Hospital bed dynamic prediction and intelligent allocation system based on big data

Abstract

The application relates to the field of information and communication technology specially used for management or supervision purposes, and particularly relates to a hospital bed dynamic prediction and intelligent allocation system based on big data, which comprises the following steps: acquiring bed big data of all departments in a hospital; determining the acquisition performance of the bed big data according to a conventional treatment process and an unconventional treatment process; analyzing the data reference of the bed big data according to the acquisition performance; expanding a patient behavior operation data set according to the data reference, and constructing a time series prediction model according to the expanded patient behavior operation data set; predicting the bed occupation time length of all beds currently used based on the time series prediction model, obtaining the bed occupation residual time length, and allocating beds to patients to be treated according to the bed occupation residual time length. The scheme in the application can improve the operation efficiency of the hospital.

Description

Technical Field

[0001] This invention relates to the field of information and communication technology, specifically for management or supervision purposes, and more specifically to a big data-based dynamic prediction and intelligent allocation system for hospital beds. Background Technology

[0002] Current hospital bed allocation is mostly based on a static allocation model on a departmental basis. However, bed demand varies significantly across departments. Departments like respiratory medicine and pediatrics experience large seasonal fluctuations in demand, orthopedics has an average length of stay of 5.8 days, and internal medicine typically has longer stays. Surgical departments rely on phased bed allocation, while internal medicine focuses more on nursing duration. The complexity of a patient's condition affects bed occupancy time; for example, critically ill patients require longer occupancy. Interdepartmental collaboration influences bed turnover; for instance, the ICU (Intensive Care Unit) needs to receive critically ill patients from various departments.

[0003] Therefore, the existing method of allocating beds based on departments has the problem of low hospital operational efficiency. Summary of the Invention

[0004] To address the issue of improving hospital operational efficiency, the present invention aims to provide a hospital bed dynamic prediction and intelligent allocation system based on big data. The specific technical solution adopted is as follows: In one embodiment of this application, a hospital bed dynamic prediction and intelligent allocation system based on big data is provided, the system comprising: The data acquisition unit is used to acquire big data on the number of beds in all departments of the hospital. The first analysis unit is used to determine the acquireability of the bed big data based on routine treatment processes and unconventional treatment processes, wherein the acquireability is used to indicate the degree to which the bed control phase is affected by the unconventional treatment processes; The second analysis unit is used to analyze the data reference value of the bed big data based on the acquisition performance, wherein the data reference value is used to indicate the credibility of the bed big data in predicting the actual bed turnover demand. The model building unit is used to expand the patient behavior operation dataset based on the data reference and to build a time series prediction model based on the expanded patient behavior operation dataset. The prediction unit is used to predict the bed occupancy time of all currently used beds based on the time-series prediction model, obtain the remaining bed occupancy time, and allocate beds to patients to be treated according to the remaining bed occupancy time.

[0005] The acquisition of big data on bed availability across all departments within the hospital includes: Vital sign data of patients corresponding to beds in each of the aforementioned departments are collected using pre-set bed monitoring equipment. By using pre-set bed sensors, the occupancy status data of beds in each department is collected, and each bed operation and location change is timestamped to obtain bed operation records and timestamp information; The vital signs data, the occupancy status data, the bed operation records, and the timestamp information are aggregated into the bed big data.

[0006] The determination of the acquireability of the bed big data based on routine and unconventional treatment processes includes: Obtain all current operation records for a single bed, including all occupancy control behaviors for that single bed, such as admission procedures, inpatient treatment, surgery, and discharge. Based on all the operation records, the occupancy control behavior with stage characteristics is taken as a bed control stage. All the bed control stages are numbered in the order of occurrence to obtain a stage vector. Construct a multidimensional vector space by taking the stage type of each element of the stage vector as a single dimension; Using N as the initial clustering number, clustering operations are performed on all stage vectors in the multidimensional vector space to obtain k clusters. When k>N, the processing procedure corresponding to the largest cluster is taken as the department's routine treatment procedure, and the processing procedures corresponding to the other clusters besides the largest cluster are taken as the department's non-routine treatment procedures. The acquisition performance of the bed big data is obtained by analyzing the stage vectors of the conventional treatment process and the unconventional treatment process.

[0007] The analysis of the stage vectors in the conventional and unconventional treatment processes to obtain the appreciatory characteristics of the bed big data includes: Obtain the number of stage vectors in the target treatment process that include the i-th bed control stage, and the total number of stage vectors in the target treatment process, wherein the target treatment process is the conventional treatment process or the unconventional treatment process; Based on the ratio of the number of stage vectors to the total number of stage vectors, determine the maximum percentage performance of the i-th bed control stage under different clusters; The maximum value of the percentage is taken as the acquisition performance of the i-th bed control stage.

[0008] The step of analyzing the data reference value of the bed big data based on the acquisition performance includes: Obtain the bed control stage sequence corresponding to a single bed, wherein the bed control stage sequence is composed of the numbers of each bed control stage in chronological order; Obtain the first patient operation stage node value and the second patient operation stage node value of the i-th bed control stage in the bed control stage sequence, wherein the second patient operation stage node value is the patient operation stage node value corresponding to the time point before the first patient operation stage node value, and the patient operation stage node value is a value that quantitatively represents the patient operation progress at different time points within the bed control stage. Based on the absolute value of the difference between the node values ​​of the first patient operation stage and the node values ​​of the second patient operation stage, normalization processing is performed to obtain the significance of the operation amplitude of the patient operation in the i-th bed control stage. The significance of the operation amplitude is used to indicate the degree of change of the patient operation at adjacent time points. The data reference value of the bed big data is determined based on the significance of the operational amplitude and the acquisition performance.

