Nursing task scheduling method and system based on internet of things and intelligent algorithm

Abstract

This invention discloses a nursing task scheduling method and system based on the Internet of Things (IoT) and intelligent algorithms, belonging to the field of smart healthcare technology. Its key technical solutions encompass a complete process including data acquisition, preprocessing, intelligent allocation, dynamic scheduling, anomaly warning, and record traceability. This achieves a paradigm shift in nursing task scheduling from experience-driven to data-driven intelligence. By integrating multiple technologies such as IoT positioning, competency profile matching, neural network prediction, and blockchain storage, a complete intelligent scheduling ecosystem is constructed. This invention provides data support for scheduling decisions through real-time data acquisition and processing; achieves optimal matching of tasks and nursing staff through multi-dimensional intelligent algorithms; ensures resource balance through predictive scheduling and dynamic support mechanisms; and ensures scheduling quality through a full-process monitoring and traceability mechanism. This improves the overall efficiency of nursing task scheduling, reduces error rates, and promotes the intelligent upgrading of the nursing management system.

Description

Technical Field

[0001] This invention relates to the field of smart healthcare technology, and more specifically, to a nursing task scheduling method and system based on the Internet of Things and intelligent algorithms. Background Technology

[0002] With the continuous growth in demand for medical services and the increasing demands for quality, the workload of nursing care is rising sharply. Currently, most medical institutions still use the traditional manual nursing task scheduling model, which relies heavily on manual operation, such as assigning tasks via telephone notifications from nurses' stations or based on paper schedules. However, this traditional scheduling model has significant technical limitations in practice and is no longer suitable for the demands of modern medical environments for efficiency, safety, and accuracy in nursing management.

[0003] Specifically, the traditional nursing task scheduling model has the following core problems: First, the lack of transparent location information leads to low response efficiency. In this model, the real-time location of nursing staff cannot be automatically obtained. In emergency situations, the location of each staff member must be determined manually, with an average response time exceeding 15 minutes. Furthermore, because task assignment does not fully consider the actual distance between the nursing staff and the execution location, ineffective movement distances increase, significantly reducing work efficiency. Second, insufficient accuracy in matching capabilities poses compliance risks. Traditional scheduling methods lack an automated matching mechanism between nursing staff qualifications and task requirements. The competency files of nursing staff, such as qualification certifications, work experience, and training records, are not effectively linked to task requirements, easily leading to unqualified personnel performing special tasks, such as assigning ordinary nurses to operate specialized equipment like ventilators. Such behavior violates nursing operation compliance requirements and constitutes a serious medical safety hazard. Third, the lack of a workload prediction mechanism leads to an imbalance in human resource allocation. Current scheduling methods rely heavily on the experience and judgment of managers, failing to achieve scientific workload prediction based on historical task data, especially in accurately identifying peak daily workload periods. The consequences are a shortage of human resources during peak hours and idle staff during off-peak hours, resulting in a severe imbalance in resource allocation. Fourth, delayed cross-departmental collaborative support affects overall efficiency. When the workload of a department increases sharply, it is necessary to manually contact other departments for coordination and support, a process that takes an average of more than 30 minutes. Such delayed responses can easily miss the best nursing opportunities, such as delayed postoperative wound dressing changes, thus adversely affecting the quality of medical services and patient safety. Fifth, fragmented task management increases the difficulty of process traceability. The traditional model lacks a complete task execution record and archiving mechanism, and nursing task execution details rely on manual records, which pose a risk of omissions or tampering. At the same time, due to the lack of an effective task conflict detection mechanism, when the same nurse is assigned multiple tasks at the same time, it is difficult to identify task overlap, leading to task execution delays.

[0004] In summary, the existing technologies suffer from problems such as opaque nurse location information, insufficient accuracy in ability matching, lack of workload prediction mechanisms, lagging cross-departmental collaborative support, and fragmented task management. These issues severely restrict the efficiency and quality of nursing task scheduling, necessitating the development of a new solution that can achieve intelligent management throughout the entire process. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this invention is to provide a nursing task scheduling method and system based on the Internet of Things and intelligent algorithms. It covers the entire process of data acquisition, preprocessing, intelligent allocation, dynamic scheduling, abnormal early warning and record traceability, realizing a paradigm shift in nursing task scheduling from "experience-driven" to "data intelligence-driven". By integrating multiple technologies such as IoT positioning, competency profile matching, neural network prediction and blockchain storage, a complete intelligent scheduling ecosystem is constructed.

[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution: Firstly, a nursing task scheduling method based on the Internet of Things and intelligent algorithms is provided, including the following steps: The location data of nursing staff is collected in real time through IoT-based positioning badges, as well as their competency profile data and nursing task data. After cleaning and mapping, the location data is converted into department-ward text information. The competency profile data is classified according to qualification type and expired qualifications are marked. Qualification keywords are extracted and priority is marked from the nursing task data. Based on the preprocessed data, the location proximity, ability matching and task saturation of nursing staff are determined, and the optimal nursing staff are selected to assign tasks, while the shortest movement path is planned. The system predicts future workload through neural networks and dynamically optimizes the work schedule. When the system detects that the task saturation of a department has reached its upper limit, it automatically triggers a cross-departmental support mechanism and sends support requests to neighboring departments. Real-time detection of task overlap, qualification mismatch, and support timeout anomalies, and triggering alerts to the management backend, supporting manual intervention; Record all information about the entire task process and store the data in a blockchain database.

