A postoperative recovery room virtual bed automatic transfer method and system

Abstract

The application provides a postoperative recovery room virtual bed automatic transfer method and system, wherein the method comprises the following steps: acquiring postoperative recovery room bed state detection data, bed corresponding patient vital sign data, and vital sign information and medical information of a patient to be used in a bed; inputting the bed state detection data and the bed corresponding patient vital sign data into a pre-established digital twin model to obtain a predicted use state of each bed; determining bed demand data of the patient to be used in the bed based on the vital sign information of the patient to be used in the bed; performing priority sorting on the patient to be used in the bed based on the medical information of the patient to be used in the bed; and determining a bed scheduling instruction based on the priority sorting, the bed demand data, and the predicted use state of each bed, wherein the bed scheduling instruction is used for automatically transferring and controlling the postoperative recovery room beds. Through implementation of the application, the automation level of postoperative recovery room bed management can be improved.

Description

Technical Field

[0001] This invention belongs to the field of medical data processing technology, specifically relating to a method and system for automatic transfer of virtual beds in a postoperative recovery room. Background Technology

[0002] As modern medical systems evolve towards greater refinement and efficiency, the Post-Anesthesia Care Unit (PACU), serving as a crucial buffer between operating rooms and general wards, has become a key variable affecting the smoothness of hospital surgical scheduling, the rationality of medical resource allocation, and patients' postoperative recovery experience. During the golden recovery period after surgery, patients' demand for PACU beds exhibits significant characteristics of high concentration, time sensitivity, and frequent turnover. Delays in bed allocation can not only lead to postoperative bed congestion but also directly cause delays in subsequent surgical scheduling, idle and wasted medical resources, and in extreme cases, even trigger patient safety incidents, causing a chain reaction of disruptions to the overall hospital's medical order.

[0003] However, most hospitals still use traditional bed management models, which rely on manual scheduling and registration or static bed information records. These models lack the ability to adapt to the dynamic treatment environment of the PACU: they cannot handle the high-frequency data streams generated instantaneously in postoperative scenarios, nor can they cope with the dynamic priority adjustment needs in cross-scheduling tasks. They are gradually becoming unable to meet the higher requirements for medical service efficiency and quality in the context of smart hospital construction.

[0004] To address this predicament, some hospitals have begun to explore the introduction of "virtual beds" or "digital scheduling" systems, hoping to optimize bed status management and usage prioritization through information technology, thereby alleviating the pressure of traditional manual scheduling. However, in terms of practical application, existing systems still have significant limitations, and their core capabilities fall short of actual clinical needs. The systems are insufficient in capturing and adapting to dynamic changes in the treatment scenario, making it difficult to form accurate and stable scheduling criteria, leading to frequent deviations and repeated adjustments in scheduling arrangements. This disconnect between the system and clinical practice prevents the expected optimization effects from being fully realized and fails to fundamentally resolve the core challenges of PACU bed management. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a method and system for automatic transfer of virtual beds in the postoperative recovery room, so as to meet the need to improve the automation level of PACU bed management.

[0006] To achieve the above objectives, the present invention provides the following technical solution: According to a first aspect, the present invention provides a method for automatic transfer of virtual beds in a postoperative recovery room, comprising: acquiring postoperative recovery room bed status detection data, vital sign data of patients corresponding to the beds, and vital sign information and medical information of patients in beds to be used; inputting the bed status detection data and vital sign data of patients corresponding to the beds into a pre-established digital twin model to obtain the predicted usage status of each bed; determining the bed demand data of patients in beds to be used based on the vital sign information of patients in beds to be used; prioritizing patients in beds to be used based on the medical information of patients in beds to be used; and determining bed scheduling instructions based on the priority ranking, bed demand data, and predicted usage status of each bed, wherein the bed scheduling instructions are used to automatically transfer and control the beds in the postoperative recovery room.

[0007] According to a second aspect, the present invention provides an automatic virtual bed transfer system for a postoperative recovery room, comprising: multiple sensors for collecting postoperative recovery room bed status detection data, vital sign data of patients corresponding to the beds, and medical information of patients in beds to be used; a main control module for executing any one of the automatic virtual bed transfer methods for a postoperative recovery room as described in the first aspect to obtain bed scheduling instructions; and multiple movable beds for automatic transfer based on the bed scheduling instructions.

[0008] According to a third aspect, the present invention provides an automatic virtual bed transfer device for a postoperative recovery room, comprising: an information acquisition module for acquiring postoperative recovery room bed status detection data, vital sign data of patients corresponding to the beds, and vital sign information and medical information of patients in beds to be used; a usage status prediction module for inputting the bed status detection data and vital sign data of patients corresponding to the beds into a pre-established digital twin model to obtain the predicted usage status of each bed; a demand determination module for determining the bed demand data of patients in beds to be used based on the vital sign information of patients in beds to be used; a sorting module for prioritizing patients in beds to be used based on the medical information of patients in beds to be used; and a scheduling instruction generation module for determining bed scheduling instructions based on the priority sorting, bed demand data, and the predicted usage status of each bed, wherein the bed scheduling instructions are used for automatic transfer control of postoperative recovery room beds.

[0009] According to a fourth aspect, an embodiment of the present invention provides an electronic device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor performing the steps of the automatic transfer method and system for virtual beds in a postoperative recovery room as described in the first aspect or any embodiment of the first aspect.

[0010] According to a fifth aspect, embodiments of the present invention provide a computer storage medium storing computer instructions that, when executed by a processor, implement the steps of the automatic transfer method and system for virtual beds in a postoperative recovery room as described in the first aspect or any embodiment of the first aspect.

