Intelligent resource scheduling optimization method based on artificial intelligence

Abstract

The invention belongs to the technical field of artificial intelligence, and discloses an intelligent resource scheduling optimization method based on artificial intelligence. The method comprises the following specific steps: S1, multi-modal data acquisition and integration; s2, unstructured data analysis is introduced; s3, constructing and applying a dynamic demand prediction model; s4, carrying out cross-department resource priority modeling and optimization; s5, deploying a real-time elastic scheduling engine; s6, designing and interacting a man-machine collaborative decision-making interface; and S7, performing closed-loop feedback optimization and block chain evidence storage auditing. The system has three outstanding advantages that firstly, multi-modal data acquisition and integration help to analyze illness conditions, realize precise medical treatment and optimize operation; 2, cross-department resource priority modeling and optimization are carried out, sorting is carried out according to an algorithm, equipment is shared, ethics are embedded, treatment is guaranteed, and efficiency and satisfaction are improved; and thirdly, man-machine collaborative decision-making interface design and interaction, AR visualization, man-machine game, real-time feedback, management improvement, scheduling guarantee and coping capability enhancement are realized.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology, and specifically relates to an intelligent resource scheduling optimization method based on artificial intelligence. Background Art

[0002] In the traditional hospital operating structure, resource allocation is mired in numerous difficulties, hindering the quality of medical services from reaching a new level. From the perspective of the allocation model, it mainly relies on medical staff to make judgments based on past experience. Once the emergency department encounters a sudden peak, a large number of wounded people pour in, or key diagnostic equipment such as CT and MRI (magnetic resonance imaging) suddenly malfunctions, it is difficult to quickly and accurately coordinate manpower and allocate materials, delaying the treatment of patients.

[0003] The disadvantages of static scheduling and bed allocation are obvious. The workload of medical staff is unbalanced, the utilization efficiency of medical resources is greatly reduced, and the idle rate of MRI equipment reaches 30%, resulting in waste of resources. In addition, information is not smooth and collaboration is sluggish among multiple departments. There is a lack of real-time data to support dynamic decision-making. Faced with the sudden outbreak of the epidemic and the periodic outbreak of seasonal diseases, the hospital is unable to make advance arrangements based on data and predict the fluctuations in demand. It can only respond passively, which restricts the improvement of medical service quality. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent resource scheduling optimization method based on artificial intelligence to solve the problems raised in the above background technology.

[0005] In order to achieve the above objectives, the present invention provides the following technical solutions: The specific steps of an intelligent resource scheduling optimization method based on artificial intelligence are as follows:

[0006] S1: Multimodal data collection and integration: By fully integrating HIS systems to extract structured data, deploying IoT devices to capture patients' physiological status, deeply integrating electronic medical records to mine key diagnosis and treatment information, and collecting hospital environment and patient flow data, all-round information aggregation lays a solid foundation for creating an accurate digital twin of hospital resources;

[0007] S2: Introducing unstructured data analysis: In emergency scenarios, we focus on medical staff voice recordings and use NLP technology to identify key information in emergency treatment communications. We deeply mine large amounts of emergency voice data to build a critical case prediction model, convert unstructured voice into quantitative indicators, predict resource needs, and effectively improve the timeliness and accuracy of emergency response.

[0008] S3: Construction and Application of Dynamic Demand Forecasting Model: Using the spatiotemporal graph neural network (ST-GNN), we construct a spatiotemporal graph of hospital department layout, patient flow, and time trends. Using historical data to train the model, we learn the demand patterns for patients, drugs, and equipment across departments over time, predicting details for the next 24 hours and improving operating room utilization efficiency.

[0009] S4: Modeling and Optimizing Cross-Departmental Resource Priority: Using the Q-learning algorithm, dynamic priority rules are established based on medical urgency and resource scarcity, prioritizing emergency care, critical care, and outpatient care. This system optimizes cross-departmental sharing of high-value da Vinci robotic devices, allocating time slots based on surgical schedules and medical complexity. Ethical weights are also embedded, with waiting penalty factors set for elderly patients, balancing fairness and humanity.

[0010] S5: Real-time elastic scheduling engine deployment: Edge computing nodes are deployed in key areas of the hospital, updating multi-source data every 5 minutes. Based on real-time data, a medical staff skill matrix matching algorithm is run to dynamically allocate manpower based on skills and needs. A path planning algorithm is also used to plan routes for mobile medical equipment, achieving precise deployment and supporting real-time optimization of massive concurrent variables.

[0011] S6: Human-machine collaborative decision-making interface design and interaction: Build an AR visualization system to display key hospital information in 3D, such as operating room-ICU linkage heat maps to facilitate insights. AI-powered scheduling recommendations are generated based on data algorithms, game theory is introduced to take into account the director's experience, and when disagreements arise, the system interacts to find a balance. Managers can manually correct and provide feedback, optimize the model, and improve the effectiveness of human-machine collaborative decision-making.

