the procedure performed by nurse j. For patient i's risk level, For patient i, the standard nursing hours per shift The threshold for the remaining nursing hours that nurse j can undertake per shift.

[0042] ; in, The standard deviation of total nursing hours allocated to all nursing staff in the same shift. Distribute working hours equally among nursing staff in the same shift. The coefficient of variation is the smallest value, indicating a more balanced load. The closer to 1; ; in, The overall work status level of nursing staff j (Level V optimal status = 5 points, Level IV normal status = 4 points... Level I prohibited = 0 points); This indicates that the self-state score is normalized; This indicates the extent to which one has surplus workload; The individual fit score is represented as follows: good condition + light workload → high individual score, poor condition + heavy workload → extremely low individual score. The F3 overall condition fit score is obtained by summing and averaging the scores of all nursing staff for the entire shift.

[0043] The static optimization function is ,in For overall compatibility.

[0044] It should be further explained that, in the specific implementation process, the process of trigger-based dynamic real-time optimization and matching to output the optimal resource matching solution includes: Dynamic matching requires high real-time performance, local optimization, and strong constraints. An improved adaptive genetic algorithm is selected, which has the advantages of fast convergence speed and the ability to quickly find local optimal solutions, fully meeting the requirement of generating a solution within 10 seconds in emergency scenarios. Add real-time constraints to the veto hard constraint rule (1. Real-time status constraint: Nurses in a Level I prohibited state are forcibly excluded from the matching range; nurses in a Level II high-risk state are prohibited from being assigned new critically ill patients; 2. Response time constraint: In a Level I emergency response scenario, the matched nurse must arrive at the patient's bedside within 3 minutes. The arrival time is calculated based on the nurse's real-time physical location. Matches exceeding the time limit are directly invalid; 3. Operation interruption constraint: Nurses performing high-risk operations (such as thrombolysis administration, blood transfusion, and intracranial pressure tubing care) are prohibited from being assigned new emergency tasks to avoid safety risks caused by operation interruption), generating dynamic matching hard constraint rules. Based on real-time monitoring results, the local optimization domain in the current optimal resource matching scheme is locked, specifically: Only patients affected by the event and available nurses are included in the optimization scope; all other stable nurse-patient matching pairs are locked and not included in the optimization calculation. Example 1: When a new patient is admitted through the emergency green channel, the optimization domain is limited to the new patient plus all on-duty nurses in the ward who meet the hard constraints, and all existing patient-nurse matching pairs are locked. Example 2: If a nurse suddenly leaves her post, the optimization domain is limited to all patients under that nurse's care plus the remaining nurses in the ward who meet the hard constraints. All other matching pairs are locked. Subsequently, based on the dynamic matching hard constraint rules, invalid matching pairs in the local optimization domain are eliminated to generate feasible matching pairs. Several resource matching schemes are randomly generated based on the feasible matching pairs. Chromosome encoding and population initialization are performed on the several resource matching schemes to generate an initial feasible solution population. Add a perturbation objective function to the three types of objective functions. , , This value quantifies the degree to which a proposed solution disturbs existing stable matching relationships. A larger value indicates a smaller adjustment range and higher clinical stability, serving as the core constraint objective of dynamic matching. It constitutes four types of objective functions, with weight coefficients set based on response levels. A dynamic optimization function is constructed using these functions and weight coefficients. The initial feasible solution population is substituted into the dynamic optimization function to obtain fitness values. Subsequently, iterative optimization through adaptive selection, crossover, and mutation operations, along with secondary verification based on clinical practice habits, outputs the optimal resource matching solution (including only showing the adjusted nurse-patient pairings, the reason for the adjustment, and the responsibility relationship before and after the adjustment; the total working hours, remaining available working hours, and details of the responsible patients for the corresponding nurses after the adjustment). The original optimal resource matching solution is updated based on the output optimal solution. The optimal resource matching solution is then pushed to the devices of the corresponding nurses and head nurses via mobile nursing terminals within seconds, clearly specifying task requirements, treatment time limits, and core patient information. Once the nurses review and confirm, the solution takes effect immediately.