[0009] The determination of the data reference value of the bed big data based on the significance of the operational amplitude and the acquisition performance includes: Obtain the acquisitive performance, operational significance, and mean vector of the bed control stage sequence for the i-th bed control stage. Determine the product of the acquired performance and the significance of the operational amplitude, and then determine the cosine similarity between the bed control stage sequence and the mean vector; The ratio of the product to the cosine similarity is normalized to obtain the data reference value of the bed big data.

[0010] The expansion of the patient behavior operation dataset based on the data reference includes: For the patient behavior operation data in the i-th bed control stage, obtain the stage node value and data reference value corresponding to any two adjacent operation data; By comparing the stage node values ​​corresponding to the two adjacent operation data, the stage node value difference is obtained; By comparing the data reference values ​​corresponding to the two adjacent operation data, the data reference value difference is obtained; Based on the difference between the stage node values ​​and the difference in data reference, the number of data points that need to be interpolated is obtained; Based on the number of data points to be interpolated, the two adjacent operation data are uniformly interpolated to generate supplementary operation data, thus obtaining the first expanded dataset. The first expanded dataset is divided into segments to obtain multiple segments, and spline interpolation is performed on the multiple segments to obtain a second expanded dataset, thereby expanding the patient behavior operation dataset.

[0011] The step of constructing a time-series prediction model based on the expanded patient behavior dataset includes: Use the expanded patient behavior dataset as the model input data; The model input data is trained using a time-series prediction algorithm. During training, the goal is to minimize the occupancy time of a single bed control phase sequence. Prediction weight parameters are constructed, and the time-series prediction model is obtained by adjusting the prediction weight parameters.

[0012] The step of predicting the occupancy duration of all currently used beds based on the time-series prediction model includes: Obtain real-time data for all currently used beds, including current bed occupancy status, current patient treatment stage, and duration of occupancy. The expanded patient behavior operation dataset and the real-time data are input into the time series prediction model so that the time series prediction model can output the estimated total occupancy time of each currently used bed based on the characteristics of the expanded patient behavior operation dataset and the real-time data. Based on the estimated total occupancy time and the occupancy time of each currently used bed, the remaining occupancy time of the bed is determined to complete the prediction of the occupancy time of all currently used beds.

[0013] The step of allocating beds to patients based on the remaining occupancy time includes: Obtain the medical needs information of patients awaiting treatment in each department, including the urgency of the patient's condition, appointment time, and expected length of hospital stay; The priority of bed allocation for the patients to be treated is determined based on the urgency of their condition and the appointment time. The remaining time of bed occupancy is matched with the expected length of hospital stay of the patient awaiting treatment, and a bed is allocated to the patient according to the bed allocation priority.

[0014] The present invention has the following beneficial effects: First, the data acquisition unit acquires big data on bed availability across all departments within the hospital. Then, the first analysis unit determines the acquireability of the big data based on routine and non-routine treatment processes, whereby the acquireability indicates the degree to which non-routine treatment processes affect bed allocation. Next, the second analysis unit analyzes the data reference value of the big data based on the acquireability, whereby the data reference value indicates the reliability of the big data in predicting actual bed turnover demand. Then, the model building unit expands the patient behavior and operation dataset based on the data reference value and constructs a time-series prediction model based on the expanded dataset. Finally, the prediction unit predicts the remaining occupancy time for all currently used beds based on the time-series prediction model, and allocates beds to patients awaiting treatment based on the remaining occupancy time. In this invention, acquiring comprehensive bed-related big data provides rich information for subsequent analysis, facilitating an accurate understanding of hospital bed occupancy. Determining acquisition patterns based on routine and non-routine treatment processes allows for a deeper understanding of the impact of special circumstances on bed control, providing a basis for optimized management. The reliance on analytical data enhances the credibility of bed turnover demand predictions, making them more aligned with reality. Expanding the patient behavior dataset and constructing a time-series prediction model significantly improves the accuracy of bed occupancy duration predictions. Allocating beds to patients based on predicted remaining occupancy duration effectively increases bed turnover rate, achieves rational allocation of medical resources, improves hospital operational efficiency, reduces patient waiting time, enhances the patient experience, and ensures timely access to medical services. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of a hospital bed dynamic prediction and intelligent allocation system based on big data, provided in one embodiment of the present invention. Figure 2 This is a flowchart illustrating a method for dynamic prediction and intelligent allocation of hospital beds based on big data, provided as an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a big data-based dynamic prediction and intelligent allocation system for hospital beds proposed according to the present invention. In the following description, different embodiments or one embodiment do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0020] The following description, in conjunction with the accompanying drawings, details a specific solution for a hospital bed dynamic prediction and intelligent allocation system based on big data provided by the present invention.

[0021] In one embodiment of this application, a hospital bed dynamic prediction and intelligent allocation system based on big data is provided. Figure 1 This is a schematic diagram of a hospital bed dynamic prediction and intelligent allocation system based on big data, provided in one embodiment of the present invention. (See attached diagram.) Figure 1 The hospital bed dynamic prediction and intelligent allocation system based on big data includes the following units as described in 110 to 150: Data acquisition unit 110 is used to acquire big data on the number of beds in all departments of the hospital.