[0011] This embodiment provides a method and system for automatic virtual bed transfer in a postoperative recovery room. By simultaneously acquiring bed status, vital signs of in-bed patients, vital signs of patients awaiting admission, and medical information, it overcomes the limitations of traditional manual scheduling that relies on single pieces of information. Utilizing a digital twin model, it predicts future bed occupancy, shifting from passive response to proactive prediction, reducing bed vacancy or patient congestion. Then, based on the vital signs of patients awaiting admission, it determines bed demand data, ensuring precise matching of bed resources with patients' postoperative recovery needs. Furthermore, it prioritizes patients based on medical information, giving priority to urgent and high-risk patients, balancing scheduling efficiency and medical equity. Finally, it automatically generates scheduling instructions, replacing cumbersome manual processes and improving the automation level and scheduling efficiency of postoperative recovery room bed management.

[0012] Other advantages, objectives, and features of the invention will be set forth in the following description and will be apparent to those skilled in the art in some respects, or may be learned by practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0013] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the following figures are provided for illustration: Figure 1 This is a flowchart illustrating a specific example of an automatic virtual bed transfer method for postoperative recovery rooms according to the present invention. Figures 2-7 This is a diagram of the display interface of the target display device in this invention; Figure 8 This is a simulated block diagram of an automated virtual bed transfer system for a postoperative recovery room according to the present invention; Figure 9 This is a schematic diagram of a virtual module of an automatic virtual bed transfer device for a postoperative recovery room according to the present invention; Figure 10 This is a schematic block diagram of a specific example of an electronic device in an embodiment of the present invention. Detailed Implementation

[0014] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can also refer to the internal connection of two components; and they can refer to a wireless connection or a wired connection. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0016] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0017] This invention provides a method for automatic transfer of virtual beds in a postoperative recovery room, such as... Figure 1 As shown, it includes: S101, acquire postoperative recovery room bed status monitoring data, bed-related patient vital sign data, and vital sign and medical information of patients in beds to be used; S102, input the bed status detection data and the vital sign data of the patients corresponding to the beds into the pre-established digital twin model to obtain the predicted usage status of each bed; S103, Based on the vital signs information of patients waiting to use beds, determine the bed demand data of patients waiting to use beds; S104, Prioritize patients for beds based on their medical information; S105 determines bed scheduling instructions based on priority ranking, bed demand data, and the predicted usage status of each bed. These instructions are used to automatically transfer and control the beds in the postoperative recovery room.

[0018] For example, this method is applied to the main control module. To complete the above method, various peripheral devices are also required, such as multiple sensors and multiple movable beds. Postoperative recovery room bed status monitoring data, vital sign data of patients corresponding to each bed, and medical information of patients waiting to use beds can all be acquired by multiple sensors. Specifically: Bed status monitoring data can include bed status, patient data assigned to the bed, and medical equipment equipped at the bed. Postoperative recovery room bed status monitoring data can be acquired through embedded sensing devices or handheld devices, such as pressure-sensing pads or capacitive sensors deployed on the bed, to determine whether the bed is in a continuously pressure-free or continuously pressured state, thus determining whether the bed is vacant or occupied. It can also determine changes in bed status by comparing with the previous status, such as from occupied to vacant, at which point the status can be defined as awaiting cleaning. The handheld device can be a mobile phone or a dedicated device used to identify the patient assigned to the bed. Patient assignment can be identified using RFID tags, meaning that RFID tags are pre-installed on each bed.

[0019] Patients assigned to beds are defined as those already occupying beds in the postoperative recovery room, while patients with beds awaiting use are defined as those not currently occupied but requiring a bed in the postoperative recovery room. Patients with beds awaiting use can be pre-marked. Patient vital signs data include heart rate, blood pressure, body temperature, SpO2, etc., and can be obtained from existing hospital monitoring equipment, such as multi-parameter monitors. Medical information for patients with beds awaiting use can be obtained from the hospital's information management system, including surgical type, disease, etc. These existing monitoring devices can sample data at a frequency of at least 1 Hz, and use an edge filtering mechanism to eliminate interference signals. Finally, the data is uploaded to edge nodes through a dedicated gateway, such as a BLE / LoRa gateway.

[0020] Postoperative recovery room bed status monitoring data, bed-related patient vital sign data, and medical information of patients waiting to use beds are all preprocessed at the edge node, uniformly formatted into JSON format, and then uploaded to the main control module via HTTPS.

[0021] A digital twin model is pre-built in the main control module. The goal of this digital twin model is to achieve real-time mapping, predictive simulation, and interactive visualization of postoperative recovery room beds and patient entities. Specifically, a virtual bed twin is constructed based on data input from the acquisition module, establishing a one-to-one mapping relationship with the physical bed. Each twin model has multi-dimensional attributes: current state, historical trajectory, expected idle time, etc., supports real-time refreshing and visual interaction, and embeds a state evolution prediction model (such as time series neural network, Bayesian state transition model) to predict future fluctuations in bed resource supply and demand.