[0045] The dynamic optimization function is: ,in, The fitness value indicates the optimal solution; a higher fitness value indicates a better solution. When the response level is Level I emergency response, =0.45、 =0.15、 =0.15、 =0.25; when the response level is a Level II conventional adjustment response, =0.25、 =0.2、 =0.1、 =0.45.

[0046] In the process of outputting the optimal resource matching scheme through iterative optimization of adaptive selection, crossover, and mutation operations, and secondary verification based on clinical practice habits, this invention employs a multi-objective genetic algorithm for iterative optimization of adaptive selection, crossover, and mutation operations, and secondary verification based on clinical practice habits. First, resource matching schemes are encoded as chromosomes. For example, a string of numbers represents information such as nurse-patient matching pairs, the total working hours of the corresponding nurses after adjustment, the remaining available working hours, and the details of the patients they are responsible for. A certain number of chromosomes are randomly generated to form an initial feasible solution population. Each chromosome represents a possible resource matching scheme. The fitness function is obtained, and a tournament selection method is used to select chromosomes with higher fitness values ​​from the current population as parents. The parent chromosomes are crossovered, exchanging some genes to generate new offspring chromosomes. This simulates biological genetic gene exchange to generate new resource matching schemes. The offspring chromosomes are mutated, randomly changing some genes to increase population diversity and avoid getting trapped in local optima. The above steps are repeated iteratively until the termination condition is met. If the maximum number of iterations is reached or the fitness value no longer increases significantly, the optimal resource matching scheme is output. The optimal resource matching scheme is then subjected to a second verification based on clinical practice habits. If it does not meet the requirements, adaptive selection, crossover, and mutation operations are repeated iteratively to optimize until the output optimal resource matching scheme passes the second verification based on clinical practice habits.

[0047] It should be further explained that, in the specific implementation process, the process of real-time monitoring of the dynamic nursing needs profiles of all patients and the real-time work status profiles of all nursing staff, and determining whether to generate a response level based on the real-time monitoring results, includes: A Level I emergency response is triggered when the patient's dynamic nursing needs profile shows an upgrade in the severity / risk level of the patient's condition, or when the nurse's real-time work status profile shows a drop in the overall work status level to Level I (disabled) / Level II (high-risk), or when a new dynamic nursing needs profile of the patient is generated (e.g., a new patient admitted through the emergency stroke green channel requires urgent matching with a nurse). When the severity / risk level of the patient's condition is downgraded, the overall status level is upgraded / downgraded, or the patient's dynamic nursing needs profile is marked as discharged / transferred, a Level II routine adjustment response is triggered. When a Level I emergency response or a Level II routine adjustment response is triggered, the dynamic real-time optimization matching of the resource dynamic intelligent matching module is activated.