[0022] Among them, bed big data refers to various data sets related to hospital bed usage, including but not limited to patients' vital signs data, bed occupancy status data, bed operation records, and timestamp information. These data can be collected and aggregated through bedside monitoring systems, smart bed sensors, etc., to provide basic data for subsequent analysis and prediction.

[0023] The first analysis unit 120 is used to determine the acquireability of the bed big data based on routine treatment processes and unconventional treatment processes, wherein the acquireability is used to indicate the degree to which the bed control phase is affected by the unconventional treatment processes.

[0024] In this context, "routine treatment process" refers to the common and standard treatment procedures and methods used for a specific disease or symptom. In this embodiment, the routine treatment process can be the processing procedure corresponding to the largest cluster determined through cluster analysis, representing a common and universal treatment process in the department.

[0025] Among them, unconventional treatment process refers to a special treatment process or situation that is different from conventional treatment. In this embodiment, it can be the processing process corresponding to the other clusters besides the largest cluster, reflecting the patient's treatment performance under special circumstances.

[0026] Among them, the acquisition performance refers to the presentation of the acquired data in terms of completeness, authenticity, and timeliness. It can be used to indicate the degree to which unconventional treatment processes affect the bed control stage. It can be determined by analyzing the data completeness of each occupancy control behavior and the interference of non-treatment factors.

[0027] Among them, the bed control stage refers to a series of stages related to the use of beds by patients, from the preparation to receive patients to the readiness to be ready again after the patients leave. The occupancy control behavior with stage characteristics, such as admission procedures, inpatient treatment, surgery, and discharge, can be regarded as a bed control stage.

[0028] The second analysis unit 130 is used to analyze the data reference value of the bed big data based on the acquisition performance, wherein the data reference value is used to indicate the credibility of the bed big data in predicting the actual bed turnover demand.

[0029] Data referenceability refers to the availability and reliability of data for a specific analytical or predictive objective. In this embodiment, data referenceability is used to indicate the reliability of big data on bed availability in predicting actual bed turnover demand. Bed turnover demand refers to the actual need for beds to be used in different departments and among different patients in a hospital. This demand can be predicted through the analysis of big data on bed availability.

[0030] The model building unit 140 is used to expand the patient behavior operation dataset based on the data reference and to build a time series prediction model based on the expanded patient behavior operation dataset.

[0031] Among them, the patient behavior operation dataset is a collection of data recording various behaviors and operations of patients during their use of beds. Expanding the patient behavior operation dataset based on data reference can improve prediction accuracy.

[0032] In this context, expansion involves adding new data or information to the existing data to make it richer and more comprehensive. In this embodiment, the patient behavior operation dataset can be expanded by performing interpolation and other processing on the original operation data.

[0033] Among them, the time series prediction model is a model built based on time series data to predict future data change trends. In this embodiment, the expanded patient behavior operation dataset can be used to construct a time series prediction model using a time series prediction algorithm to predict bed occupancy time.

[0034] The prediction unit 150 is used to predict the bed occupancy time of all currently used beds based on the time-series prediction model, obtain the remaining bed occupancy time, and allocate beds to patients to be treated according to the remaining bed occupancy time.

[0035] Bed occupancy duration prediction estimates the length of time a patient will use the bed. Remaining bed occupancy duration refers to the estimated time a bed will remain occupied after it has already been used for some time. Bed allocation assigns beds to suitable patients according to certain rules.

[0036] For example, in a general hospital, the data acquisition unit collects vital signs data such as ECG and blood oxygen saturation from patients through bedside monitoring systems installed in various wards. Simultaneously, smart bed sensors collect data on bed occupancy status, summarizing these data to form a large bed-related big data dataset. The first analysis unit uses cluster analysis of a large number of bed operation records to determine that treatment procedures for most common cold patients are routine, while complex treatment procedures for some rare disease patients are unconventional, thus clarifying the acquisition characteristics. The second analysis unit analyzes the reliability of this data in predicting bed turnover demand. The model building unit expands the patient behavior operation dataset based on data reference value, such as interpolating between two bed operation data points. After constructing the time-series prediction model, the prediction unit uses this model combined with real-time bed data to predict bed occupancy duration, obtains the remaining occupancy time, and then allocates beds based on information such as the urgency of the patient's condition.

[0037] As can be seen from the above implementation methods, the solution in this application, by acquiring comprehensive bed-related big data, can provide rich information for subsequent analysis, which is conducive to accurately grasping the hospital bed utilization situation; based on the acquisition performance determined by routine and non-routine treatment processes, it can gain a deeper understanding of the degree of influence of special circumstances on the bed control stage, providing a basis for optimizing management; the reference value of the analyzed data can improve the credibility of the prediction of bed turnover demand, making the prediction more in line with the actual situation; expanding the patient behavior operation dataset and constructing a time-series prediction model greatly improves the accuracy of bed occupancy time prediction; and allocating beds to patients based on the predicted remaining bed occupancy time can effectively improve bed turnover rate, achieve rational allocation of medical resources, thereby improving hospital operational efficiency, while reducing patient waiting time for beds, improving patient medical experience, and ensuring that patients receive timely medical services.

[0038] In some implementations, obtaining the big data on bed availability across all departments within the hospital includes: Vital sign data of patients corresponding to beds in each of the aforementioned departments are collected using pre-set bed monitoring equipment. By using pre-set bed sensors, the occupancy status data of beds in each department is collected, and each bed operation and location change is timestamped to obtain bed operation records and timestamp information; The vital signs data, the occupancy status data, the bed operation records, and the timestamp information are aggregated into the bed big data.