[0022] The digital twin model uses BIM / IFC standards or UML modeling language to construct the static attributes and dynamic status of each resuscitation bed. Each bed corresponds to a twin with the following attributes: real-time occupancy status; current patient number and estimated awakening time; cleanliness status and total usage time; and synchronized environmental variable values. The digital twin model includes a twin data synchronization controller that synchronizes its status with the data acquisition module periodically (every second). In the event of a sudden change in status (such as a patient's sudden abnormality or premature bed release), the twin status is immediately and forcibly refreshed. The digital twin model also includes a predictive simulation submodule. This module uses algorithms such as LSTM and random forest to predict a patient's estimated awakening time (±5 minutes error) based on historical data; it simulates the predicted usage status of each bed within the next 5 hours, such as whether it is idle or occupied during a certain time period. The bed twins and data synchronization provide highly consistent and scenario-based data support (including real-time and historical data) for predictive simulation, improving the accuracy of awakening time prediction and the realism of bed vacancy simulation, and serving not only a demonstrative purpose. Synchronized data serves as input features and training samples for the prediction model, generating personalized prediction results. The twin binds the prediction results to the physical resources of the beds, providing the scheduling module with a dual decision-making basis of "time + resources".

[0023] Bed demand data can be determined based on bed demand instructions issued when a patient is transferred out of the operating room. A bed demand instruction can be triggered upon transfer out of the operating room, and the current patient enters a bed-awaiting state. In this embodiment, for multiple demand instructions, the bed demand data can also be customized and matched based on the patient's individual characteristics. For example, bed demand data can include bed demand time and bed resource type. The resource type can include medical resources, such as medical equipment. The method for determining the bed demand data based on the vital sign information of the patient awaiting bed use can be to automatically match the corresponding parameter intervals in the vital sign information using a clinical threshold library pre-stored in the main control module, and output the patient's baseline tolerance time. In addition, the baseline tolerance time can be further adjusted according to the trend of vital signs, such as calculating the rate of change of certain parameters, and the tolerance time interval can be used as the demand time. Then, based on the patient's vital sign data, it can be analyzed which medical equipment is needed. This embodiment does not limit this process; those skilled in the art can determine it as needed.

[0024] Based on the medical information of patients awaiting bed occupancy, patients are prioritized for bed occupancy. One implementation method is to use a pre-established standardized scoring system that covers urgency, disease risk, recovery speed, and surgical type. Based on the scoring system and the medical information of patients awaiting bed occupancy, quantifiable scores, i.e., individual indicator scores, are obtained. Each dimension is assigned corresponding weights based on departmental attributes. Finally, the score for each patient is calculated according to the formula: comprehensive score = individual indicator score × indicator weight. Patients are then ranked from highest to lowest score.

[0025] After obtaining the priority ranking, bed demand data, and the predicted occupancy status of each bed, bed scheduling instructions can be derived based on these three factors. Specifically, genetic algorithms, reinforcement learning, and other methods can be used to obtain bed scheduling instructions, which include the correspondence between patients and beds, such as allocating a particular bed to a particular patient. Taking the use of a genetic algorithm to obtain bed scheduling instructions as an example, firstly, the patient-bed matching schemes are encoded as chromosomes, and a population containing candidate schemes is initialized. Then, a fitness function that integrates priority, time conflict, bed utilization rate, and bed fit is used for scoring. Time conflict can be determined based on the bed demand data and the predicted occupancy status of each bed, that is, whether the bed demand time and the predicted occupancy status match. For example, if a patient needs a bed before 12 noon tomorrow, the predicted occupancy status of each bed is checked to see if there are any vacant beds at the corresponding time. The fitness function aims to ensure that the scheme corresponding to the chromosome can maximize the satisfaction of priority ranking requirements and the matching degree between bed demand data and predicted bed occupancy status. The population is iteratively optimized through selection, crossover, and mutation until the termination condition is met; finally, the optimal chromosome is selected and converted into scheduling instructions corresponding to patients and beds. The genetic algorithm formula is an existing formula and will not be elaborated here.

[0026] This invention provides a method for automatically transferring virtual beds in a postoperative recovery room. By simultaneously acquiring bed status, vital signs of in-bed patients, vital signs of patients awaiting admission, and medical information, it overcomes the limitations of traditional manual scheduling that relies on single pieces of information. Utilizing a digital twin model, it predicts future bed occupancy, shifting from passive response to proactive prediction, reducing bed vacancies or patient congestion. Then, based on the vital signs of patients awaiting admission, it determines bed demand data, ensuring a precise match between bed resources and patients' postoperative recovery needs. Furthermore, it prioritizes patients based on medical information, giving priority to urgent and high-risk patients, balancing scheduling efficiency and medical equity. Finally, it automatically generates scheduling instructions, replacing cumbersome manual processes and improving the automation level and scheduling efficiency of postoperative recovery room bed management.

[0027] As an optional implementation, the digital twin model integrates a target prediction model. Bed status detection data and corresponding patient vital sign data are input into the pre-established digital twin model to predict the occupancy status of each bed, including: The vital signs data of each patient are input into the target prediction model to obtain the expected awakening time of each patient. Based on the expected awakening time of each patient and the corresponding bed status detection data, the predicted usage status of each bed is determined. The predicted usage status of each bed is represented by the idle / occupied time axis of the bed at future time.

[0028] For example, the target prediction model may include algorithms such as LSTM and random forest. Then, vital sign data are input into the prediction model trained and optimized based on historical medical data. The model calculates and outputs the estimated awakening time for each patient, with an error controlled within ±5 minutes; subsequently, the predicted awakening time is output. The target prediction model is trained on historical data, which contains a large amount of vital sign data and actual awakening times from past patients. Using vital sign data as features and actual awakening time as labels, the larger the data volume and the more patient types covered, the more accurate the model prediction.

[0029] This invention provides a method for automatic virtual bed transfer in a postoperative recovery room. It applies a digital twin model to predict bed status, and achieves accurate estimation of patient awakening time by integrating a target prediction model. The future occupancy status of the bed is visualized in the form of a timeline, which solves the pain points of information lag and inaccurate prediction in traditional bed status management.