[0012] S7: Closed-loop feedback optimization and blockchain evidence audit: Relying on the digital twin of hospital resources, simulate multi-scenario scheduling effects, use GAN to create an extreme data strong model, introduce blockchain, record the entire chain information of resource scheduling, and distribute epidemic prevention materials in a compliant and transparent manner in line with JCI standards, which not only ensures refined management, but also builds a solid data trust foundation for improving medical quality.

[0013] Preferably, the specific steps of multimodal data collection and integration in S1 are as follows:

[0014] Step 1: Structured medical data extraction: Fully access the HIS system to accurately extract basic patient information, detailed diagnosis and treatment process records, and detailed cost details. This data is clearly organized and easy to analyze, building a basic medical process framework and providing key underlying information support for subsequent in-depth research on disease trends and resource consumption.

[0015] Step 2: Real-time patient physiological monitoring: Deploy IoT devices to closely monitor the patient's bed status, turning frequency, and subtle changes. Wearable devices can continuously collect heart rate, blood pressure, and blood oxygen saturation 24 hours a day, capturing the patient's physiological dynamics in real time and from all angles, laying a solid data foundation for precision medical decision-making.

[0016] Step 3: Integration of environmental and crowd flow data: The system collects environmental data from all parts of the hospital, including the temperature and humidity in wards, operating rooms, and waiting areas, which are related to patients' physical sensations and the stable operation of medical equipment. At the same time, crowd density monitoring accurately reflects the busyness of operations and provides early warning of the risk of crowd gathering.

[0017] Preferably, the introduction of unstructured data analysis in S2 refers to focusing on carrying out sophisticated NLP processing on voice records during the emergency process, and accurately capturing the key points in the communication between medical staff with the help of cutting-edge natural language processing technology, and accurately extracting keywords describing the patient's symptoms such as chest pain and difficulty breathing, as well as verbal expressions of the severity of the disease, and the general direction of the preliminary diagnosis. Then, relying on big data thinking, in-depth analysis of massive emergency voice data, and full efforts to build a critical case prediction model based on semantic features, in-depth exploration of the inherent logical connection between different expression patterns and the diagnosis of critical illness, and cleverly transforming scattered, unstructured voice information into quantitative data indicators with great predictive value.

[0018] Preferably, the construction and application of the dynamic demand forecasting model in S3 refers to the use of the spatiotemporal graph neural network ST-GNN to integrate the hospital department layout, patient flow path and time trend, and construct a complex spatiotemporal graph structure, so that the model can learn the dynamic laws related to patients, drugs and equipment in each department at different time periods, laying the foundation for accurate prediction; training ST-GNN with rich historical data allows it to master the demand change pattern of each department, accurately predict the patient flow per hour in the next 24 hours, estimate drug consumption by category and dosage form, and calculate demand based on equipment use and failure conditions, and output a detailed demand list; strictly verify the prediction results to ensure that the overall error rate is less than 8%, apply the model to actual scenarios, and deploy the anesthesia team in advance based on the operating room occupancy rate prediction, optimize the surgical process, and effectively improve the hospital resource allocation efficiency and department operation efficiency.

[0019] Preferably, the specific steps of cross-departmental resource priority modeling and optimization in S4 are as follows:

[0020] Step 1: Establish a dynamic priority system: Using the Q-learning algorithm, we fully consider the degree of medical urgency and resource scarcity, and clearly prioritize each department. The emergency department is ranked first due to its criticality and timeliness, followed by the intensive care unit, and the outpatient department is ranked relatively low. This ensures that critically ill patients have priority access to medical resources and guarantees timely life-saving treatment.

[0021] Step 2: Build a high-value equipment sharing model: For the high-end da Vinci robotic device, we will break down departmental barriers and establish a sharing optimization model based on the patient's surgical schedule and the complexity of the condition. This model will comprehensively weigh the difficulty of the surgery and the expected efficacy, scientifically plan the equipment usage period, and promote the circulation of equipment throughout the hospital, thereby improving utilization efficiency, reducing idle rates, and meeting the needs of more patients.

[0022] Step 3: Embed ethical weights to ensure fairness: Innovatively integrate ethical considerations into the resource allocation model, and set a waiting time penalty factor specifically for vulnerable groups such as elderly patients. When resource allocation decisions are made, this factor will come into play to avoid excessive resource allocation, balance fairness and efficiency, and make medical resource allocation more in line with the concept of humanistic care.