[0048] It should be further explained that, in the specific implementation process, a cerebrovascular disease specialty data prediction model is constructed based on comprehensive diagnosis, treatment, and nursing data of cerebrovascular disease patients. This model outputs multi-dimensional feature prediction indicators, and the process of determining whether to generate an early warning response level based on these indicators includes: Regarding the cerebrovascular disease data prediction model mentioned in this application, we hereby clarify that: the core architecture, implementation logic, input-output paradigm, and scenario-based adaptation method of this model have all been fully disclosed in the prior art and belong to conventional prior art in the fields of medical time-series data prediction and stroke clinical auxiliary decision-making. Those skilled in the art can implement it without creative effort. The inventive contribution of this invention does not lie in the model itself, but in the deep coupling of the model's prediction results with dynamic matching of nursing resources and closed-loop management of safety early warning. It extracts multi-dimensional feature indicators from the patient's adjacent historical collection period (288) and the current collection period (15 minutes) of the full-dimensional diagnosis and treatment nursing data of cerebrovascular disease patients, inputs the multi-dimensional feature indicators into the cerebrovascular disease data prediction model, outputs the multi-dimensional feature prediction indicators for the next period according to the cerebrovascular disease data prediction model, feeds the multi-dimensional feature prediction indicators back to the cerebrovascular disease-specific multi-dimensional assessment module, and outputs the patient's dynamic nursing needs prediction profile according to the cerebrovascular disease-specific multi-dimensional assessment module. When the dynamic nursing needs prediction profile of a patient shows an escalation in the severity of their condition / risk level, a Level I emergency warning level is generated for the patient. When a patient’s dynamic nursing needs profile shows a downgrade in the severity of their condition / risk level, a Level II routine adjustment warning level is generated for the patient. When a patient's Level I emergency alert level or Level II routine adjustment alert level is generated, the dynamic real-time optimization matching of the resource dynamic intelligent matching module is activated in advance. Specifically, based on the Level I emergency alert level or Level II routine adjustment alert level, the local optimization domain in the current optimal resource matching scheme is locked. Then, based on the dynamic matching hard constraint rules, invalid matching pairs in the local optimization domain are eliminated to generate feasible matching pairs. Several resource matching schemes are randomly generated based on the feasible matching pairs. Chromosome encoding and population initialization are performed on several resource matching schemes to generate an initial feasible solution population. The initial feasible solution population is substituted into the dynamic optimization function to obtain the fitness value. Then, through adaptive selection, crossover, and mutation operations, iterative optimization and secondary verification based on clinical practice habits are performed to output the optimal resource matching scheme for the next collection cycle.

[0049] like Figure 2 As shown, the method for dynamic matching and safety early warning management of nursing resources in cerebrovascular centers includes the following steps: Step s1: Collect comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients and comprehensive management data of nursing staff, mark the collection time, and set the collection cycle; Step s2: Based on the comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients, construct a multi-dimensional assessment system specifically for cerebrovascular disease, and conduct real-time quantitative assessment of the severity of the patient's condition, the risk of specialized nursing care, and the workload of nursing care, and output a dynamic nursing needs profile of the patient; Step s3: Based on the full-dimensional management data of nursing staff, conduct real-time quantitative assessment of cognitive load and physiological state dimensions, and output a real-time work status profile of nursing staff. The real-time work status profile of nursing staff includes comprehensive work status level and resource matching hard constraint rules. Step s4: Based on the dynamic nursing needs profile of patients and the dynamic capability profile of nursing staff, perform static basic scheduling matching and trigger-based dynamic real-time optimization matching of nursing resources, and output the optimal resource matching scheme; Step s5: Monitor the dynamic nursing needs profiles of all patients and the real-time work status profiles of all nursing staff in real time. Determine whether to generate a response level based on the real-time monitoring results. At the same time, construct a cerebrovascular specialty data prediction model based on the comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients, output multi-dimensional feature prediction indicators, and determine whether to generate an early warning response level based on the multi-dimensional feature prediction indicators. If a response level or early warning response level is generated, trigger dynamic real-time optimization matching.