[0030] As an optional implementation, patients using beds are prioritized based on their medical information, including: The system receives pre-scored results from experts based on the medical information of patients awaiting bed allocation, covering multiple target indicators including surgical type, recovery speed, urgency, and disease risk. Based on the scoring results, a subjective weight vector is obtained using the analytic hierarchy process (AHP). Then, based on the medical information, the system determines the quantitative values ​​for the current patient's surgical type, recovery speed, urgency, and disease risk. These values ​​are then input into a pre-built target model to determine the objective weights for each of the four indicators. The target model is constructed based on historical transfer data. Finally, based on each patient's quantitative values ​​for surgical type, recovery speed, urgency, and disease risk, along with the corresponding subjective and objective weight vectors, the system determines the priority ranking of patients for bed allocation.

[0031] For example, in this implementation example, four core indicators—"surgery type," "recovery speed," "urgency level," and "disease risk"—were pre-established. The analytic hierarchy process (AHP) was used, calibrated with clinical data. Specifically, firstly, using AHP, clinical experts were invited to perform pairwise comparisons and scoring of the indicators, forming a judgment matrix. After passing a consistency test, a subjective weight vector was obtained. Next, based on the hospital's massive historical transfer data, a multiple linear regression model was constructed, with the actual urgency of transfer (such as standardized waiting time) as the dependent variable Y and the quantified values ​​of the aforementioned indicators as the independent variable X. The obtained standardized regression coefficients are normalized and used as the objective data weight vector. Finally, the final weights are synthesized using the following formula: ; in, It is an adjustable parameter used to balance experience and data.

[0032] For each patient, their various indicator values ​​are standardized (e.g., "recovery speed" is mapped backward from the previously predicted estimated awakening time), and then substituted into the linear weighted summation formula: Priority score = Σ(standardized index value) ); Based on this, the patients were initially sorted.

[0033] This invention provides a method for automatically transferring virtual beds in a postoperative recovery room. By integrating expert clinical judgment with historical transfer data modeling, it achieves the quantification and standardization of patient prioritization. Overall, it avoids the subjectivity and randomness of traditional manual prioritization, making patient priority allocation more scientific and consistent. This ensures that emergency and high-risk patients can obtain bed resources first, and the prioritization logic can be dynamically optimized with the accumulation of historical data, adapting to long-term operational changes in the recovery room.

[0034] As an optional implementation method, the priority ranking of patients' bed usage is determined based on each patient's surgical type, recovery speed, urgency, and disease risk quantification values, as well as corresponding subjective and objective weight vectors, including: Based on the quantitative values ​​of each patient's surgical type, recovery speed, urgency, and disease risk, as well as the corresponding subjective and objective weight vectors, a preliminary priority ranking of patient bed usage is determined. Based on the predicted usage status of each bed and bed demand data, a conflict risk value for the preliminary priority ranking is determined. If the risk value exceeds the preset value, the ranking is re-ranked, and the conflict risk value is determined based on the re-ranking until the conflict risk value meets the requirements, thus obtaining the priority ranking of patient bed usage.

[0035] For example, the process of determining the initial priority ranking of patients' bed usage based on the quantitative values ​​of each patient's surgical type, recovery speed, urgency, disease risk, and corresponding subjective and objective weight vectors is as described in the above embodiment and will not be repeated here. The initial priority ranking only considers patient attributes. To ensure the executability of the scheduling plan, this embodiment introduces a forward-looking conflict risk assessment mechanism. Specifically, in the above embodiment, the predicted bed usage status has been obtained, and the bed demand data is known. Based on this, the risk is quantified. The indicators for quantifying the risk include the waiting index and the recovery window overlap rate. The waiting index can be determined by calculating the patient's expected waiting time / the average tolerable waiting time for this type of patient. This expected waiting time is dynamically estimated based on the current queue and bed prediction. The recovery window overlap rate can be determined by calculating the proportion of the overlap time between the bed demand data of patients waiting to use the bed and the idle state in the predicted bed usage status to the larger span of the two intervals. Combining the above two, the conflict risk value is calculated as follows: Conflict risk value = (1 - window overlap rate) × waiting index × 100; Ultimately, based on conflict risk values, the data is categorized into low-risk (<30), medium-risk (30-60), and high-risk (>60). For medium- and high-risk matching items, it is determined that they may cause scheduling bottlenecks or resource competition, automatically triggering targeted replanning. The evaluation strategy will be dynamically adjusted. Specifically, based on the original evaluation indicator system, the weights of indicators such as "conflict correction cost" and "resource utilization rate" are temporarily and significantly increased. Around high-risk nodes, multiple alternative paths are simulated through a digital twin model, including adjusting patient order, changing bed matching, and inserting cleaning buffer time. Finally, the optimized new sorting scheme that can effectively avoid high-risk conflicts is fed back, thereby achieving closed-loop correction of the initial queue and ultimately outputting an intelligent queue that has both high clinical priority and high execution feasibility.

[0036] This invention provides a method for automatically transferring virtual beds in a postoperative recovery room. By quantitatively assessing the risk of matching conflicts between beds and patient needs, the method adjusts the patient queue accordingly, solving the problem of unexecutable plans that may occur when simply prioritizing patients. This improves the feasibility of implementing scheduling instructions. The method proposed in this embodiment achieves optimal matching between patient queues and bed resources while ensuring the needs of high-priority patients, effectively reducing scheduling delays and resource waste caused by resource conflicts, and improving the stability of the overall scheduling plan.