[0023] Preferably, the deployment of the real-time elastic scheduling engine in S5 refers to the widespread deployment of edge computing nodes in key areas of the hospital, including nurse stations, operating room scheduling centers, and logistics support departments. With its proximity to data sources and low latency, it quickly collects multi-source data from HIS systems and IoT devices every 5 minutes, grasps the hospital's operational dynamics in real time, and provides immediate basis for subsequent decision-making. Based on the collected real-time data, the medical staff skill matrix matching algorithm is activated to systematically sort out the medical staff's intubation and emergency resuscitation professional skills. At the same time, combined with the urgency of patient needs and the busyness of the department, the most suitable personnel are intelligently and dynamically assigned to the corresponding positions to ensure efficient and high-quality medical services. For mobile medical equipment such as defibrillators and infusion pumps, a path planning algorithm is used to comprehensively consider the current location of the equipment, the urgency of the needs issued by various departments, and the congestion status of the hospital passages to plan the optimal movement route for the equipment, so that the equipment can be accurately deployed in the shortest time. Technically, real-time optimization of massive concurrent variables is achieved to ensure that patients can receive timely treatment.

[0024] Preferably, the specific steps of designing and interacting with the human-machine collaborative decision-making interface in S6 are as follows:

[0025] Step 1: Build an AR visualization platform: Create an AR visualization system that converts the hospital's resource layout, including detailed information about the distribution of wards, operating rooms, and medical equipment storage locations, as well as real-time patient flow and pre-set scheduling plans, into intuitive 3D images. Using an operating room-ICU bed linkage heat map, managers can clearly see the connection between surgery and postoperative monitoring, accurately grasping key processes.

[0026] Step 2: Human-machine collaborative recommendation generation: The AI model uses complex algorithms based on massive amounts of medical data to quickly generate and push preliminary scheduling and planning recommendations. Game theory is also incorporated to fully respect the clinical director's intuitive judgment, formed through years of experience. When disagreements arise, both parties interact through a game-like game, weighing the pros and cons, working together to find the optimal balance and formulate scientific and practical decisions.

[0027] Step 3: Real-time feedback optimization model: After receiving the fusion solution, if managers find that there are still areas for improvement, they can manually correct them directly in the visual interface and provide real-time feedback to the AI model. The model receives new instructions, uses deep learning to correct the causes, and dynamically adjusts its own decision-making logic, continuously improving its ability to respond to complex medical scenarios and achieving continuous improvement in the effectiveness of human-machine collaborative decision-making.

[0028] Preferably, the closed-loop feedback optimization and blockchain evidence audit in S7 refer to the simulation and evaluation of scheduling strategies: with the help of the digital twin of hospital resources, different scheduling strategies are simulated for daily peaks and extreme scenarios of large-scale traffic accidents and infectious disease outbreaks, and resource allocation, patient treatment, and cost-effectiveness indicators are comprehensively evaluated to understand the pros and cons of each strategy; strengthen the extreme adaptability of the model: use GAN to generate extreme scenario data, train the model to adapt to rare and critical situations, and output reliable scheduling plans even under extreme pressure to ensure the hospital's emergency response capabilities; blockchain ensures transparent decision-making: introduce blockchain to record information on the entire process of resource scheduling, strictly comply with the distribution of epidemic prevention materials, ensure fine management standards, and build a solid foundation of trust for improving medical quality.

[0029] The beneficial effects of the present invention are as follows:

[0030] 1. The present invention has multiple significant benefits through multimodal data collection and integration. First, structured medical data extraction and access to the HIS system, the patient, diagnosis and treatment, and cost information obtained to build a basic process framework, help analyze the condition and resource utilization, and make medical decisions more targeted. Second, real-time patient physiological monitoring relies on IoT devices and wearable devices to accurately grasp the patient's dynamics, lay a solid foundation for precision medicine, and can promptly detect subtle changes in the patient's body, facilitating early intervention. Finally, environmental and crowd data are integrated to collect temperature, humidity, and crowd density, which not only ensures patient comfort and equipment stability, but also warns of gathering risks and optimizes hospital operations.

[0031] 2. This invention achieves remarkable results through cross-departmental resource priority modeling and optimization. First, the dynamic priority system uses the Q-learning algorithm to rationally sort departments according to urgency and scarcity, with emergency departments given priority, critical cases second, and outpatient departments last. This allows critically ill patients to receive treatment quickly, seizes the golden rescue time, and greatly enhances life support. Second, the high-value equipment sharing model breaks down departmental boundaries and plans the use period of the Da Vinci robot according to surgical schedules and patient conditions. As a result, equipment utilization rates soar, idle rates drop significantly, and more patients can benefit. Third, ethical weights are embedded, and a penalty factor is set for elderly patients, taking into account fairness, making resource allocation more heartwarming, and improving patient satisfaction with medical treatment.