[0007] Furthermore, determining the location proximity, ability matching, and task saturation of caregivers includes: Based on the Euclidean distance between the nurse's current location and the task execution location, the proximity is quantified using a distance decay function or a first piecewise function to obtain the location proximity; where the shorter the distance, the higher the proximity, and the longer the distance, the lower the proximity. The matching degree is quantified based on the Jaccard similarity between the qualification set required for the task and the qualification set of nursing staff to obtain the ability matching degree; the higher the similarity, the higher the matching degree, and the matching degree is further optimized by combining the perfect match indicator and the years of nursing staff's work experience. The task saturation is obtained by quantifying the saturation based on the ratio of the current number of tasks a nurse is currently handling to the maximum number of tasks they can handle, using a second piecewise function. The lower the ratio, the lower the saturation; and the higher the ratio, the higher the saturation.

[0008] Furthermore, the selection of the optimal caregiver for task allocation includes: A comprehensive evaluation score is generated for each caregiver by comprehensively assessing the location proximity, ability matching, and task saturation. From the set of available nursing staff, select nursing staff whose comprehensive evaluation score is not lower than a preset threshold; From the selected nursing staff, the nurse with the highest comprehensive evaluation score was chosen as the task performer; When multiple caregivers have the same highest overall assessment score, the caregiver closest to the task location should be selected first.

[0009] Furthermore, the planning of the shortest movement path includes: Based on the hospital floor plan, an improved A-type... The algorithm plans a path from the nurse's current location to the task execution location, where the path cost is calculated using a heuristic function that combines the actual cost from the starting point to the node and the heuristic cost from the node to the destination. The path cost is dynamically adjusted, taking into account three factors: distance between nodes, elevator waiting time, and corridor congestion. The three factors are then weighted and summed using weighting coefficients. When the congestion level of any node in the path exceeds a preset threshold, path replanning is triggered, and the optimal path is recalculated based on the updated cost.

[0010] Furthermore, the prediction of future workload using a neural network includes: A bidirectional long short-term memory network is used to train historical nursing workload data to predict the workload for a specified future period. The bidirectional long short-term memory network includes an input layer, multiple hidden layers, and an output layer. The input layer receives historical workload data, and the output layer generates predicted values ​​for future workload. During network training, time series data are processed through forward and backward propagation, and robust loss functions are used to optimize model parameters, combined with regularization techniques to prevent overfitting.

[0011] Furthermore, the dynamically optimized shift schedule includes: Based on the predicted future workload, a scheduling optimization model is constructed, where the optimization objective is to minimize labor costs, and the constraints include ensuring that the number of people scheduled for each time period meets the predicted workload demand, and that the scheduling time of each nurse does not exceed their maximum working time. The optimization model is solved using integer programming to obtain the scheduling decision variables, and the scheduling table is dynamically adjusted based on the solution results.

[0012] Furthermore, the automatic triggering of the cross-departmental support mechanism includes: When the department's task saturation is detected to reach the first preset threshold and remain so for the first time threshold, a support request is automatically sent to a neighboring department within a preset distance range. The support request is directed to available nursing staff in neighboring departments whose workload is not higher than a second preset threshold, and the qualifications of the nursing staff must meet the task requirements of the department that issued the request. If no response is received within the second time threshold, the scope of the support request will be expanded. For available nursing staff, a comprehensive assessment is conducted based on their distance from the requesting department, their ability matching, and current availability, and the nursing staff with the best assessment results are selected as support personnel.

[0013] Furthermore, the anomalies of overlapping real-time detection tasks include: For each nurse, the time interval between the start of their assigned tasks was monitored. When multiple tasks are assigned to the same caregiver within the same time period, and the time interval between the start of any two tasks is less than a preset time threshold, a task overlap warning is triggered.

[0014] Furthermore, the recording of the entire task process information and storing the data in a blockchain database includes: Record the execution process data of each nursing task in real time, including the executor's identification, task type, priority, start time, end time, time taken, and execution result; The task records are organized in chronological order into a linked storage structure, where each storage unit contains the data hash value, timestamp, and association identifier with the previous storage unit. Based on the chained storage structure, it supports record querying and data statistical analysis by multiple dimensions such as personnel, department, time range, and task type.

[0015] Secondly, a nursing task scheduling system based on the Internet of Things and intelligent algorithms is provided, including: The data acquisition module is configured to collect the location data of nursing staff in real time through IoT positioning badges, as well as the nursing staff's competency profile data and nursing task data; The data processing module is configured to clean and map the location data and convert it into department-ward text information, classify the competency profile data according to qualification type and mark expired qualifications, and extract qualification keywords and label priorities from the nursing task data. The intelligent scheduling module is configured to determine the location proximity, ability matching and task saturation of nursing staff based on preprocessed data, and select the optimal nursing staff to assign tasks, while planning the shortest movement path. The scheduling optimization module is configured to predict future workload through neural networks and dynamically optimize the schedule. When the task saturation of a department reaches the upper limit, it automatically triggers a cross-department support mechanism and sends a support request to neighboring departments. The anomaly warning module is configured to detect task overlap, qualification mismatch, and support timeout anomalies in real time, and trigger warnings to the management backend, supporting manual intervention; The record traceability module is configured to record information throughout the entire task process and store the data in a blockchain database.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention proposes a nursing task scheduling method based on the Internet of Things (IoT) and intelligent algorithms. This method encompasses a complete technical solution covering data acquisition, preprocessing, intelligent allocation, dynamic scheduling, anomaly warning, and record traceability. It achieves a paradigm shift in nursing task scheduling from "experience-driven" to "data-driven intelligence." By integrating multiple technologies such as IoT positioning, competency profile matching, neural network prediction, and blockchain storage, it constructs a complete intelligent scheduling ecosystem. Specifically, the system provides data support for scheduling decisions through real-time data acquisition and processing; achieves optimal matching of tasks and nursing staff through multi-dimensional intelligent algorithms; ensures resource balance through predictive scheduling and dynamic support mechanisms; and ensures scheduling quality through a full-process monitoring and traceability mechanism. This systematic technical solution improves the overall efficiency of nursing task scheduling, reduces the error rate, and promotes the intelligent upgrading of the nursing management system.