[0037] As an optional implementation, before determining the bed scheduling instruction based on priority ranking, bed demand data, and the predicted occupancy status of each bed, the following steps are included: The algorithm determines whether the number of patients waiting for bed use exceeds the target number and whether a resource mutation has occurred. If the number exceeds the target number but no resource mutation occurs, a first objective algorithm is used based on priority ranking, bed demand data, and the predicted usage status of each bed to determine the bed scheduling instruction. If the number does not exceed the target number and no resource mutation occurs, a second objective algorithm is used based on priority ranking, bed demand data, and the predicted usage status of each bed to determine the bed scheduling instruction. If a resource mutation occurs, a third objective algorithm is used based on priority ranking, bed demand data, and the predicted usage status of each bed to determine the bed scheduling instruction.

[0038] For example, this embodiment differentiates the methods for determining scheduling instructions under different circumstances. When considering whether the number of patients waiting for bed use exceeds the target number and whether a resource mutation occurs, the target number can be 5 patients. If the number exceeds the target number but no resource mutation occurs, a first objective algorithm (reinforcement learning algorithm) is used. If the number does not exceed the target number and no resource mutation occurs, the problem complexity is low, and a second objective algorithm (simulated annealing algorithm, SA) is used to quickly obtain a high-quality solution with less computational resources, improving response speed. When a resource mutation occurs, such as a bed being temporarily disabled, a third objective algorithm (greedy algorithm, combining rules for local adjustments) is used to first generate a feasible solution, ensuring continuous system operation. In addition, this embodiment retains a human-computer interaction interface, allowing administrators to switch pre-configured specific rule algorithm packages under special circumstances to address the problem of all algorithms failing due to significant changes in clinical scenarios.

[0039] In this embodiment, the second objective algorithm can use simulated annealing to escape local optima and achieve globally optimal bed allocation based on patient priority ranking, patient bed demand, and predicted bed occupancy status, ultimately outputting scheduling instructions. First, based on patient priority ranking, patient bed demand, and predicted bed occupancy status, an initial solution for the Alternative Scheduling (SA) is generated as the starting point for SA iterations. For example, the initial solution could be a scheme such as ranking patients from highest to lowest priority; allocating beds to the highest priority patients first; not re-allocating already occupied beds, and sequentially matching all patients.

[0040] The design minimizes the cost function F, meaning the more unreasonable the solution, the larger the value of F; the more reasonable the solution, the smaller the value of F. F = w1×A + w2×B + w3×C Where A represents the lack of beds for high-priority patients; B represents the total predicted waiting time for all patients awaiting allocation; C represents the bed vacancy rate; and w1, w2, and w3 are weights, increasing sequentially. The goal is to prioritize ensuring beds for emergency patients, then reduce waiting times, and finally improve bed utilization.

[0041] Fine-tuning can be done based on the current plan. For example, two allocated patients' compliant beds can be swapped and their release times rematched, a patient can be replaced with another compliant pre-release bed with an earlier release date, or any of the newly added vacant pre-release beds can be used to replace unassigned high-priority patients.

[0042] After obtaining the adjusted plan, the value of the minimum cost function F of the new plan is calculated according to the above formula. The difference between this value and the minimum cost function F of the current plan is then calculated to obtain the cost difference. If the cost difference is less than 0, the new plan is considered an optimized plan, and the current plan is replaced. If the cost difference is greater than 0, the new plan is considered a degraded plan, and a new plan is generated. The temperature decay setting is then implemented according to the simulated annealing algorithm. For example, a general geometric cooling method can be used to gradually reduce the temperature and tighten the screening criteria. The iteration stops when the temperature reaches the termination temperature, and a formal bed scheduling instruction is generated.

[0043] This invention provides a method for automatically transferring virtual beds in a postoperative recovery room. It divides the scenario into three categories based on the number of patients and the status of resources and matches them with a dedicated algorithm. This method breaks through the efficiency bottleneck of a single algorithm in different scenarios, ensuring optimal resource allocation under normal conditions and coping with special situations such as fluctuations in the number of patients and sudden changes in resources. It improves the adaptability and robustness of the scheduling system and ensures that the management of postoperative recovery room beds can operate efficiently and stably under various operational conditions.

[0044] As an optional implementation, the first objective algorithm is a reinforcement learning algorithm. Based on priority ranking and the predicted occupancy status of each bed, the first objective algorithm determines the bed scheduling instructions, including: Priority ranking, bed demand data, and predicted occupancy status of each bed are input into a pre-trained reinforcement learning algorithm to obtain the highest-value action. The action represents the patient-bed matching combination. The reward function in the reinforcement learning algorithm is determined based on changes in bed turnover, changes in average patient waiting time, and scheduling conflict rate. Based on the highest-value action, a patient-bed allocation scheduling instruction is generated.

[0045] For example, the following content is predefined in a pre-trained reinforcement learning algorithm: State (S): is a composite feature vector, including: a list of patients with beds to be used and a list of available beds. The list of patients with beds to be used includes the priority and demand start time TP for each patient. The list of available beds includes the expected idle time TS, medical resource type, and expected available duration TE-TS for each bed.

[0046] Action (A): A discrete action, i.e., selecting a patient from the current queue. and assigned to a specific bed. This forms an allocation pair ( ).