[0039] Among them, bedside monitoring equipment is a medical device used to monitor a patient's physical condition, capable of detecting and recording various physiological indicators in real time. In this embodiment, the bedside monitoring equipment is pre-installed on the hospital bed to specifically collect vital sign data of patients corresponding to beds in various departments, providing basic information about the patient's health status for subsequent analysis. Vital sign data refers to various indicators reflecting the basic life state of the human body, such as body temperature, blood pressure, heart rate, and respiration.

[0040] Among them, the bed sensor is a device installed on the hospital bed to sense changes in the bed's status. It is responsible for collecting occupancy status data for beds in various departments. Occupancy status data refers to whether a bed is currently occupied and related occupancy information, which is collected by the bed sensor and directly reflects the bed's usage status.

[0041] Bed-related operations refer to various actions related to the use of hospital beds, such as patient admission, discharge, and bed transfer. Each bed-related operation needs to be recorded for subsequent comprehensive analysis of bed utilization. Location changes indicate changes in the spatial position of the bed, which may be due to patient transfers to other departments or ward adjustments. Timestamp recording is a method of recording events by specifying the exact time, down to the precise point in time. Each bed-related operation and location change is timestamped to facilitate subsequent analysis of the sequence and time intervals of events, providing temporal data support for bed utilization analysis.

[0042] For example, in various wards of a hospital, each bed is equipped with bed monitoring equipment, such as a multi-parameter monitor, to collect vital signs data such as heart rate and blood oxygen in real time. Simultaneously, pressure sensors installed at the bottom of the beds act as bed sensors, detecting whether the bed is occupied in real time and collecting occupancy status data. When bed operations such as patient admission, discharge, or transfer occur, or when the bed's location changes due to ward adjustments, the system records the event with a timestamp. Finally, by aggregating this data from different sources, the hospital's bed big data is formed, which is used for subsequent dynamic bed allocation prediction and intelligent allocation analysis.

[0043] In this embodiment, by setting up bed monitoring equipment to collect vital sign data, data support can be provided for bed usage analysis from the perspective of patient health. For example, patients with serious conditions may need to occupy beds for a long time, which can help predict bed demand. By using bed sensors to collect occupancy status data, the real-time bed usage can be intuitively understood, facilitating timely adjustment of bed allocation. Time-stamping the bed operation and location change records can clearly present the dynamic process of bed usage, which helps to analyze the bed turnover pattern. The aggregation of various data forms bed big data, providing a comprehensive and detailed data foundation for the entire big data-based hospital bed dynamic prediction and intelligent allocation system, making subsequent analysis, prediction and allocation more accurate and scientific, effectively improving the efficiency of hospital bed management, optimizing resource allocation, and indirectly improving the patient's medical experience.

[0044] In some implementations, determining the acquireability of the bed-related big data based on routine and unconventional treatment processes includes: Obtain all current operation records for a single bed, including all occupancy control behaviors for that single bed, such as admission procedures, inpatient treatment, surgery, and discharge. Based on all the operation records, the occupancy control behavior with stage characteristics is taken as a bed control stage. All the bed control stages are numbered in the order of occurrence to obtain a stage vector. Construct a multidimensional vector space by taking the stage type of each element of the stage vector as a single dimension; Using N as the initial clustering number, clustering operations are performed on all stage vectors in the multidimensional vector space to obtain k clusters. When k>N, the processing procedure corresponding to the largest cluster is taken as the department's routine treatment procedure, and the processing procedures corresponding to the other clusters besides the largest cluster are taken as the department's non-routine treatment procedures. The acquisition performance of the bed big data is obtained by analyzing the stage vectors of the conventional treatment process and the unconventional treatment process.

[0045] Since hospital bed allocation is limited by the static departmental bed allocation, and inter-departmental bed transfers are also restricted by the actual floor zoning of hospital departments and wards, the main improvement in bed utilization efficiency should focus on enhancing the accuracy of predictions based on the existing bed allocation model to increase bed turnover efficiency and reduce patient waiting time. Analysis of big data on hospital beds reveals that the factors influencing the reference value of bed data across different departments are primarily the impact of the treatment process of the diseases handled by the department on bed occupancy and the impact of departmental management stages on bed operation. The degree of adjustability varies among these factors, and their reference impact on the needs of different departments also differs. Therefore, further analysis of the changes in hospital bed big data, and the detailed changes under the influence of bed turnover in different departments, is needed to achieve accurate dynamic predictions of hospital bed availability.

[0046] Occupation control behavior refers to the operational behaviors related to a series of processes from the availability of beds to their occupation, use, and eventual release. These behaviors can include admission procedures, inpatient treatment, surgery, and discharge, and directly affect the occupancy status of beds.

[0047] The control phase refers to the different stages that a bed goes through throughout its entire usage cycle, based on different operational behaviors.

[0048] Among them, the stage characteristics refer to the unique features or iconic operations of each control stage, which can be used to distinguish different control stages. In this embodiment, different behaviors such as admission procedures and surgery can be used as characteristics to distinguish different stages.

[0049] The bed allocation control phase is a specific stage defined by the phased nature of occupancy control behavior. For example, the admission procedure can be considered a bed allocation control phase, and so on, the bed usage process can be divided into phases.

[0050] The stage vector is a vector formed by numbering all bed control stages in chronological order, recording the sequence of bed control stages for convenient subsequent analysis. The element stage type represents the stage category for each element in the stage vector, such as admission treatment stage type, discharge stage type, etc. A single dimension refers to a direction or the measurement direction of a variable; each element stage type is used as a single dimension to construct a multidimensional vector space. A multidimensional vector space is a space composed of multiple single dimensions, enabling the description and analysis of data in multiple directions.