[0017] 2. This invention constructs a high-quality data cleaning and standardization process by introducing Z-score outlier filtering, qualification validity determination, and Jaccard similarity matching algorithm. This solution effectively solves problems such as location jumps, expired qualifications, and keyword mismatches in the original data, laying a reliable data foundation for subsequent intelligent scheduling. Specific methods include: using sliding window mean filtering to eliminate positioning anomalies, controlling position accuracy error within 5 meters; automatically marking qualification validity periods to prevent unqualified personnel from participating in task allocation; and using keyword similarity calculation to achieve precise matching between task requirements and personnel capabilities.

[0018] 3. This invention achieves scientific screening of nursing staff through a comprehensive evaluation of location proximity, ability matching, and task saturation, overcoming the limitations of traditional scheduling that relies on only a single factor (such as distance or qualification), and constructs a multi-objective optimization decision-making model. In specific implementation, the distance decay function ensures the implementation of the principle of proximity allocation, the ability matching score ensures operational compliance, and the task saturation assessment effectively prevents staff overload. This comprehensive scoring mechanism improves the accuracy of staff matching and shortens the average response time to less than 3 minutes.

[0019] 4. This invention is based on improved A The algorithm incorporates real-world factors such as elevator waiting time and corridor congestion to achieve intelligent planning of nursing staff movement paths. It overcomes the limitations of traditional navigation systems that only consider distance, dynamically adjusting path selection through a cost function to ensure nursing staff reach their destinations in the shortest possible time. When the system detects path congestion, it automatically triggers a replanning mechanism to prevent nursing staff from getting stuck in congested areas. This dynamic path planning strategy reduces the average daily travel distance of nursing staff and improves time efficiency.

[0020] 5. This invention constructs a workload prediction model based on a bidirectional long short-term memory (LSTM) neural network. By analyzing 90 days of historical data, it achieves accurate prediction of the workload of each department within the next 24 hours. This model overcomes the subjective arbitrariness of experience-based scheduling, identifies workload fluctuation patterns through time series analysis, and provides a basis for precise allocation of human resources. The mean absolute percentage error (MAPE) of the prediction is controlled within 8.7%, increasing the matching degree between the scheduling plan and actual demand from 70% to 93%, effectively alleviating the problem of manpower shortage during peak hours and idle personnel during off-peak hours.

[0021] 6. The integer linear programming scheduling optimization model proposed in this invention aims to minimize labor costs and achieve scientific scheduling while meeting task requirements and individual working time constraints. This model, through quantifying constraints, ensures that the scheduling scheme meets both business needs and labor regulations, thereby improving labor utilization and significantly reducing the incidence of overtime work. Attached Figure Description

[0022] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart from Embodiment 1 of the present invention; Figure 2 This is a system block diagram in Embodiment 3 of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0024] Example 1: A nursing task scheduling method based on the Internet of Things and intelligent algorithms, such as Figure 1 As shown, it includes the following steps: S1: Real-time location data of nursing staff is collected through IoT positioning work badges, as well as nursing staff competency profile data and nursing task data; S2: After cleaning and mapping the location data, it is converted into department-ward text information. The competency profile data is classified according to the qualification type and expired qualifications are marked. Qualification keywords are extracted and priority is marked from the nursing task data. S3: Based on the preprocessed data, determine the location proximity, ability matching and task saturation of nursing staff, select the optimal nursing staff to assign tasks, and plan the shortest movement path. S4: Predict future workload through neural networks and dynamically optimize the work schedule. When the task saturation of a department reaches the upper limit, it automatically triggers a cross-departmental support mechanism and sends a support request to neighboring departments. S5: Real-time detection of task overlap, qualification mismatch, and support timeout anomalies, and triggers alerts to the management backend, supporting manual intervention; S6: Records the entire process of the task and stores the data in a blockchain database.

[0025] In step S1, nursing staff are equipped with IoT positioning badges (supporting Bluetooth + WiFi dual-mode positioning), and their location coordinates are collected in real time and transmitted to the dispatch system through the hospital intranet.

[0026] In some optional examples, Kalman filtering can be used to fuse multi-source positioning data to improve positioning accuracy: ; in, The optimal estimated position; To predict the location; Kalman gain; For the observation location; For observation matrix; positioning accuracy error rice.

[0027] The data collection for competency profiles mainly covers basic information of nursing staff (including name and title), professional qualification certifications (such as intravenous puncture, intensive care nursing, pediatric specialty nursing, etc.), clinical work experience (including years of service in each department and frequency of special clinical tasks performed), and continuing education training records (training content and corresponding assessment results for the past three years). All of the above data are uniformly entered and centrally stored in the nursing staff competency database.