[0032] 3. The present invention brings many benefits through the design and interaction of the human-machine collaborative decision-making interface. First, the AR visualization platform presents the hospital's complex information in three dimensions. The operating room-ICU bed linkage heat map allows managers to have a clear view of the surgical connection, accurately control key processes, and greatly improve management efficiency. Secondly, in the human-machine collaborative suggestion generation link, AI and clinical directors interact with each other, which not only gives play to the advantages of AI big data algorithms, but also respects experience and judgment, and jointly produces scientific and practical decisions to ensure the rationality of scheduling. Finally, the real-time feedback optimization model allows managers to manually correct and provide real-time feedback as needed, prompting AI to dynamically learn and optimize logic, and continuously improve the ability of human-machine collaboration to cope with complex medical scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is the process of the intelligent resource scheduling optimization method based on artificial intelligence of the present invention. DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0035] like Figure 1 As shown, the embodiment of the present invention provides an intelligent resource scheduling optimization method based on artificial intelligence, and the specific steps are as follows:

[0036] S1: Multimodal data collection and integration: By fully integrating HIS systems to extract structured data, deploying IoT devices to capture patients' physiological status, deeply integrating electronic medical records to mine key diagnosis and treatment information, and collecting hospital environment and patient flow data, all-round information aggregation lays a solid foundation for creating an accurate digital twin of hospital resources;

[0037] S2: Introducing unstructured data analysis: In emergency scenarios, we focus on medical staff voice recordings and use NLP technology to identify key information in emergency treatment communications. We deeply mine large amounts of emergency voice data to build a critical case prediction model, convert unstructured voice into quantitative indicators, predict resource needs, and effectively improve the timeliness and accuracy of emergency response.

[0038] S3: Construction and Application of Dynamic Demand Forecasting Model: Using the spatiotemporal graph neural network (ST-GNN), we construct a spatiotemporal graph of hospital department layout, patient flow, and time trends. Using historical data to train the model, we learn the demand patterns for patients, drugs, and equipment across departments over time, predicting details for the next 24 hours and improving operating room utilization efficiency.

[0039] S4: Modeling and Optimizing Cross-Departmental Resource Priority: Using the Q-learning algorithm, dynamic priority rules are established based on medical urgency and resource scarcity, prioritizing emergency care, critical care, and outpatient care. Cross-departmental sharing of high-value da Vinci robotic equipment is optimized, with usage time allocated based on surgical schedules and disease complexity. Ethical weights are also embedded, with waiting penalty factors set for elderly patients, balancing fairness and humanity.

[0040] S5: Real-time elastic scheduling engine deployment: Edge computing nodes are deployed in key areas of the hospital, updating multi-source data every 5 minutes. Based on real-time data, a medical staff skill matrix matching algorithm is run to dynamically allocate manpower based on skills and needs. A path planning algorithm is also used to plan routes for mobile medical equipment, achieving precise deployment and supporting real-time optimization of massive concurrent variables.

[0041] S6: Human-machine collaborative decision-making interface design and interaction: Build an AR visualization system to display key hospital information in 3D, such as operating room-ICU linkage heat maps to facilitate insights. AI-powered scheduling recommendations are generated based on data algorithms, game theory is introduced to take into account the director's experience, and when disagreements arise, the system interacts to find a balance. Managers can manually correct and provide feedback, optimize the model, and improve the effectiveness of human-machine collaborative decision-making.

[0042] S7: Closed-loop feedback optimization and blockchain evidence audit: Relying on the digital twin of hospital resources, simulate multi-scenario scheduling effects, use GAN to create an extreme data strong model, introduce blockchain, record the entire chain information of resource scheduling, and distribute epidemic prevention materials in a compliant and transparent manner in line with JCI standards, which not only ensures refined management, but also builds a solid data trust foundation for improving medical quality.

[0043] The specific steps of multimodal data collection and integration in S1 are as follows:

[0044] Step 1: Structured medical data extraction: Fully access the HIS system to accurately extract basic patient information, detailed diagnosis and treatment process records, and detailed cost details. This data is clearly organized and easy to analyze, building a basic medical process framework and providing key underlying information support for subsequent in-depth research on disease trends and resource consumption.

[0045] Step 2: Real-time patient physiological monitoring: Deploy IoT devices to closely monitor the patient's bed status, turning frequency, and subtle changes. Wearable devices can continuously collect heart rate, blood pressure, and blood oxygen saturation 24 hours a day, capturing the patient's physiological dynamics in real time and from all angles, laying a solid data foundation for precision medical decision-making.

[0046] Step 3: Integration of environmental and crowd flow data: The system collects environmental data from all parts of the hospital, including the temperature and humidity in wards, operating rooms, and waiting areas, which are related to patients' physical sensations and the stable operation of medical equipment. At the same time, crowd density monitoring accurately reflects the busyness of operations and provides early warning of the risk of crowd gathering.

[0047] First, structured medical data extraction is connected to the HIS system to obtain key information and build a basic framework to help analyze the condition and resource utilization; second, real-time patient physiological monitoring relies on IoT and wearable devices to accurately grasp the patient's physical dynamics and lay a solid foundation for precision medicine; finally, environmental and crowd data are integrated to collect temperature, humidity, crowd density, etc. to ensure patient physical sensation and equipment stability, and can also warn of risks and optimize hospital operations.