[0050] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A dynamic matching and safety early warning management system for nursing resources in cerebrovascular centers, characterized in that, This includes a monitoring center, which is connected to a data acquisition module, a multi-dimensional assessment module specifically for cerebrovascular diseases, a comprehensive nursing resource profiling module, a dynamic intelligent resource matching module, and a real-time nursing monitoring and early warning module. The data acquisition module is used to collect comprehensive diagnosis, treatment, and nursing data of patients with cerebrovascular diseases and comprehensive management data of nursing staff, and to mark the collection time and set the collection cycle; The cerebrovascular disease-specific multidimensional assessment module is used to construct a cerebrovascular disease-specific multidimensional assessment system based on the full-dimensional diagnosis, treatment and nursing data of cerebrovascular disease patients. It can quantitatively assess the severity of the patient's condition, the risk of specialized nursing care and the workload of nursing care in real time, and output a dynamic nursing needs profile of the patient. The nursing resource full-dimensional profile module is used to conduct real-time quantitative assessment of cognitive load and physiological status based on the full-dimensional management data of nursing staff, and output a real-time work status profile of nursing staff, which includes a comprehensive work status level and hard constraint rules for resource matching. The resource dynamic intelligent matching module is used to perform static basic scheduling matching and trigger-based dynamic real-time optimization matching of nursing resources based on the dynamic nursing needs profile of patients and the dynamic capability profile of nursing staff, and output the optimal resource matching scheme. The process of performing static basic scheduling matching of nursing resources and outputting the optimal resource matching scheme includes: Construct a veto hard constraint rule, import the dynamic nursing needs profile of all patients and the real-time work status profile of all nursing staff, and based on the veto hard constraint rule, eliminate invalid matching pairs between patients and nursing staff in advance and generate feasible matching pairs. Several nurse-patient allocation and scheduling schemes are randomly generated based on feasible matching pairs. Three types of objective functions are constructed. A static optimization function is constructed based on the three types of objective functions. For each group of nurse-patient allocation and scheduling schemes, the static optimization function is substituted to obtain the comprehensive fit. The top 10% of nurse-patient allocation and scheduling schemes with the highest overall suitability were selected as the candidate optimal solution set. The candidate optimal solution set was then subjected to a second verification based on clinical practice habits. Candidate optimal solutions that did not conform to clinical practice habits were eliminated. From the candidate optimal solutions that passed the second verification based on clinical practice habits, the candidate optimal solution with the highest overall suitability was selected as the optimal resource matching scheme. The process of trigger-based dynamic real-time optimization matching and outputting the optimal resource matching solution includes: Add real-time constraint rules to the veto hard constraint rules to generate dynamic matching hard constraint rules. Based on real-time monitoring results, lock the local optimization domain in the current optimal resource matching scheme. Then, based on the dynamic matching hard constraint rules, remove invalid matching pairs in the local optimization domain to generate feasible matching pairs. Based on the feasible matching pairs, randomly generate several resource matching schemes. Perform chromosome encoding and initialize the population for several resource matching schemes to generate an initial feasible solution population. Based on the three types of objective functions, a perturbation objective function is added to form a four types of objective functions. The weight coefficients of the four types of objective functions are set according to the response level. A dynamic optimization function is constructed based on the four types of objective functions and the weight coefficients. The initial feasible solution population is substituted into the dynamic optimization function to obtain the fitness value. Then, through adaptive selection, crossover, mutation operations, iterative optimization and secondary verification based on clinical practice habits, the optimal resource matching scheme is output. The original optimal resource matching scheme is updated based on the output optimal resource matching scheme. The real-time nursing monitoring and early warning module is used to monitor the dynamic nursing needs profiles of all patients and the real-time work status profiles of all nursing staff in real time. Based on the real-time monitoring results, it determines whether to generate a response level. At the same time, based on the comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients, it constructs a cerebrovascular specialty data prediction model and outputs multi-dimensional feature prediction indicators. Based on the multi-dimensional feature prediction indicators, it determines whether to generate an early warning response level. If a response level or an early warning response level is generated, it triggers the dynamic real-time optimization matching of the resource dynamic intelligent matching module.

2. The dynamic matching and safety early warning management system for cerebrovascular center nursing resources according to claim 1, characterized in that, The process of real-time quantitative assessment of the severity of a patient's condition includes: Specialty features were extracted and standardized preprocessing was performed on the collected comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients to generate multi-dimensional feature indicators; A severity assessment system for patients' conditions was constructed, comprising primary sub-dimensions, weighting percentages of these sub-dimensions, secondary core assessment indicators, and specialized quantitative rules. Secondary core assessment indicators were extracted from the multi-dimensional feature indicators based on this system and standardized scoring was applied. A weighted average was then calculated for each secondary core assessment indicator based on its weighting percentage to obtain a comprehensive score for the patient's severity condition. Threshold points were selected within the comprehensive score range to divide the patient's condition into sub-intervals representing different severity levels. The severity level corresponding to the patient's comprehensive score was determined based on the sub-interval to which it belonged.