[0028] Nursing task data collection: Receive and integrate nursing task requests from multiple sources, including doctor's orders, patient calls, and instructions issued by the nurses' station. Standardize and label each task, with specific dimensions including: task type (e.g., infusion, medication administration, health education, wound care, etc.), priority (emergency tasks must be responded to within 3 minutes, routine tasks within 15 minutes), execution requirements (e.g., "must have professional competence to operate an indwelling intravenous catheter"), and execution location (specifically down to the ward number and bed number).

[0029] In step S2, location data preprocessing: filtering outlier values, such as coordinate jumps caused by signal loss, and mapping the coordinates to department-ward text information through departmental area division (such as internal medicine ward, surgical ward) to facilitate understanding by dispatchers.

[0030] Outlier detection algorithm: The Z-score method is used to detect outliers. ; when Values ​​that are identified as outliers are removed. This is the current coordinate value; The average coordinates within the sliding window (window size 10 seconds); This represents the standard deviation of the coordinates within the sliding window.

[0031] Competency profile preprocessing: Categorize competency data by "qualification type" (e.g., "critical care nursing qualification" and "pediatric nursing qualification"), and mark each qualification with an "expiration date". Expired qualifications are automatically marked as "invalid".

[0032] Determination of Qualification Validity: ; in, This is an indicator function; it returns 1 if the condition is met, and 0 otherwise. The expiration date of the qualification; The current date; for .

[0033] Task data preprocessing: Extract "qualification keywords" (such as "venipuncture") from the task execution requirements and match them with keywords in the competency profile; assign color labels to tasks according to priority (red: emergency; blue: routine) for easy visual identification.

[0034] Keyword matching algorithms can use Jaccard similarity to calculate the matching degree. : ; in, A set of qualifications required for the task; A collection of qualifications for nursing staff; when "time" indicates a complete match. Time indicates partial matching. The time indicates a mismatch.

[0035] In step S3, the determination of location proximity is specifically as follows: based on the Euclidean distance between the current location of the nursing staff and the task execution location, the proximity is quantified using a distance decay function or a first piecewise function to obtain the location proximity; where the shorter the distance, the higher the proximity, and the longer the distance, the lower the proximity.

[0036] In some optional examples, a distance decay function is used to calculate the location score: ; in, The Euclidean distance (in meters) between the nurse's current location and the location where the task is performed; Distance attenuation coefficient (determined through clinical optimization) ); 30 is the maximum score for position.

[0037] Optimize using 500 sets of historical scheduling data The value that minimizes the response time: ; Constraint: Average distance traveled Rice (reduced by 30%).

[0038] In some alternative examples, a piecewise linear approximation can also be used to calculate the position score: .

[0039] The determination of competency matching degree is as follows: the matching degree is quantified based on the Jaccard similarity between the qualification set required for the task and the qualification set of nursing staff to obtain the competency matching degree; the higher the similarity, the higher the matching degree, and the matching degree is weighted and optimized by combining the perfect match indicator and the years of nursing staff's work experience.

[0040] For example, calculating the ability matching score: ; in, Jaccard similarity ( ); Weighting for qualifications ( ); To achieve a perfect match bonus score ( ); For empirical coefficients ( ); Experience score (normalized to) ).

[0041] and: ; in, For nursing staff, the length of service is required. The upper limit of experience ( Year).

[0042] Total score formula: ; Ability score range: .

[0043] The task saturation is determined as follows: based on the ratio of the current number of tasks to the maximum number of tasks that a nursing staff member can handle, the saturation is quantified by a second piecewise function to obtain the task saturation; the lower the ratio, the lower the saturation, and the higher the ratio, the higher the saturation.

[0044] For example, task saturation score calculation: ; in, Task saturation is calculated as follows: ; in, This represents the number of tasks currently being performed by nursing staff. Maximum workload for nursing staff (set according to nursing level: head nurse) Head Nurse nurse ,Nurse ).

[0045] The process of selecting the optimal nurses for task assignment includes: comprehensively assessing location proximity, ability matching, and task saturation to generate a comprehensive assessment value for each nurse; selecting nurses with comprehensive assessment values ​​not lower than a preset threshold from the set of assignable nurses; selecting the nurse with the highest comprehensive assessment value from the selected nurses as the task performer; and prioritizing the nurse closest to the task execution location when multiple nurses have the same highest comprehensive assessment value.

[0046] For example, total score Calculation and personnel selection: ; The total score range is as follows: .

[0047] Selection criteria: ; in, A pool of assignable nursing staff; For the optimal caregiver, argmax is the independent variable corresponding to the maximum value (finding the person with the highest score). The total score is the sum of scores for position, ability, and saturation. In mathematics, | represents a conditional constraint, meaning "the following conditions must be met". Stotal(i)≥60 is the screening threshold; the total score must be greater than or equal to 60. Nurses with scores below 60 are directly excluded and do not participate in the selection process.

[0048] When total scores are the same, the candidate with the closer distance will be given priority: ; in, The subset of nurses with the highest scores.

[0049] Planning the shortest movement path includes: based on the hospital floor plan, using an improved A... The algorithm plans the path from the nurse's current location to the task execution location. The path cost is calculated using a heuristic function, which combines the actual cost from the starting point to the node and the heuristic cost from the node to the destination. The path cost is dynamically adjusted, taking into account three factors: distance between nodes, elevator waiting time, and corridor congestion. These three factors are then weighted and summed using weighting coefficients. When the congestion of any node in the path exceeds a preset threshold, path replanning is triggered, and the optimal path is recalculated based on the updated cost.