[0047] Reward (R): Designed as a multi-objective weighted reward function: ; in, , , Indicates the corresponding weight. This indicates changes in bed turnover rate, which characterizes the frequency of bed utilization and release per unit time. It reflects resource flow efficiency, with positive values ​​indicating faster turnover and increased rewards. This indicates the change in average patient waiting time, used to characterize the change in the average time from when a patient enters the allocation queue to when they are actually allocated a bed. The shorter the waiting time, the higher the reward. ConflictRate represents the scheduling conflict rate after this decision. It is used to characterize the proportion of the total number of allocations that conflict with patient needs or resource constraints in the scheduling scheme. ConflictRate∈[0,1], and the lower the value, the better. , , The values ​​can be 0.4, 0.3, and 0.3, respectively. The weighting coefficients are derived from the hospital's historical operational data through multi-objective optimization and calibration to balance efficiency, fairness, and stability.

[0048] A deep Q-network (DQN) is trained offline using historical transport data to learn how to maximize long-term cumulative rewards. During online operation, the engine receives the system state S(t) in real time, and the trained policy network outputs the action A with the highest Q value, which is the optimal allocation scheme.

[0049] This invention provides a method for automatic virtual bed transfer in a postoperative recovery room. The method uses a reinforcement learning algorithm to generate scheduling instructions and uses a reward function to coordinate key indicators such as bed turnover, patient waiting time, and conflict rate, thereby achieving global optimization of the scheduling scheme. At the same time, the algorithm has online learning capabilities and can continuously optimize the scheduling strategy as clinical data accumulates, maintaining adaptability to changes in the recovery room.

[0050] As an optional implementation method, an automatic transfer method for virtual beds in a postoperative recovery room further includes: when the scheduling conflict rate corresponding to the patient-bed allocation scheduling instruction obtained based on the reinforcement learning algorithm meets the preset requirements, the priority ranking and the predicted usage status of each bed are input into a pre-constructed genetic algorithm to obtain the patient-bed allocation scheduling instruction.

[0051] For example, when the real-time scheduling conflict rate (i.e. the proportion of resource contention in actual or simulated allocation) continuously exceeds a threshold (e.g., 20%), it indicates that the current reinforcement learning strategy may be trapped in a local optimum or has not adapted to the new pattern. The main control module engine automatically switches to the genetic algorithm (GA) and uses its global search capability to find the optimal solution again.

[0052] This invention provides an automatic virtual bed transfer method for postoperative recovery rooms. When the conflict rate of the reinforcement learning output scheme exceeds the standard, it automatically switches to genetic algorithm optimization to ensure that the scheduling scheme always maintains a low conflict rate and high executability.

[0053] As an optional implementation, bed scheduling instructions are determined based on priority ranking and the predicted occupancy status of each bed, including: Based on priority ranking, bed demand data, and predicted occupancy status of each bed, multiple preliminary bed scheduling instructions are determined. For each preliminary bed scheduling instruction, the corresponding bed transfer process is simulated using a pre-built digital twin model to obtain target parameters during the simulation. Based on the target parameters and preset evaluation rules, multiple indicator evaluation values ​​are obtained. Based on the multiple indicator evaluation values, the evaluation score of the preliminary bed scheduling instruction is determined. The preliminary bed scheduling instruction with the highest evaluation score is selected as the bed scheduling instruction.

[0054] For example, in this embodiment, when multiple preliminary bed scheduling instructions are generated, a digital twin module is used to simulate the complete execution path of each preliminary bed scheduling instruction. This includes: transfer sequence simulation: simulating the order in which different patients are transferred; physical route planning: based on the hospital's electronic map, planning specific routes for each transfer task and assessing intersection conflicts, elevator usage conflicts, etc.; resource occupancy simulation: calculating the required transfer personnel, equipment, and their timelines; and timeline extrapolation: accurately extrapolating the entire process of "occupancy-cleaning-idleness-reoccupancy" for each bed. During the simulation, target parameters are obtained, and based on the target parameters and pre-set evaluation rules, multiple indicator evaluation values ​​are obtained, specifically including: Resource utilization rate (30 points): Score = 30 (Total actual bed occupancy time during the simulation period / (Total number of beds)) Scheduling cycle length) directly measures the utilization efficiency of the core resource of hospital beds.

[0055] Execution efficiency (25 points): Score = 25 (Standard baseline time / simulated total scheduling time). The standard baseline time is derived from historical data statistics, and this metric measures process smoothness.

[0056] Vacancy rate (20 points): Score = 20 (1 - Total bed vacancy time during the simulation period / (Total number of beds)) (Schedule duration). Specifically targeting the core pain point of "zero beds waiting for surgery" in the resuscitation room, minimizing ineffective idle time.

[0057] Conflict Correction Cost (15 points): Score = 15 (1 - number of conflict events predicted in the simulation) (Average cost of handling a single conflict / estimated total operating costs). This is a deduction item, preemptively penalizing "tight" solutions that seem efficient but have low fault tolerance and are prone to causing on-site chaos and additional costs.

[0058] Clinical fit (10 points): A veto criterion. The fit is assessed against a pre-defined clinical rule set (e.g., "Critically ill patients must be assigned to beds equipped with full-function monitors," "Transfer distance must not exceed XX meters"). Complete compliance earns 10 points; failure to meet any key rule results in 0 points, and the proposed solution will typically be rejected.

[0059] Finally, assuming the clinical suitability score is not zero, calculate the total score for each candidate protocol: =Σ(Scores for each item).

[0060] The scheme with the highest total score is selected as the final execution scheme. When high scores are close, the scheme with a higher "clinical suitability" score and a higher "conflict correction cost" score is given priority, reflecting the principle of "safety and robustness first". Finally, all simulation details (sequence, path, timeline) of the optimal scheme will be encapsulated into structured scheduling instructions.