[0051] The initial cluster size is the number of clusters preset at the start of cluster analysis, with N as the initial value, providing the starting condition for the clustering operation. Clustering is a data analysis method that divides data in a dataset into different categories or clusters based on similarity. A cluster is a group of different categories or groups obtained after the clustering operation. k clusters are obtained through the clustering operation, and different clusters represent different treatment processes. The largest cluster is the cluster containing the most data.

[0052] The treatment process refers to the series of operations and treatment procedures a patient undergoes from admission to discharge. Data integrity refers to whether the data is complete and free of missing values. In this embodiment, the data integrity of each occupancy control behavior during routine and routine treatment processes is analyzed to determine the acquireability of big data on bed usage. Non-treatment factor interference refers to the impact of factors other than the patient's disease treatment itself on the bed usage process, such as processing time and bed cleaning time.

[0053] Before dividing the routine treatment process into routine and non-routine treatment processes, it is necessary to perform empty cluster detection and removal on the clustering results: traverse all clusters, select the valid clusters that contain a number of stage vectors greater than 0, recount the number of valid clusters, and retain the stage vector set of each valid cluster.

[0054] For example, in the internal medicine department of a hospital, the occupancy control behaviors for a single bed include patient admission procedures, post-admission treatment, possible surgeries, and final discharge. These behaviors constitute different bed control stages, which are sequentially numbered to obtain stage vectors. A multi-dimensional vector space is constructed using each stage type, such as admission procedures or surgery, as a single dimension. The initial cluster size N is set to 3. Clustering operations are performed on these stage vectors, resulting in 5 clusters (k=5>N). The largest cluster, containing the most data, corresponds to the processing procedures, such as the treatment process for most common internal medicine patients, and is considered a routine treatment process. The remaining clusters, such as the complex treatment processes for some rare disease patients, are considered unconventional treatment processes. Then, the completeness of the data for each occupancy control behavior in these processes is analyzed, including whether there are missing data at certain stages and interference from non-treatment factors, such as delays in discharge procedures. This yields the achievable performance of the bed big data.

[0055] In this embodiment, by analyzing the occupancy control behavior of individual beds, dividing bed control stages and constructing a multi-dimensional vector space, and then performing cluster analysis to determine routine and non-routine treatment processes, and further analyzing data integrity and interference from non-treatment factors, a deeper understanding of the accessibility of bed big data in different treatment processes can be achieved. This helps to identify problems in bed usage data and the impact of non-treatment factors on bed turnover, providing more accurate data support for subsequent big data-based dynamic bed prediction and intelligent allocation, improving the precision of hospital bed management, optimizing resource allocation, increasing bed utilization efficiency, and indirectly improving the patient's medical experience.

[0056] In some implementations, the analysis of stage vectors in the conventional treatment process and the unconventional treatment process to obtain the apprehension of the bed big data includes: Obtain the number of stage vectors in the target treatment process that include the i-th bed control stage, and the total number of stage vectors in the target treatment process, wherein the target treatment process is the conventional treatment process or the unconventional treatment process; Based on the ratio of the number of stage vectors to the total number of stage vectors, determine the maximum percentage performance of the i-th bed control stage under different clusters; The maximum value of the percentage is taken as the acquisition performance of the i-th bed control stage.

[0057] The number of stage vectors refers to the number of stage vectors that contain the i-th bed control stage in a specific cluster (e.g., the k-th cluster). In this embodiment, the frequency of a certain bed control stage in a specific cluster can be understood by counting this number.

[0058] The total number of stage vectors refers to the sum of all stage vectors contained in the k-th cluster. It represents the total amount of bed control stage-related data covered by the cluster and is used for subsequent calculation of the proportion.

[0059] The maximum percentage is determined by calculating the ratio of the number of stage vectors of the i-th bed control stage in different clusters to the total number of stage vectors in the corresponding clusters, and then finding the maximum value among these ratios. This reflects the relative prominence of the i-th bed control stage in each cluster.

[0060] Among them, the accessibility performance is used to intuitively indicate the strength of the accessibility performance of the i-th bed control stage in the bed big data. The larger the coefficient value, the less obvious the process-related influence is during the daily use of the bed control stage, and the higher the data accessibility.

[0061] For example, since the acquired bed-related big data only pertains to bed usage, it is a comprehensive dataset resulting from patient usage and departmental management processes. However, it doesn't fully represent the detailed bed usage process. For instance, after treatment, patients need to wait for discharge procedures to be completed and the bed cleaned before the bed occupancy status can be updated. But no data is acquired to update the status for these stages. Therefore, the acquired bed-related big data inherently includes the processing phases within the corresponding timeframes, and the changes at these stages are due to management processes rather than the patient's treatment process. Thus, it's necessary to define the acquireability of bed-related big data by examining the performance of data collected from multiple stages.

[0062] The following steps can be used to obtain the accessibility of big data on hospital beds: Retrieve all current operation records for a single bed, including all occupancy control behaviors; classify a single occupancy control behavior with distinct stage characteristics as a bed control stage, such as known bed control stages like admission procedures, inpatient treatment, surgery, and discharge; number all stages and record them as a stage vector according to their order of occurrence. The stage vector can be represented as: in, Let be the stage vector for the i-th bed control stage; The specific stage type code for bed control corresponds to known bed control stages such as admission procedures, inpatient treatment, surgery, and discharge.