[0050] For example, based on hospital floor plans Using an improved A The algorithm plans the optimal path.

[0051] Heuristic function design: ; in, From the starting point to the node The actual cost; For the node The heuristic cost of reaching the destination.

[0052] Dynamic cost function Considering dynamic factors such as elevator waiting time and corridor congestion: ; in, For nodes arrive Euclidean distance; Average elevator waiting time (real-time data collection, historical average) ); Corridor congestion coefficient ( (1 indicates smooth traffic, 3 indicates severe congestion). Weighting coefficients (optimized via grid search).

[0053] Path replanning strategy: When a node on the path is detected... When this occurs, path replanning is triggered: .

[0054] in, The congestion cost (or congestion index) for node / segment n is an indicator that quantifies the degree of congestion (e.g., the higher the value, the more severe the congestion). The currently planned path (consisting of a series of nodes / paths n); This is the starting point of the path; The endpoint of the path; The "updated weights" refer to the weights of edges / nodes that the A* algorithm relies on (such as distance, time, congestion cost, etc.). Here, it means that the path weights have been updated with the latest congestion data (or other dynamic information).

[0055] In step S4, the future workload is predicted using a neural network, including: training a bidirectional long short-term memory network on historical nursing task data to predict the workload for a specified future period; wherein the bidirectional long short-term memory network includes an input layer, multiple hidden layers and an output layer, the input layer receives historical task data, and the output layer generates predicted values ​​for future workload; during network training, time series data is processed through forward and backward propagation, and a robust loss function is used to optimize model parameters, combined with regularization techniques to prevent overfitting.

[0056] LSTM neural network architecture: Bi-LSTM is used to predict the workload of each department in the next 24 hours.

[0057] Network structure: Input layer: history Daily task volume (statistics based on 24 hours, input dimension) ); Hidden layers: 2 layers of Bi-LSTM, number of hidden units per layer ; Output layer: Task volume prediction for the next 24 hours (output dimension 24).

[0058] Forward propagation formula: ; ; ; ; in, For a moment Input features; The states are hidden in the forward and backward directions; For splicing operations; For predicting output; The hidden state at time t; As input to the hidden weights; To hide the weight; For bias terms; Weights from hidden layer to output layer; This is the output layer bias.

[0059] Loss function: Using the Huber loss function to improve robustness: ; ; in, This is the total loss function of the model (the smaller the value, the better the model fits). The total number of training samples (e.g., the total number of historical data for the task). Let be the Huber loss value for the i-th sample; This represents the actual value (actual task volume / workload). This is the Huber loss threshold. The model prediction value for the i-th sample.

[0060] Model training configuration: Optimizer: Adam, initial learning rate ; Batch size: ; Training rounds: Early stop strategy (patience=20); Regularization: Dropout (dropout_rate=0.3) + L2 regularization ).

[0061] Performance on the test set: Mean Absolute Percentage Error (MAPE) is The root mean square error (RMSE) is Tasks per hour; Determination coefficient for .

[0062] Dynamically optimizing the shift schedule includes: constructing a shift optimization model based on the predicted future workload, where the optimization objective is to minimize labor costs, and the constraints include ensuring that the number of people on shift in each time period meets the predicted workload requirements, and that the shift duration of each nurse does not exceed their maximum working hours; solving the optimization model using integer programming to obtain the shift decision variables, and dynamically adjusting the shift schedule based on the solution results.

[0063] Specifically, define the optimization problem: ; Constraints: ; ; ; Where t represents the hour. Number the nursing staff; For a moment No. The number of shifts for each nursing staff member (0 or 1); For a moment No. The labor cost of one nursing staff; For the predicted time Task volume; For target human resource utilization rate ( ); For the first The maximum daily working hours for each nursing staff member are determined using integer linear programming (ILP) with branch and bound.

[0064] An automatic cross-departmental support mechanism is triggered, including: when the task saturation of a department reaches a first preset threshold and remains so for a first time threshold, a support request is automatically sent to neighboring departments within a preset distance range; the support request is directed to available nursing staff in neighboring departments whose task saturation is not higher than a second preset threshold, and the qualifications of the nursing staff must meet the task requirements of the requesting department; if no response is received within the second time threshold, the scope of the support request is expanded; for available nursing staff, a comprehensive evaluation is conducted based on their distance from the requesting department, their ability matching degree, and their current availability, and the nursing staff with the best evaluation result is selected as the support staff.

[0065] For example, when the workload of a certain department is saturated And for 10 minutes, the system automatically sends notifications to nearby departments (distance) (meters) of available nursing staff (saturation) Send a support request. The support personnel must meet the department's task requirements. If there is no response within 5 minutes, expand the scope of the request.

[0066] Define support personnel matching score: ; in, , is the distance score ( rice); For ability matching degree, To support nurses' set of qualifications / skills (such as professional titles, certificates, and operational procedures). To support the required set of qualifications / skills for the mission; The matching degree between the support nurse's ability and the task, ranging from 0 to 1. , is the usability score ( (for task saturation). All of these are weighting coefficients.

[0067] Choose a strategy: ; in, The best support personnel selected in the end; It serves as a gathering place for supportive nursing staff from nearby departments.

[0068] In step S5, the real-time detection of task overlap anomalies includes: for each nurse, detecting the task start time interval of the assigned tasks; when it is detected that the same nurse is assigned multiple tasks in the same time period, and the task start time interval between any two tasks is less than a preset time threshold, a task overlap warning is triggered.