[0061] This invention provides a method for automatic virtual bed transfer in a postoperative recovery room. By generating multiple preliminary scheduling instructions and simulating their execution, and combining standardized evaluation rules to select the optimal solution, the method achieves pre-performance verification and selection of the scheduling scheme. This avoids potential problems such as transfer congestion and time coordination issues that may occur during actual execution, improves the efficiency of automatic bed transfer, reduces conflicts, and ensures that the final selected scheme is the comprehensive optimal solution.

[0062] As an optional implementation method, a virtual bed transfer method for postoperative recovery room further includes: sending bed status detection data, predicted usage status, and patient bed demand data and priority sorting to a target display device.

[0063] For example, the target display device can be a display terminal for medical staff, providing a web / touchscreen interface, supporting the display of bed status detection data, predicted usage status, and patient bed demand data and priority sorting, such as... Figures 2-7As shown, the system includes all displayed information and display methods, with each bed presented as a status map; it allows doctors to select beds to obtain patient recovery progress curves; and it dynamically updates bed "heat maps" to help identify short-term congestion areas. In addition, medical staff can "pause scheduling," "change beds," and "set surgical time slots"; all intervention actions are recorded and incorporated into the system's learning samples. This embodiment achieves real-time sharing and visualization of scheduling-related information through display devices, improving the collaborative efficiency of multi-role medical staff in the resuscitation room, avoiding collaborative errors caused by information asymmetry, and facilitating rapid manual intervention by medical staff in special scenarios due to information transparency.

[0064] This embodiment provides an automated virtual bed transfer system for postoperative recovery rooms, such as... Figure 8 As shown, it includes: Multiple sensors 101 are used to collect postoperative recovery room bed status detection data, vital sign data of patients corresponding to the beds, and medical information of patients in beds to be used; The main control module 102 is used to execute an automatic transfer method for virtual beds in the postoperative recovery room as described in the above embodiment, and to obtain bed scheduling instructions; Multiple movable beds 103 are used for automatic transfer based on bed scheduling commands. The movable beds can be AGV bed vehicles, rail transport beds, programmable mobile robots, lifting rail transfer units, etc.

[0065] As an alternative implementation, a virtual bed transfer system for postoperative recovery rooms, such as... Figure 7 As shown, it also includes: Display device 104 is used to display bed status detection data, predicted usage status, and patient bed demand data and priority ranking.

[0066] This embodiment provides an automatic virtual bed transfer device for postoperative recovery rooms, such as... Figure 9 As shown, it includes: The information acquisition module 201 is used to acquire postoperative recovery room bed status monitoring data, vital sign data of patients corresponding to the beds, vital sign information and medical information of patients in beds to be used; The status prediction module 202 is used to input bed status detection data and vital sign data of patients corresponding to the beds into a pre-established digital twin model to obtain the predicted usage status of each bed. The demand determination module 203 is used to determine the bed demand data of patients in need of use based on their vital sign information. The sorting module 204 is used to prioritize patients for beds based on their medical information. The scheduling instruction generation module 205 is used to determine bed scheduling instructions based on priority ranking, bed demand data, and the predicted usage status of each bed. The bed scheduling instructions are used to automatically transfer and control the beds in the postoperative recovery room.

[0067] This application also provides an electronic device, such as... Figure 10 As shown, processor 501 and memory 502 are connected via a bus or other means.

[0068] Processor 501 can be a central processing unit (CPU). Processor 501 can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or combinations of the above types of chips.

[0069] The memory 502, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as the program instructions / modules corresponding to the automatic transfer method for virtual beds in a postoperative recovery room in this embodiment of the invention. The processor executes various functional applications and data processing by running the non-transitory software programs, instructions, and modules stored in the memory.

[0070] Memory 502 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created by the processor, etc. Furthermore, the memory may include high-speed random access memory and non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory 502 may optionally include memory remotely located relative to the processor, which can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0071] The one or more modules are stored in the memory 502, and when executed by the processor 501, they perform actions such as... Figure 1 The embodiment shown illustrates an automatic virtual bed transfer method and system for postoperative recovery rooms.

[0072] For specific details regarding the aforementioned electronic devices, please refer to the relevant documentation. Figure 1 The relevant descriptions and effects in the illustrated embodiments are for understanding purposes only and will not be repeated here.

[0073] This embodiment also provides a computer storage medium storing computer-executable instructions that can execute an automatic virtual bed transfer method for postoperative recovery rooms in any of the above-described method embodiments. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), etc.; the storage medium may also include combinations of the above types of memory.

[0074] Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail through the above preferred embodiments, those skilled in the art should understand that various changes can be made to it in form and detail without departing from the scope defined by the claims of the present invention.

Claims

1. A method for automatic transfer of virtual beds in a postoperative recovery room, characterized in that, include: Acquire postoperative recovery room bed status monitoring data, vital sign data of patients corresponding to the beds, and vital sign and medical information of patients in beds to be used; By inputting bed status detection data and vital sign data of patients corresponding to each bed into a pre-established digital twin model, the predicted usage status of each bed can be obtained. Based on the vital signs information of patients waiting for beds to be used, determine the bed demand data of patients waiting for beds to be used; Based on the medical information of patients waiting to use beds, patients are prioritized for beds. Based on priority ranking, bed demand data, and the predicted usage status of each bed, bed scheduling instructions are determined. These instructions are used to automatically transfer beds in the postoperative recovery room.