[0063] For all bed control processes, the possible types of occupancy control are limited, and control behaviors tend to exhibit a clustered distribution based on the types of influence on the patient's treatment process. Therefore, the vector at each stage... Using the element stage type as a single dimension, a multidimensional vector space is constructed; using N as the initial cluster size, clustering operations are performed to obtain k clusters. Then, when... In this case, the largest cluster can be considered as the routine treatment process of the current department, and the remaining clusters can be considered as the patient's performance under unconventional treatment processes.

[0064] Then the first The acquisition performance during the individual bed control phase can be expressed as: in, This represents the acquisition performance during the i-th bed control phase; Indicates the first The bed control phase in the first The number of stage vectors in each cluster; Indicates the first The total number of stage vectors contained in each cluster.

[0065] in, This represents the maximum percentage of the i-th bed control stage under different clusters. The larger this value is, the less obvious the process-related impact of this bed control stage is in daily use, and the greater the value of its acquisition performance.

[0066] It should be noted that, in order to ensure that the calculation results are meaningful, when performing fractional operations, if the denominator is 0, a parameter adjustment factor greater than 0 needs to be added to the denominator to prevent the denominator from being 0. The value of the parameter adjustment factor is set by the implementer according to the actual situation, and is set to 0.1 in this application.

[0067] In this embodiment, by acquiring the number of stage vectors, the total number of stage vectors, and calculating the maximum percentage performance, the acquisition performance of each bed control stage under different clusters can be quantified. This quantification method makes the analysis of bed big data more accurate, helping hospitals to clearly understand the interference and integrity of data from non-treatment factors in each bed control stage. Based on this, hospitals can specifically improve data collection methods or optimize bed management processes to improve the quality of bed big data, providing strong support for more accurate dynamic prediction and intelligent allocation of beds, thereby improving hospital operational efficiency, rationally allocating medical resources, and ultimately improving the patient's medical experience.

[0068] In some implementations, analyzing the data reference value of the bed big data based on the acquisition performance includes: Obtain the bed control stage sequence corresponding to a single bed, wherein the bed control stage sequence is composed of the numbers of each bed control stage in chronological order; Obtain the first patient operation stage node value and the second patient operation stage node value of the i-th bed control stage in the bed control stage sequence, wherein the second patient operation stage node value is the patient operation stage node value corresponding to the time point before the first patient operation stage node value, and the patient operation stage node value is a value that quantitatively represents the patient operation progress at different time points within the bed control stage. Based on the absolute value of the difference between the node values ​​of the first patient operation stage and the node values ​​of the second patient operation stage, normalization processing is performed to obtain the significance of the operation amplitude of the patient operation in the i-th bed control stage. The significance of the operation amplitude is used to indicate the degree of change of the patient operation at adjacent time points. The data reference value of the bed big data is determined based on the significance of the operational amplitude and the acquisition performance.

[0069] The bed control phase sequence refers to the sequence formed by numbering each bed control phase according to the order in which they occur. It records the order of each control phase during the process of a bed's use from start to end, and is an important basis for analyzing the bed usage process and patient operation.

[0070] The first patient operation stage node value refers to the stage node value of the patient's operation at a specific time point within the i-th bed control stage. It represents the specific operational status of the patient within the bed control stage at that moment. The second patient operation stage node value refers to the patient operation stage node value corresponding to the time point within the i-th bed control stage that precedes the first patient operation stage node value. By comparing it with the first patient operation stage node value, it is used to analyze the changes in the patient's operation over time.

[0071] The absolute value of the difference refers to the absolute value of the subtraction between the node value of the first patient's operation stage and the node value of the second patient's operation stage. It eliminates the positive or negative influence of the subtraction result of the two values ​​and only focuses on the size of the difference between the two, so as to measure the change of the patient's operation at adjacent time points.

[0072] Among them, the significance of the operation amplitude is a value obtained by normalizing the absolute value of the difference between the node values ​​of the first and second patient operation stages. This value is used to intuitively indicate the degree of change of the patient's operation at adjacent time points. The larger the value, the more obvious the degree of change.

[0073] Among them, the degree of variation in patient operations describes how the patient's operations at adjacent time points change during the bed control phase. It is quantified by the significance of the operation amplitude. The degree of variation reflects the stability and regularity of the patient's operations, thus affecting the data reference value of bed big data.

[0074] The acquisition of node values ​​during patient operation stages relies on the linkage data between the hospital's medical and nursing operation management system and the intelligent bed sensing terminal: when medical staff complete key actions such as admission procedures, treatment operations, and discharge preparations, they need to enter operation nodes in the system; at the same time, the intelligent bed sensor will collect the bed usage status synchronously, and the system will automatically generate node values ​​based on the operation entry information and bed status data according to the preset linear assignment rules.

[0075] For example, in the obstetrics and gynecology department of a hospital, the bed control stages for a single bed include admission procedures, prenatal care, delivery, postpartum care, and discharge, sequentially numbered 1, 2, 3, 4, and 5, forming a bed control stage sequence. In the i-th bed control stage of prenatal care, assuming that at time t, the node value of the first patient's operation stage is 5, and at time t-1, the node value of the second patient's operation stage is 3, then the absolute value of the difference between the two is |5-3|=2. Normalizing this 2 yields the significance of the patient's operation amplitude in this bed control stage, indicating that within the prenatal care stage, there is a certain degree of variation in patient operations at adjacent time points. By analyzing the significance of the operation amplitude in each bed control stage, the data reference value of the large bed control dataset can be determined. It should be noted that the times t and t-1 here are the timing nodes that trigger the operation event, rather than equal intervals of physical time. The time difference between adjacent nodes can cover several minutes to several hours (e.g., t-1 is the drug injection operation node in the middle of prenatal care, and t is the fetal heart monitoring completion node in the end of prenatal care). Therefore, it can span different operation states within the same control phase.