[0069] For example, exception type definition: Task overlap: The same caregiver at the same time (time interval) (minutes) assigned 2 or more tasks; Qualification mismatch: Nursing staff whose qualifications do not meet the requirements for task execution (e.g., being assigned ICU nursing tasks without ICU qualifications). Support timeout: No response was received within 5 minutes of the cross-departmental support request being sent.

[0070] Task overlap detection algorithm: For nursing staff Define task time overlap detection: ; in, For nursing staff Task overlap identifier; For nursing staff The task set; For the task The start time; For the task The start time; For time threshold ( minute); For logical OR operation; For indicator functions; when The task overlap warning will be triggered at any time.

[0071] Early warning mechanism: The system detects the above anomalies in real time, triggers an early warning (background pop-up window + audio-visual reminder), and pushes it to the nursing management backend, displaying the details of the abnormal task (task ID, personnel involved, and anomaly type).

[0072] Manual intervention support: The back-end supports nursing managers to view abnormal details, manually adjust task allocation (such as reassigning overlapping tasks to available staff), assign support staff, and automatically save adjustment records into the system.

[0073] In step S6, the entire task process information is recorded and stored in a blockchain database, including: real-time recording of the execution process data for each nursing task, including the executor's identifier, task type, priority, start time, end time, time consumed, and execution result; organizing the task records into a chain storage structure in chronological order, where each storage unit contains the data hash value, timestamp, and association identifier with the previous storage unit; based on the chain storage structure, record queries and data statistical analysis are supported by multiple dimensions such as personnel, department, time range, and task type.

[0074] Specifically, real-time recording: After a task is assigned, the system automatically records "executor, task type, priority, start time, end time, time taken, and execution result (success / failure and reason)". Nursing staff can confirm the task completion status through a mobile app.

[0075] Data storage: Records are stored in a blockchain database (ensuring immutability), supporting multi-dimensional queries by "personnel, department, time, and task type".

[0076] Blockchain storage structure: Each task record is recorded as a block: ; ; ; in, For the first One task block; For block headers; For blocks; The hash value of the previous block; For timestamps; For Merkle tree root hash; Used as a unique identifier for the task; It serves as a unique identifier for nursing staff; Task type; These are the start and end times, respectively. The result of the execution.

[0077] Hash calculate: .

[0078] Statistical analysis: Automatically generates nursing workload reports (such as the number of infusion tasks a nurse performs in a month), response efficiency reports (average response time for emergency tasks), and exception handling reports (number of overlapping tasks in a month), providing data support for performance accounting and management optimization.

[0079] Example 2: Taking the internal medicine department of a tertiary hospital (3 wards, 60 beds, 25 nursing staff) as a pilot project, the intelligent nursing task scheduling system of the present invention was implemented. The specific steps are as follows: (1) System deployment and data initialization (days 1-2).

[0080] We equipped 25 nursing staff with IoT-enabled location badges and completed the entry of hospital floor maps (marking the locations of departments, wards, elevators, and corridors).

[0081] Data on nursing staff competence was collected: 10 staff members were qualified for "intravenous puncture + critical care nursing", 8 staff members were qualified for "pediatric nursing", and 7 staff members were qualified for "routine nursing". Data on daily tasks in internal medicine over the past 3 months was entered (an average of 120 tasks per day, with a peak of 45 tasks per day between 8 and 10 am).

[0082] (2) Parameter configuration (day 3).

[0083] Task allocation parameters: Location score is out of 30 points (30 points within 5 meters, 5 points deducted for every additional 5 meters, 0 points beyond 50 meters).

[0084] The ability score is out of 40 (40 points for a perfect match, 20-39 points for a partial match, and 0 points for a no match).

[0085] Saturation score: 30 points (out of 10) Saturation 30 points Score 15-29 points. (0 points)

[0086] Warning threshold: Task saturation Cross-departmental support is triggered, and two tasks trigger overlapping warnings within 10 minutes of each other in the same time period.

[0087] Prediction Model: Workload prediction model trained based on LSTM neural network; Input: 90 days of historical workload data; Output: Workload prediction for the next 24 hours; Prediction Error: MAPE .

[0088] (3) Actual operation and effect verification (days 4-30).

[0089] Emergency task dispatch: Patient suddenly experiences intravenous extravasation (emergency task, execution location: Ward 2, Bed 3 of Internal Medicine). The system locates nursing staff within 50 meters in real time and selects nurse Zhang XX who is "qualified for intravenous puncture and has an intravenous saturation of 40%".

[0090] Example of three-dimensional scoring calculation: Assume the parameters for nurse Zhang XX: Distance: 25 meters; Ability Match: Perfect Match (Jaccard Similarity) ), with 5 years of experience in the industry; Task saturation: Currently executing 2 tasks. saturation .

[0091] Score Calculation: Position score: .

[0092] Ability Score: ; However, the maximum ability score is 40, therefore .

[0093] Saturation score: .

[0094] Total Score: .

[0095] The system selected Zhang XX as the executor and planned the shortest path (from the nurses' station to bed 3, a distance of 25 meters). Zhang XX arrived at the scene within 3 minutes, reducing the response time by 80% compared to before the pilot test.

[0096] Dynamic scheduling: The predicted workload was 45 tasks between 8 and 10 a.m. (35 tasks exceeding the normal capacity). The system automatically added 2 nurses to the schedule, and the manpower saturation dropped from 92% to 68% during this period, with no task delays.