[0048] Among them, the introduction of unstructured data analysis in S2 refers to focusing on the detailed NLP processing of voice records in the emergency process. With the help of cutting-edge natural language processing technology, the key points in the communication between medical staff can be accurately captured, and the patient's symptom description keywords such as chest pain and difficulty breathing can be accurately extracted, as well as verbal expressions of the severity of the disease and the general direction of the preliminary diagnosis. Then, relying on big data thinking, massive emergency voice data is deeply analyzed, and a critical case prediction model based on semantic features is fully constructed. The inherent logical connection between different expression patterns and the diagnosis of critical illness is deeply explored, and the scattered and unstructured voice information is cleverly converted into quantitative data indicators with great predictive value.

[0049] Focusing on emergency voice records, we use advanced NLP technology to accurately extract key information such as patient symptoms, severity of illness, and preliminary diagnosis. Then, with big data thinking, we dig deep into massive voice data, build a critical case prediction model, and convert scattered voice into quantitative indicators, providing a strong basis for early prediction and precise treatment.

[0050] Among them, the construction and application of the dynamic demand prediction model in S3 refers to the use of the spatiotemporal graph neural network ST-GNN to integrate the hospital department layout, patient flow path and time trend, and construct a complex spatiotemporal graph structure, so that the model can learn the dynamic laws related to patients, drugs and equipment in each department at different time periods, laying the foundation for accurate prediction; training ST-GNN with rich historical data allows it to master the demand change pattern of each department, accurately predict the patient flow per hour in the next 24 hours, estimate drug consumption by category and dosage form, and calculate demand based on equipment usage and failure conditions, and output a detailed demand list; strictly verify the prediction results to ensure that the overall error rate is less than 8%, and apply the model to actual scenarios. Based on the prediction of operating room occupancy rate, the anesthesia team is deployed in advance, the surgical process is optimized, and the hospital resource allocation efficiency and department operation efficiency are effectively improved.

[0051] First, ST-GNN is used to construct a spatiotemporal graph to learn the dynamic patterns of departments and lay the foundation for accurate predictions. Second, based on historical data training models, it can accurately estimate patient flow, drug consumption, and equipment requirements for the next 24 hours, and output a detailed list to facilitate advance preparation. Third, strict verification ensures a low error rate, which can be applied in practice to allocate resources based on operating room predictions, improving hospital resource allocation and department operation efficiency.

[0052] The specific steps for modeling and optimizing cross-departmental resource priorities in S4 are as follows:

[0053] Step 1: Establish a dynamic priority system: Using the Q-learning algorithm, we fully consider the degree of medical urgency and resource scarcity, and clearly prioritize each department. The emergency department is ranked first due to its criticality and timeliness, followed by the intensive care unit, and the outpatient department is ranked relatively low. This ensures that critically ill patients have priority access to medical resources and ensures timely life-saving treatment.

[0054] Step 2: Build a high-value equipment sharing model: For the high-end da Vinci robotic device, we will break down departmental barriers and establish a sharing optimization model based on the patient's surgical schedule and the complexity of the condition. This model will comprehensively weigh the difficulty of the surgery and the expected efficacy, scientifically plan the equipment usage period, and promote the circulation of equipment throughout the hospital, thereby improving utilization efficiency, reducing idle rates, and meeting the needs of more patients.

[0055] Step 3: Embed ethical weights to ensure fairness: Innovatively integrate ethical considerations into the resource allocation model, and set a waiting time penalty factor specifically for vulnerable groups such as elderly patients. When resource allocation decisions are made, this factor will come into play to avoid excessive resource allocation, balance fairness and efficiency, and make medical resource allocation more in line with the concept of humanistic care.

[0056] First, a dynamic priority system is established through the Q-learning algorithm. The priorities of emergency, critical care, and outpatient departments are clarified based on urgency and scarcity to ensure the treatment of critically ill patients. Second, a high-value equipment sharing model is constructed to break down departmental boundaries. The use time of the da Vinci robot is planned according to the surgery and the condition of the disease to improve equipment utilization. Finally, ethical weights are embedded and a waiting penalty factor is set for elderly patients. This balances fairness and efficiency, makes resource allocation full of humanistic care, and improves the overall quality of medical services.

[0057] The real-time elastic scheduling engine deployment in S5 involves extensively deploying edge computing nodes in key hospital areas, including nurse stations, operating room dispatch centers, and logistics support departments. Leveraging their proximity to data sources and low latency, edge computing nodes rapidly collect multi-source data from HIS systems and IoT devices every five minutes, providing real-time insights into hospital operations and providing immediate basis for subsequent decision-making. Based on this real-time data, a medical staff skill matrix matching algorithm is activated to systematically analyze medical staff's intubation and emergency resuscitation expertise. Taking into account the urgency of patient needs and the workload of each department, the most suitable personnel are intelligently and dynamically assigned to the corresponding positions, ensuring efficient and high-quality medical services. For mobile medical equipment such as defibrillators and infusion pumps, a path planning algorithm is used to comprehensively consider the device's current location, the urgency of requests from various departments, and the congestion of hospital corridors. This allows for the optimal movement of the equipment, ensuring its precise deployment in the shortest possible time. This technically achieves real-time optimization of massive concurrent variables, ensuring timely patient care.