[0076] Among them, the operation sequence node refers to the key operation trigger time point recorded in the hospital medical and nursing operation management system that is strongly related to the bed control stage, including but not limited to the admission procedure submission node, treatment operation start node, surgery end node, discharge procedure completion node, etc.; its correspondence with physical time is determined by the actual execution time of the medical operation, and there is no fixed threshold for the physical time interval between adjacent sequence nodes, which is used to characterize the sequential logic and progress changes of patient operations.

[0077] In this embodiment, by acquiring the bed control phase sequence and analyzing the node values ​​of the first and second patient operation stages in each bed control phase, the absolute value of the difference is calculated and normalized to obtain the significance of the operation amplitude, thereby determining the data reference value of the bed big data. This method can deeply analyze the credibility of bed big data in predicting actual bed turnover demand from the perspective of patient operation changes. By quantifying the degree of patient operation changes, it helps hospitals more accurately assess data quality, identify data that may be unstable or have special circumstances, thereby optimizing the bed prediction model, improving the rationality and efficiency of bed allocation, further enhancing the hospital's operational management level, and providing patients with higher quality and more efficient medical services.

[0078] In some implementations, determining the data reference value of the bed big data based on the significance of the operational amplitude and the accessibility performance includes: Obtain the acquisitive performance, operational significance, and mean vector of the bed control stage sequence for the i-th bed control stage. Determine the product of the acquired performance and the significance of the operational amplitude, and then determine the cosine similarity between the bed control stage sequence and the mean vector; The ratio of the product to the cosine similarity is normalized to obtain the data reference value of the bed big data.

[0079] For example, the differences in the reference value of actual big data due to the differences in demand between different departments mainly refer to the similarity analysis of the bed turnover process. This means analyzing the influence of factors other than the patient's condition itself, primarily the impact of each stage of the treatment process on bed turnover. Patient condition factors mainly refer to the strengths of key specialties or characteristic departments in hospitals with clear advantages, such as cardiology and oncology. These departments possess high levels of medical technology and brand recognition, employ advanced management models such as Enhanced Recovery After Surgery (ERAS), and have sufficient medical staff during the bed control phase, thus meeting patient needs, shortening the average length of stay, and increasing bed turnover. In contrast, departments treating patients with complex conditions often admit elderly patients, patients with multiple co-existing diseases, or critically ill patients, resulting in longer treatment cycles, including longer appointment times for examinations during the large bed control phase, and extended hospital stays. Therefore, it is necessary to further analyze the predictive reference value of bed data for actual turnover demand from the perspective of the overall departmental turnover performance.

[0080] The reliability of large-scale bed usage data can be determined through the following steps: Obtain the bed control phase sequence for a single bed First, the category code of the bed control stage can be mapped to a numerical code of [1, 10]. Then, the sequence of different lengths can be aligned to the same length through the dynamic time warping algorithm to obtain a numerical vector with uniform dimension. Obtain the value of the first patient operation stage node in the i-th bed control stage sequence. Second patient operation stage node value ; Based on the node value of the first patient operation stage Second patient operation stage node value The absolute value of the difference is normalized to obtain the significance of the patient's operation range in the i-th bed control stage.

[0081] For each bed control phase, the lower the availability of information, the greater the likelihood that the patient procedures at that phase are influenced by uncontrollable factors. Correspondingly, analyzing the operational variations, the more unique the differences in patient treatment within a single procedure, the lower its reliability, meaning the management behavior at that bed control phase is less reliable. The differences in patient treatment within a single procedure are represented by the significance of the operational amplitude compared to the previous procedure. The greater the change in the operational phase, the stronger the significance of the operational amplitude. The significance of the operational amplitude can be expressed as follows: in, Indicates the significance of the patient's actions during the i-th bed control phase; This represents the node value of the first patient's operation stage, which is the [number]th [stage]. The time during the individual bed control phase is The node values ​​of the patient operation stage; This represents the node value of the second patient's operation stage, which is the first... The time during the individual bed control phase is The node values ​​of the patient operation stage; This represents the absolute value of the difference between the node values ​​of the first patient's operation stage and the node values ​​of the second patient's operation stage. This indicates normalization processing.

[0082] It should be noted that, unless otherwise specified, the normalization methods in the embodiments of the present invention all adopt the minimum-maximum normalization function, which will not be further elaborated here.

[0083] Based on the significance and accessibility of the operational magnitude, data reference value is determined. Data reference value can be represented in the following ways: in, Indicates the first The first phase of bed control The data reference for this operation has a value range of [0, 1]. This represents the acquisition performance of the i-th bed control stage, which has been normalized to [0, 1]. This indicates the significance of the patient's operation in the i-th bed control stage, and has been normalized to [0, 1]. Indicates the first The sequence of bed control phases for each bed control phase; Indicates the first The mean vector of the bed control stage sequence for each bed control stage can be calculated by using the dynamic normalization algorithm to perform equal length processing for change path sequences of different lengths. This represents the cosine similarity between the stage sequence and the mean vector; This indicates normalization processing.