[0097] Scheduling optimization calculation Standard capacity: 35 items; predicted task volume: 45 items; target manpower utilization rate: 90%.

[0098] Manpower required:

[0099] Current shift: 25 people, additional staff needed: people.

[0100] Due to limitations in the department's actual staffing levels, the system will add 2 support nurses, bringing the staffing saturation point to: ; By increasing the workload per nurse (from 35 items / 8 hours to 40 items / 8 hours), staff saturation is achieved: ; By dynamically allocating resources, we can meet the needs of tasks during peak periods.

[0101] Cross-departmental support: With a surge in postoperative patients in internal medicine and a task saturation of 85%, the system sends a support request to the nearest surgical department (within 150 meters).

[0102] Assuming there are 3 available nurses in the surgical department, the support matching calculation is shown in Table 1.

[0103] Table 1 Support Matching Calculation Results

[0104] Wang XX (highest score) and Li XX were selected as support personnel. Within 5 minutes, two available nurses with "postoperative care qualifications" responded, reducing the support response time by 70%.

[0105] Anomaly alert: The system detected that nurse Li XX was simultaneously assigned two tasks: "administer medication at 9:00" and "infuse intravenous fluid at 9:05", with a time interval of 5 minutes. Minutes later, an overlapping warning was triggered.

[0106] Overlap detection: ; , This triggered an alert.

[0107] The nursing manager reassigned the infusion task to an available nurse, Wang XX, within 5 minutes, without any delay.

[0108] Record tracking: Monthly statistics show that the completion rate of internal medicine nursing tasks increased from 90% to 99%, the average task time was shortened by 15%, and the satisfaction of nursing staff increased from 72 points (out of 100) to 89 points.

[0109] Pilot results show that the intelligent scheduling system for nursing tasks of the present invention can effectively solve the pain points of traditional scheduling and has practical application value.

[0110] Example 3: A nursing task scheduling system based on the Internet of Things and intelligent algorithms. This system is used to implement the nursing task scheduling method based on the Internet of Things and intelligent algorithms described in Example 1, such as... Figure 2 As shown, it includes a data acquisition module, a data processing module, an intelligent scheduling module, a shift optimization module, an anomaly warning module, and a record tracing module.

[0111] The system includes the following modules: a data acquisition module, configured to collect real-time location data of nursing staff via IoT-based location badges, as well as their competency profiles and nursing task data; a data processing module, configured to clean and map the location data into department-ward text information, classify competency profile data by qualification type and mark expired qualifications, and extract qualification keywords and prioritize nursing task data; an intelligent scheduling module, configured to determine the location proximity, competency matching, and task saturation of nursing staff based on preprocessed data, select the optimal nursing staff to assign tasks, and plan the shortest movement path; a scheduling optimization module, configured to predict future workload through neural networks and dynamically optimize the scheduling table, and automatically trigger a cross-departmental support mechanism to send support requests to neighboring departments when the department's task saturation reaches its upper limit; an anomaly warning module, configured to detect task overlap, qualification mismatch, and support timeout anomalies in real time, and trigger warnings to the management backend, supporting manual intervention; and a record and traceability module, configured to record the entire task process information and store the data in a blockchain database.

[0112] Working Principle: This invention encompasses a complete technical solution for data acquisition, preprocessing, intelligent allocation, dynamic scheduling, anomaly warning, and record traceability. It achieves a paradigm shift in nursing task scheduling from "experience-driven" to "data-driven intelligence." By integrating technologies such as IoT positioning, competency profile matching, neural network prediction, and blockchain storage, it constructs a complete intelligent scheduling ecosystem. Specifically, the system provides data support for scheduling decisions through real-time data acquisition and processing; achieves optimal matching of tasks and nursing staff through multi-dimensional intelligent algorithms; ensures resource balance through predictive scheduling and dynamic support mechanisms; and ensures scheduling quality through a full-process monitoring and traceability mechanism. This systematic technical solution improves the overall efficiency of nursing task scheduling, reduces error rates, and promotes the intelligent upgrading of the nursing management system.

[0113] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0114] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0115] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0116] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0117] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A nursing task scheduling method based on the Internet of Things and intelligent algorithms, characterized in that, Includes the following steps: The location data of nursing staff is collected in real time through IoT-based location badges, as well as their competency profile data and nursing task data. After cleaning and mapping, the location data is converted into department-ward text information. The competency profile data is classified according to qualification type and expired qualifications are marked. Qualification keywords are extracted and priority is marked from the nursing task data. Based on the preprocessed data, the location proximity, ability matching and task saturation of nursing staff are determined, and the optimal nursing staff are selected to assign tasks, while the shortest movement path is planned. The system predicts future workload through neural networks and dynamically optimizes the work schedule. When the system detects that the task saturation of a department has reached its upper limit, it automatically triggers a cross-departmental support mechanism and sends support requests to neighboring departments. Real-time detection of task overlap, qualification mismatch, and support timeout anomalies, and triggering alerts to the management backend, supporting manual intervention; Record all information about the entire task process and store the data in a blockchain database.

2. The nursing task scheduling method based on the Internet of Things and intelligent algorithms according to claim 1, characterized in that, The determination of the location proximity, ability matching, and task saturation of nursing staff includes: Based on the Euclidean distance between the nurse's current location and the task execution location, the proximity is quantified using a distance decay function or a first piecewise function to obtain the location proximity; where the shorter the distance, the higher the proximity, and the longer the distance, the lower the proximity. The matching degree is quantified based on the Jaccard similarity between the qualification set required for the task and the qualification set of nursing staff to obtain the ability matching degree; the higher the similarity, the higher the matching degree, and the matching degree is further optimized by combining the perfect match indicator and the years of nursing staff's work experience. The task saturation is obtained by quantifying the saturation based on the ratio of the current number of tasks a nurse is currently handling to the maximum number of tasks they can handle, using a second piecewise function. The lower the ratio, the lower the saturation; and the higher the ratio, the higher the saturation.