3. The dynamic matching and safety early warning management system for cerebrovascular center nursing resources according to claim 2, characterized in that, The process of real-time quantitative assessment of patients' specialist nursing risks and workload includes: A specialized nursing risk dimension assessment system was constructed, which includes specialized quantitative assessment indicators, the weighting of specialized quantitative assessment indicators, and evidence-based scoring rules. Based on the specialized nursing risk dimension assessment system, specialized quantitative assessment indicators corresponding to the multi-dimensional feature indicators were extracted, and standardized scores were assigned to the specialized quantitative assessment indicators based on evidence-based scoring rules. Subsequently, according to the weighting of each specialized quantitative assessment indicator, the scores of each specialized quantitative assessment indicator were weighted and averaged to obtain a comprehensive risk score. Within the comprehensive risk score range, threshold points were selected to divide the range into sub-intervals with different risk levels. Based on the sub-interval to which the comprehensive risk score belongs, the risk level corresponding to the comprehensive risk score was determined. A three-dimensional stroke-specific working hour quantification model was constructed. Multi-dimensional feature indicators, severity level of illness, and risk level were input into the three-dimensional stroke-specific working hour quantification model. Based on the three-dimensional stroke-specific working hour quantification model, the model outputs basic direct nursing hours, indirect nursing hours, and risk-additional hours, as well as the corresponding weight ratios of the three types of hours. Based on the basic direct nursing hours, indirect nursing hours, and risk-additional hours, as well as the corresponding weight ratios of the three types of hours, the total nursing hours were obtained. The nursing load level corresponding to different nursing hour intervals was preset, and the nursing load level corresponding to the working hour interval to which the total nursing hours belong was obtained. A dynamic nursing needs profile of patients is constructed based on their severity of illness, risk level, and nursing workload level.

4. The dynamic matching and safety early warning management system for cerebrovascular center nursing resources according to claim 3, characterized in that, The process of the nursing resource holistic profiling module, which uses comprehensive management data of nursing staff to quantitatively assess the physiological status dimension, includes: Core physiological indicators, working hours, and shift data are extracted from the comprehensive management data of nursing staff to construct a quantitative assessment system for physiological fatigue. This system includes secondary sub-indicators from the core physiological indicators and working hours / shift data, their corresponding weights, and specialized quantitative scoring rules. Based on these rules, standardized scores are assigned to the secondary sub-indicators. A weighted average is then calculated based on the score and weight of each secondary sub-indicator to obtain a comprehensive physiological fatigue score. Furthermore, different fatigue levels are preset for different comprehensive physiological fatigue score ranges, allowing for the acquisition of the corresponding fatigue level within each range.

5. The dynamic matching and safety early warning management system for cerebrovascular center nursing resources according to claim 4, characterized in that, The nursing resource comprehensive profiling module performs real-time cognitive load quantification assessment based on comprehensive management data of nursing staff, and outputs a real-time work status profile of nursing staff. The process includes: A cognitive load quantification model for patients is constructed. The dynamic nursing needs profiles of all patients under the care of each nurse are input into the cognitive load quantification model, and the basic patient load score is output based on the cognitive load quantification model. A real-time task cognitive load assessment model is constructed. Real-time operation data is extracted from the full-dimensional management data corresponding to each nurse. The real-time operation data is input into the real-time task cognitive load assessment model. The real-time task cognitive load assessment model performs standardized scoring on sub-dimensions based on the built-in specialized quantitative scoring rules. Then, a weighted average is performed based on the scores of each sub-dimension and the built-in sub-weights to output the real-time task cognitive load score. The cognitive load score of the real-time task and the basic patient load score are weighted and averaged to obtain the comprehensive cognitive load score. Different cognitive load scores are preset to correspond to different cognitive load levels, and the cognitive load level corresponding to the interval to which the comprehensive cognitive load score belongs is obtained. Construct a two-dimensional fusion matrix, which includes the comprehensive work status level corresponding to different fatigue levels and cognitive load levels, as well as the resource matching constraint parameters corresponding to different comprehensive work status levels. Based on the fatigue level and cognitive load level of each nurse and the two-dimensional fusion matrix, the comprehensive work status level and resource matching hard constraint rules of each nurse are obtained. Based on the comprehensive work status level and resource matching hard constraint rules of each nurse, a real-time work status profile of the nurse is output.