2. The method for automatic transfer of virtual beds in a postoperative recovery room according to claim 1, characterized in that, The digital twin model integrates a target prediction model, inputting bed status detection data and corresponding patient vital sign data into a pre-established digital twin model to predict the predicted usage status of each bed, including: The vital signs data of each patient are input into the target prediction model to obtain the expected awakening time of each patient. Based on the expected awakening time of each patient and the corresponding bed status monitoring data, the predicted usage status of each bed is determined, and the predicted usage status of each bed is represented by the vacancy / occupancy time axis of the bed at future time.

3. The method for automatic transfer of virtual beds in a postoperative recovery room according to claim 1, characterized in that, Based on the medical information of patients awaiting bed occupancy, patients are prioritized for bed occupancy, including: The system receives pre-scoring results from experts based on the medical information of patients awaiting bed allocation, covering multiple target indicators, including surgical type, recovery speed, urgency level, and disease risk. Based on the scoring results, the subjective weight vector is obtained using the analytic hierarchy process (AHP). Based on medical information, determine the quantitative values ​​of the current patient's surgical type, recovery speed, urgency level, and disease risk. The quantitative values ​​of the patient's surgical type, recovery speed, urgency, and disease risk are input into the pre-built target model to determine the objective weights of the four categories of indicators: surgical type, recovery speed, urgency, and disease risk. The target model is built based on historical transfer data. Based on the quantitative values ​​of each patient's surgical type, recovery speed, urgency, and disease risk, as well as the corresponding subjective and objective weight vectors, the priority ranking of patients' bed usage is determined.

4. The method for automatic transfer of virtual beds in a postoperative recovery room according to claim 3, characterized in that, Based on each patient's quantified values ​​for surgical type, recovery speed, urgency, and disease risk, along with corresponding subjective and objective weight vectors, the priority ranking of patients for bed allocation is determined, including: Based on the quantitative values ​​of each patient's surgical type, recovery speed, urgency, disease risk, and corresponding subjective and objective weight vectors, a preliminary priority ranking of patients' bed usage is determined. Based on the predicted occupancy status and bed demand data for each bed, a preliminary priority ranking conflict risk value is determined. If the risk value exceeds the preset value, the order is reordered. Based on the reordering, the conflict risk value is determined until the conflict risk value meets the requirements, and the priority order of patients' bed usage is obtained.

5. The method for automatic transfer of virtual beds in a postoperative recovery room according to claim 1, characterized in that, Based on priority ranking, bed demand data, and the predicted occupancy status of each bed, the process before determining bed scheduling instructions includes: Determine whether the number of patients waiting for bed use exceeds the target number and whether a resource mutation has occurred; When the target number is exceeded and no resource mutation occurs, the first target algorithm is used to determine the bed scheduling instruction based on priority ranking, bed demand data and the predicted usage status of each bed. When the target number is not exceeded and no resource mutation occurs, the second target algorithm is used to determine the bed scheduling instruction based on priority ranking, bed demand data and the predicted usage status of each bed. When resource mutations occur, a third objective algorithm is used to determine bed scheduling instructions based on priority ranking, bed demand data, and the predicted usage status of each bed.

6. The method for automatic transfer of virtual beds in a postoperative recovery room according to claim 5, characterized in that, The first objective algorithm is a reinforcement learning algorithm. Based on priority ranking, bed demand data, and the predicted occupancy status of each bed, it determines bed scheduling instructions, including: Priority ranking, bed demand data, and the predicted usage status of each bed are input into a pre-trained reinforcement learning algorithm to obtain the most valuable action. The action represents the patient-bed matching combination. The reward function in the reinforcement learning algorithm is determined based on changes in bed turnover, changes in average patient waiting time, and scheduling conflict rate. Generate patient-bed allocation and scheduling instructions based on the most valuable actions.

7. The method for automatic transfer of virtual beds in a postoperative recovery room according to claim 6, characterized in that, Also includes: When the scheduling conflict rate corresponding to the patient-bed allocation scheduling instruction obtained based on the reinforcement learning algorithm meets the preset requirements, the priority ranking and the predicted usage status of each bed are input into the pre-built genetic algorithm to obtain the patient-bed allocation scheduling instruction.

8. The method for automatic virtual bed transfer in a postoperative recovery room according to claim 1, characterized in that, Based on priority ranking, bed demand data, and the predicted occupancy status of each bed, bed scheduling instructions are determined, including: Based on priority ranking, bed demand data, and the predicted usage status of each bed, several preliminary bed scheduling instructions are determined. For each initial bed scheduling instruction, the corresponding bed transfer process is simulated based on a pre-built digital twin model to obtain the target parameters in the simulation process; Based on the target parameters and preset evaluation rules, multiple indicator evaluation values ​​are obtained; The evaluation score for the preliminary bed scheduling instruction was determined based on multiple indicator evaluation values. Select the initial bed scheduling instruction with the highest evaluation score as the bed scheduling instruction.

9. A method for automatic transfer of virtual beds in a postoperative recovery room according to any one of claims 1-8, characterized in that, Also includes: The system sends bed status detection data, predicted usage status, patient bed demand data, and priority sorting to the target display device.

10. A virtual bed transfer system for a postoperative recovery room, characterized in that, include: Multiple sensors are used to collect data on the status of postoperative recovery room beds, vital signs of patients corresponding to the beds, and medical information of patients in beds waiting to be used. The main control module is used to execute the automatic transfer method of virtual bed in postoperative recovery room as described in any one of claims 1-9, and obtain bed scheduling instructions; Multiple movable beds are available for automatic transfer based on bed scheduling instructions.

Images