[0058] On the one hand, edge computing nodes are deployed in key areas to collect HIS and IoT multi-source data every 5 minutes, to monitor operational dynamics in real time and provide immediate basis for decision-making. On the other hand, a medical skills matching algorithm is activated based on real-time data, and personnel are accurately allocated based on department needs and busyness to ensure efficient and high-quality services. Furthermore, a path planning algorithm is used for mobile medical equipment to comprehensively consider location, urgency of demand, and channel congestion to plan the optimal route, optimize concurrent variables in real time, and help patients receive timely treatment.

[0059] The specific steps for designing and interacting with the human-machine collaborative decision-making interface in S6 are as follows:

[0060] Step 1: Build an AR visualization platform: Create an AR visualization system that converts the hospital's resource layout, including detailed information about the distribution of wards, operating rooms, and medical equipment storage locations, as well as real-time patient flow and pre-set scheduling plans, into intuitive 3D images. Using an operating room-ICU bed linkage heat map, managers can clearly see the connection between surgery and postoperative monitoring, accurately grasping key processes.

[0061] Step 2: Human-machine collaborative recommendation generation: The AI model uses complex algorithms based on massive amounts of medical data to quickly generate and push preliminary scheduling and planning recommendations. Game theory is also incorporated to fully respect the clinical director's intuitive judgment, formed through years of experience. When disagreements arise, both parties interact through a game-like game, weighing the pros and cons, working together to find the optimal balance and formulate scientific and practical decisions.

[0062] Step 3: Real-time feedback optimization model: After receiving the fusion solution, if managers find that there are still areas for improvement, they can manually correct them directly in the visual interface and provide real-time feedback to the AI model. The model receives new instructions, uses deep learning to correct the causes, and dynamically adjusts its own decision-making logic, continuously improving its ability to respond to complex medical scenarios and achieving continuous improvement in the effectiveness of human-machine collaborative decision-making.

[0063] First, an AR visualization platform is built to present all kinds of key information of the hospital in three dimensions. With the help of linked heat maps, managers can accurately grasp the connection process between surgery and monitoring, thereby improving management efficiency. Second, human-machine collaboration generates suggestions. The AI model combines massive data algorithms with the experience of clinical directors, introduces game theory to weigh the pros and cons, and formulates scientific and practical decisions. Finally, real-time feedback is provided to optimize the model. Managers can make corrections and provide feedback on the visualization interface, prompting AI to deeply learn and dynamically adjust, and continuously improving the ability of human-machine collaboration to cope with complex medical scenarios.

[0064] Among them, the closed-loop feedback optimization and blockchain evidence audit in S7 refer to the simulation and evaluation of scheduling strategies: with the help of the digital twin of hospital resources, different scheduling strategies are simulated for daily peaks and extreme scenarios such as large-scale traffic accidents and infectious disease outbreaks, and resource allocation, patient treatment, and cost-effectiveness indicators are comprehensively evaluated to understand the pros and cons of each strategy; strengthen the extreme adaptability of the model: use GAN to generate extreme scenario data, train the model to adapt to rare and critical situations, and output reliable scheduling plans even under extreme pressure to ensure the hospital's emergency response capabilities; blockchain ensures transparent decision-making: introduce blockchain to record information on the entire process of resource scheduling, strictly comply with the distribution of epidemic prevention materials, ensure fine management standards, and build a solid foundation of trust for improving medical quality.

[0065] First, simulate and evaluate scheduling strategies, use digital twins to simulate multiple scenarios, and comprehensively consider resources, treatment, and cost-effectiveness indicators to help select the optimal strategy; second, strengthen the extreme adaptability of the model, and use GAN to generate extreme data training models so that they can operate reliably under extreme pressure and improve emergency response capabilities; third, blockchain ensures transparent decision-making, records the entire scheduling process, and strictly follows standards for the distribution of epidemic prevention materials to ensure meticulous and standardized management, laying a solid foundation of trust for improving medical quality.