6. The dynamic matching and safety early warning management system for cerebrovascular center nursing resources according to claim 5, characterized in that, The process of real-time monitoring of dynamic nursing needs profiles for all patients and real-time work status profiles for all nursing staff, and determining whether to generate a response level based on the real-time monitoring results, includes: A Level I emergency response is triggered when the patient's dynamic nursing needs profile shows an upgrade in the severity / risk level of the patient's condition, or when the comprehensive work status level of the nursing staff in the real-time work status profile drops to Level I (disabled) / Level II (high-risk), or when a new dynamic nursing needs profile of the patient is generated. When the severity / risk level of the patient's condition is downgraded, the overall status level is upgraded / downgraded, or the patient's dynamic nursing needs profile is marked as discharged / transferred, a Level II routine adjustment response is triggered. When a Level I emergency response or a Level II routine adjustment response is triggered, the dynamic real-time optimization matching of the resource dynamic intelligent matching module is activated.

7. The dynamic matching and safety early warning management system for cerebrovascular center nursing resources according to claim 6, characterized in that, Based on comprehensive diagnosis, treatment, and nursing data of cerebrovascular disease patients, a cerebrovascular specialty data prediction model is constructed, outputting multi-dimensional feature prediction indicators. The process of determining whether to generate an early warning response level based on these multi-dimensional feature prediction indicators includes: Multidimensional feature indicators are extracted from the comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients in the patient's recent historical collection period and the current collection period. The multidimensional feature indicators are input into the cerebrovascular specialty data prediction model. Based on the cerebrovascular specialty data prediction model, the multidimensional feature prediction indicators for the next period are output. The multidimensional feature prediction indicators are fed back to the cerebrovascular disease-specific multidimensional assessment module. Based on the cerebrovascular disease-specific multidimensional assessment module, the dynamic nursing needs prediction profile of the patient is output. When the dynamic nursing needs prediction profile of a patient shows an escalation in the severity of their condition / risk level, a Level I emergency warning level is generated for the patient. When a patient’s dynamic nursing needs profile shows a downgrade in the severity of their condition / risk level, a Level II routine adjustment warning level is generated for the patient. When a patient's Level I emergency warning level or Level II routine adjustment warning level is generated, the dynamic real-time optimization matching of the resource dynamic intelligent matching module is activated in advance.

8. A method for dynamic matching and safety early warning management of nursing resources in cerebrovascular centers, implemented based on the dynamic matching and safety early warning management system for nursing resources in cerebrovascular centers as described in any one of claims 1 to 7, characterized in that, Includes the following steps: Step s1: Collect comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients and comprehensive management data of nursing staff, mark the collection time, and set the collection cycle; Step s2: Based on the comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients, construct a multi-dimensional assessment system specifically for cerebrovascular disease, and conduct real-time quantitative assessment of the severity of the patient's condition, the risk of specialized nursing care, and the workload of nursing care, and output a dynamic nursing needs profile of the patient; Step s3: Based on the full-dimensional management data of nursing staff, conduct real-time quantitative assessment of cognitive load and physiological state dimensions, and output a real-time work status profile of nursing staff. The real-time work status profile of nursing staff includes comprehensive work status level and resource matching hard constraint rules. Step s4: Based on the dynamic nursing needs profile of patients and the dynamic capability profile of nursing staff, perform static basic scheduling matching and trigger-based dynamic real-time optimization matching of nursing resources, and output the optimal resource matching scheme; Step s5: Monitor the dynamic nursing needs profiles of all patients and the real-time work status profiles of all nursing staff in real time. Determine whether to generate a response level based on the real-time monitoring results. At the same time, construct a cerebrovascular specialty data prediction model based on the comprehensive diagnosis and treatment nursing data of cerebrovascular disease patients, output multi-dimensional feature prediction indicators, and determine whether to generate an early warning response level based on the multi-dimensional feature prediction indicators. If a response level or early warning response level is generated, trigger dynamic real-time optimization matching.

Images