3. The nursing task scheduling method based on the Internet of Things and intelligent algorithms according to claim 1, characterized in that, The selection of the optimal nursing staff for task allocation includes: A comprehensive evaluation score is generated for each caregiver by comprehensively assessing the location proximity, ability matching, and task saturation. From the set of available nursing staff, select nursing staff whose comprehensive evaluation score is not lower than a preset threshold; From the selected nursing staff, the nurse with the highest comprehensive evaluation score was chosen as the task performer; When multiple caregivers have the same highest overall assessment score, the caregiver closest to the task location should be selected first.

4. The nursing task scheduling method based on the Internet of Things and intelligent algorithms according to claim 1, characterized in that, The shortest travel path to be planned includes: Based on the hospital floor plan, an improved A-type... The algorithm plans a path from the nurse's current location to the task execution location, where the path cost is calculated using a heuristic function that combines the actual cost from the starting point to the node and the heuristic cost from the node to the destination. The path cost is dynamically adjusted, taking into account three factors: distance between nodes, elevator waiting time, and corridor congestion. The three factors are then weighted and summed using weighting coefficients. When the congestion level of any node in the path exceeds a preset threshold, path replanning is triggered, and the optimal path is recalculated based on the updated cost.

5. The nursing task scheduling method based on the Internet of Things and intelligent algorithms according to claim 1, characterized in that, The method of predicting future workload using neural networks includes: A bidirectional long short-term memory network is used to train historical nursing workload data to predict the workload for a specified future period. The bidirectional long short-term memory network includes an input layer, multiple hidden layers, and an output layer. The input layer receives historical workload data, and the output layer generates predicted values ​​for future workload. During network training, time series data are processed through forward and backward propagation, and robust loss functions are used to optimize model parameters, combined with regularization techniques to prevent overfitting.

6. The nursing task scheduling method based on the Internet of Things and intelligent algorithms according to claim 1, characterized in that, The dynamically optimized work schedule includes: Based on the predicted future workload, a scheduling optimization model is constructed, where the optimization objective is to minimize labor costs, and the constraints include ensuring that the number of people scheduled for each time period meets the predicted workload demand, and that the scheduling time of each nurse does not exceed their maximum working time. The optimization model is solved using integer programming to obtain the scheduling decision variables, and the scheduling table is dynamically adjusted based on the solution results.

7. The nursing task scheduling method based on the Internet of Things and intelligent algorithms according to claim 1, characterized in that, The automatic triggering of the cross-departmental support mechanism includes: When the department's task saturation is detected to reach the first preset threshold and remain so for the first time threshold, a support request is automatically sent to a neighboring department within a preset distance range. The support request is directed to available nursing staff in neighboring departments whose workload is not higher than a second preset threshold, and the qualifications of the nursing staff must meet the task requirements of the department that issued the request. If no response is received within the second time threshold, the scope of the support request will be expanded. For available nursing staff, a comprehensive assessment is conducted based on their distance from the requesting department, their ability matching, and current availability, and the nursing staff with the best assessment results are selected as support personnel.

8. The nursing task scheduling method based on the Internet of Things and intelligent algorithms according to claim 1, characterized in that, The anomalies of overlapping real-time detection tasks include: For each nurse, the time interval between the start of their assigned tasks was monitored. When multiple tasks are assigned to the same caregiver within the same time period, and the time interval between the start of any two tasks is less than a preset time threshold, a task overlap warning is triggered.

9. The nursing task scheduling method based on the Internet of Things and intelligent algorithms according to claim 1, characterized in that, The process of recording the entire task information and storing the data in a blockchain database includes: Record the execution process data of each nursing task in real time, including the executor's identification, task type, priority, start time, end time, time taken, and execution result; The task records are organized in chronological order into a linked storage structure, where each storage unit contains the data hash value, timestamp, and association identifier with the previous storage unit. Based on the chained storage structure, it supports record querying and data statistical analysis by multiple dimensions such as personnel, department, time range, and task type.

10. A nursing task scheduling system based on the Internet of Things and intelligent algorithms, characterized in that, include: The data acquisition module is configured to collect the location data of nursing staff in real time through IoT positioning badges, as well as the nursing staff's competency profile data and nursing task data; The data processing module is configured to clean and map the location data and convert it into department-ward text information, classify the competency profile data according to qualification type and mark expired qualifications, and extract qualification keywords and label priorities from the nursing task data. The intelligent scheduling module is configured to determine the location proximity, ability matching and task saturation of nursing staff based on preprocessed data, and select the optimal nursing staff to assign tasks, while planning the shortest movement path. The scheduling optimization module is configured to predict future workload through neural networks and dynamically optimize the schedule. When the task saturation of a department reaches the upper limit, it automatically triggers a cross-department support mechanism and sends a support request to neighboring departments. The anomaly warning module is configured to detect task overlap, qualification mismatch, and support timeout anomalies in real time, and trigger warnings to the management backend, supporting manual intervention; The record traceability module is configured to record information throughout the entire task process and store the data in a blockchain database.

Images