[0066] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0067] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent resource scheduling optimization method based on artificial intelligence, characterized by: The specific steps of the intelligent resource scheduling optimization method based on artificial intelligence are as follows: S1: Multimodal data collection and integration: By fully integrating HIS systems to extract structured data, deploying IoT devices to capture patients' physiological status, deeply integrating electronic medical records to mine key diagnosis and treatment information, collecting hospital environment and patient flow data, and comprehensively gathering information, we lay a solid foundation for creating an accurate digital twin of hospital resources. S2: Introducing unstructured data analysis: In emergency scenarios, we focus on medical staff voice recordings and use NLP technology to identify key information in emergency treatment communications. We deeply mine large amounts of emergency voice data to build a critical case prediction model, convert unstructured voice into quantitative indicators, predict resource needs, and effectively improve the timeliness and accuracy of emergency response. S3: Construction and Application of Dynamic Demand Forecasting Model: Using the spatiotemporal graph neural network (ST-GNN), we construct a spatiotemporal graph of hospital department layout, patient flow, and time trends. Using historical data to train the model, we learn the demand patterns for patients, drugs, and equipment across departments over time, predicting details for the next 24 hours and improving operating room utilization efficiency. S4: Modeling and Optimizing Cross-Departmental Resource Priority: Using the Q-learning algorithm, dynamic priority rules are established based on medical urgency and resource scarcity, prioritizing emergency care, critical care, and outpatient care. This system optimizes cross-departmental sharing of high-value da Vinci robotic devices, allocating time slots based on surgical schedules and medical complexity. Ethical weights are also embedded, with waiting penalty factors set for elderly patients, balancing fairness and humanity. S5: Real-time elastic scheduling engine deployment: Edge computing nodes are deployed in key areas of the hospital, updating multi-source data every 5 minutes. Based on real-time data, a medical staff skill matrix matching algorithm is run to dynamically allocate manpower based on skills and needs. A path planning algorithm is also used to plan routes for mobile medical equipment, achieving precise deployment and supporting real-time optimization of massive concurrent variables. S6: Human-machine collaborative decision-making interface design and interaction: Build an AR visualization system to display key hospital information in 3D, such as operating room-ICU linkage heat maps to facilitate insights. AI-powered scheduling recommendations are generated based on data algorithms, game theory is introduced to take into account the director's experience, and when disagreements arise, the system interacts to find a balance. Managers can manually correct and provide feedback, optimize the model, and improve the effectiveness of human-machine collaborative decision-making. S7: Closed-loop feedback optimization and blockchain evidence audit: Relying on the digital twin of hospital resources, simulate multi-scenario scheduling effects, use GAN to create an extreme data strong model, introduce blockchain, record the entire chain information of resource scheduling, and distribute epidemic prevention materials in a compliant and transparent manner in line with JCI standards, which not only ensures refined management, but also builds a solid data trust foundation for improving medical quality.

2. The method for intelligent resource scheduling optimization based on artificial intelligence according to claim 1, characterized in that: The specific steps of multimodal data collection and integration in S1 are as follows: Step 1: Structured medical data extraction: Fully access the HIS system to accurately extract basic patient information, detailed diagnosis and treatment process records, and detailed cost details. This data is clearly organized and easy to analyze, building a basic medical process framework and providing key underlying information support for subsequent in-depth research on disease trends and resource consumption. Step 2: Real-time patient physiological monitoring: Deploy IoT devices to closely monitor the patient's bed status, turning frequency, and subtle changes. Wearable devices can continuously collect heart rate, blood pressure, and blood oxygen saturation 24 hours a day, capturing the patient's physiological dynamics in real time and from all angles, laying a solid data foundation for precision medical decision-making. Step 3: Integration of environmental and crowd flow data: The system collects environmental data from all parts of the hospital, including the temperature and humidity in wards, operating rooms, and waiting areas, which are related to patients' physical sensations and the stable operation of medical equipment. At the same time, crowd density monitoring accurately reflects the busyness of operations and provides early warning of the risk of crowd gathering.

3. The method for intelligent resource scheduling optimization based on artificial intelligence according to claim 1, characterized in that: The introduction of unstructured data analysis in S2 refers to focusing on the detailed NLP processing of voice records during the emergency process. With the help of cutting-edge natural language processing technology, the key points in the communication between medical staff can be accurately captured, and keywords describing patients' symptoms such as chest pain and difficulty breathing can be accurately extracted, as well as verbal expressions of the severity of the disease and the general direction of the preliminary diagnosis. Then, relying on big data thinking, massive emergency voice data is deeply analyzed, and a critical case prediction model based on semantic features is fully constructed. The inherent logical connection between different expression patterns and the diagnosis of critical illness is deeply explored, and the scattered and unstructured voice information is cleverly converted into quantitative data indicators with great predictive value.

4. The method for intelligent resource scheduling optimization based on artificial intelligence according to claim 1, characterized in that: The construction and application of the dynamic demand forecasting model in S3 refers to the use of the spatiotemporal graph neural network ST-GNN to integrate the hospital department layout, patient flow paths and time trends, and construct a complex spatiotemporal graph structure, so that the model can learn the dynamic laws related to patients, drugs, and equipment in each department at different time periods, laying the foundation for accurate prediction; training ST-GNN with rich historical data allows it to grasp the demand change pattern of each department, accurately predict the patient flow per hour in the next 24 hours, estimate drug consumption by category and dosage form, and calculate demand based on equipment usage and failure conditions, and output a detailed demand list; strictly verify the prediction results to ensure that the overall error rate is less than 8%, and apply the model to actual scenarios. Based on the prediction of operating room occupancy rate, the anesthesia team is deployed in advance, the surgical process is optimized, and the hospital resource allocation efficiency and department operation efficiency are effectively improved.

5. The method for intelligent resource scheduling optimization based on artificial intelligence according to claim 1, characterized in that: The specific steps for modeling and optimizing cross-departmental resource priorities in S4 are as follows: Step 1: Establish a dynamic priority system: Using the Q-learning algorithm, we fully consider the degree of medical urgency and resource scarcity, and clearly prioritize each department. The emergency department is ranked first due to its criticality and timeliness, followed by the intensive care unit, and the outpatient department is ranked relatively low. This ensures that critically ill patients have priority access to medical resources and guarantees timely life-saving treatment. Step 2: Build a high-value equipment sharing model: For the high-end da Vinci robotic device, we will break down departmental barriers and establish a sharing optimization model based on the patient's surgical schedule and the complexity of the condition. This model will comprehensively weigh the difficulty of the surgery and the expected efficacy, scientifically plan the equipment usage period, and promote the circulation of equipment throughout the hospital, thereby improving utilization efficiency, reducing idle rates, and meeting the needs of more patients. Step 3: Embed ethical weights to ensure fairness: Innovatively integrate ethical considerations into the resource allocation model, and set a waiting time penalty factor specifically for vulnerable groups such as elderly patients. When resource allocation decisions are made, this factor will come into play to avoid excessive resource allocation, balance fairness and efficiency, and make medical resource allocation more in line with the concept of humanistic care.

6. The method for intelligent resource scheduling optimization based on artificial intelligence according to claim 1, characterized in that: The real-time elastic scheduling engine deployment in S5 involves extensively deploying edge computing nodes in key hospital areas, including nurse stations, operating room dispatch centers, and logistics support departments. Leveraging their proximity to data sources and low latency, edge computing nodes rapidly collect multi-source data from HIS systems and IoT devices every five minutes, providing real-time insights into hospital operations and providing immediate basis for subsequent decision-making. Based on this collected real-time data, a medical staff skill matrix matching algorithm is activated to systematically analyze medical staff's intubation and emergency resuscitation expertise. Furthermore, based on the urgency of patient needs and the workload of each department, the most suitable personnel are intelligently and dynamically assigned to the corresponding positions, ensuring efficient and high-quality medical services. For mobile medical equipment such as defibrillators and infusion pumps, a path planning algorithm is used to comprehensively consider the device's current location, the urgency of requests from various departments, and the congestion of hospital corridors. This allows for the optimal movement of the equipment, ensuring its precise and timely deployment. This technically achieves real-time optimization of massive concurrent variables, ensuring timely patient care.

7. The method for intelligent resource scheduling optimization based on artificial intelligence according to claim 1, characterized in that: The specific steps of designing and interacting the human-machine collaborative decision-making interface in S6 are as follows: Step 1: Build an AR visualization platform: Create an AR visualization system that converts the hospital's resource layout, including detailed information about the distribution of wards, operating rooms, and medical equipment storage locations, as well as real-time patient flow and pre-set scheduling plans, into intuitive 3D images. Using an operating room-ICU bed linkage heat map, managers can clearly see the connection between surgery and postoperative monitoring, accurately grasping key processes. Step 2: Human-machine collaborative recommendation generation: The AI model uses complex algorithms based on massive amounts of medical data to quickly generate and push preliminary scheduling and planning recommendations. Game theory is also incorporated to fully respect the clinical director's intuitive judgment, formed through years of experience. When disagreements arise, both parties interact through a game-like game, weighing the pros and cons, working together to find the optimal balance and formulate scientific and practical decisions. Step 3: Real-time feedback optimization model: After receiving the fusion solution, if managers find that there are still areas for improvement, they can manually correct them directly in the visual interface and provide real-time feedback to the AI model. The model receives new instructions, uses deep learning to correct the causes, and dynamically adjusts its own decision-making logic, continuously improving its ability to respond to complex medical scenarios and achieving continuous improvement in the effectiveness of human-machine collaborative decision-making.

8. The method for intelligent resource scheduling optimization based on artificial intelligence according to claim 1, characterized in that: The closed-loop feedback optimization and blockchain evidence audit in S7 refer to the simulation and evaluation of scheduling strategies: with the help of the digital twin of hospital resources, different scheduling strategies are simulated for daily peaks and extreme scenarios such as large-scale traffic accidents and infectious disease outbreaks, and resource allocation, patient treatment, and cost-effectiveness indicators are comprehensively evaluated to understand the pros and cons of each strategy; strengthening the extreme adaptability of the model: using GAN to generate extreme scenario data, training the model to adapt to rare and critical situations, and output reliable scheduling plans even under extreme pressure to ensure the hospital's emergency response capabilities; blockchain ensures transparent decision-making: introducing blockchain to record information on the entire process of resource scheduling, strictly complying with the distribution of epidemic prevention materials, ensuring fine management and standards, and building a solid foundation of trust for improving medical quality.

Images