[0084] This involves collecting the bed control phase sequence of all beds in the department, calculating the arithmetic mean of each position according to the vector dimension, and obtaining the mean vector. The value ranges from [0, 1] to indicate the degree of fit between the control phase sequence of the target bed and the department's baseline process sequence. A higher value indicates that the bed's usage process is more in line with the department's routine mode.

[0085] Among them, when A result of 0 indicates that the usage process of the target bed completely deviates from the department's routine. For example, if a surgical bed receives a patient with complex injuries requiring multi-departmental resuscitation, the parameter adjustment factor of 0.05 needs to be added to the denominator, resulting in a denominator of 0 + 0.05 = 0.05 and a numerator of 0.05. Because the process is unconventional, Typically below 0.3, the final calculated The value will be lower than 0.2, corresponding to a low-reference assessment result, which is completely in line with clinical reality.

[0086] In this embodiment, the data reference value of the bed control big data is determined by comprehensively considering the accessibility performance, operational amplitude significance, and the mean vector of the bed control stage sequence in the i-th bed control stage. First, the ratio of the product to the cosine similarity is calculated, and then normalization is performed. This method can comprehensively and quantitatively evaluate the reliability of bed data in predicting actual bed turnover demand. Accessibility performance reflects the interference of non-treatment factors on the data, while operational amplitude significance reflects the degree of change in patient operations. Combining these two aspects and correlating them with the mean of the bed control stage sequence allows for analysis of data reliability from multiple dimensions. Accurately determining data reference value helps hospitals more accurately utilize bed big data for dynamic bed prediction and intelligent allocation, avoiding resource waste or unreasonable allocation due to inaccurate data, improving the utilization efficiency of hospital bed resources, optimizing patient treatment processes, and ultimately enhancing the overall operational efficiency and service quality of the hospital.

[0087] In some implementations, expanding the patient behavior dataset based on the data reference includes: For the patient behavior operation data in the i-th bed control stage, obtain the stage node value and data reference value corresponding to any two adjacent operation data; By comparing the stage node values ​​corresponding to the two adjacent operation data, the stage node value difference is obtained; By comparing the data reference values ​​corresponding to the two adjacent operation data, the data reference value difference is obtained; Based on the difference between the stage node values ​​and the difference in data reference, the number of data points that need to be interpolated is obtained; Based on the number of data points to be interpolated, the two adjacent operation data are uniformly interpolated to generate supplementary operation data, thus obtaining the first expanded dataset. The first expanded dataset is divided into segments to obtain multiple segments, and spline interpolation is performed on the multiple segments to obtain a second expanded dataset, thereby expanding the patient behavior operation dataset.

[0088] Here, the stage node value refers to the position identifier value of each operation data in the patient's behavior operation data in the i-th bed control stage, corresponding to the entire bed control stage process. It can be understood as a value used to locate the specific step or state of the patient's operation within the bed control stage, and through this value, the sequence and specific position of the operation in the entire process can be clearly identified.

[0089] The number of data points to be interpolated is calculated based on the difference between the stage node values ​​corresponding to two adjacent operational data points and the difference in data reference. This number determines the number of data points that need to be inserted between two adjacent operational data points, with the aim of more meticulously filling the data and making the data distribution more continuous and complete to meet the needs of subsequent analysis and model building.

[0090] Uniform interpolation is a method that inserts new data points at equal intervals between two known data points. In this embodiment, new data points are inserted at equal intervals between two adjacent data points, based on the number of data points to be interpolated, so that the data exhibits a uniform variation trend between these two points, thereby increasing the density and detail of the data.

[0091] Segmentation involves dividing the entire data range into multiple independent sub-intervals according to certain rules. In this embodiment, the data after uniform interpolation can be divided into different segments based on specific standards or logic. This facilitates subsequent operations such as constructing spline functions for each sub-interval, improving the accuracy and specificity of data processing.

[0092] Spline interpolation is a process of fitting and interpolating data within segmented intervals using spline functions. A spline function is a piecewise defined function. By constructing appropriate spline functions for each segmented interval, the changing trends of the original data can be better approximated, further optimizing the continuity and smoothness of the data, thereby achieving a more accurate and effective expansion of the patient behavior manipulation dataset.

[0093] For example, bed prediction should be based on the overall bed turnover rate changes in the department to make adjustments to the predictive reference of patient behavior at the corresponding bed control stage. Therefore, the reference of the stage data at the obtained bed control stage position can be used to expand the dataset of patient operations between the obtained raw data, thereby improving the accuracy of the turnover rate analysis of patient operation sequences at a single bed control stage.

[0094] The patient behavior dataset can be expanded through the following steps: Taking the patient behavior operation data of the i-th bed control stage as an example, the stage node values ​​corresponding to any two adjacent operation data are obtained, and are represented as follows: and ; To obtain the data references corresponding to any two adjacent operation data, they are respectively represented as: and ; The number of data points that need to be interpolated can be represented as follows: in, This represents the number of data points that need to be interpolated between the t-th and t-1-th operations in the i-th bed control stage; This indicates rounding down to the nearest integer to ensure that the number of data points is an integer.

[0095] In the way the number of data points is represented, This represents a non-zero constant, used to avoid the extreme case where the denominator of a fraction is zero. Its empirical value can be 1.

[0096] After obtaining the number of data points to be interpolated, uniform interpolation is performed to process the discrete data points, and subsequent spline interpolation processing steps are completed, including segmentation of intervals and construction of spline functions, to process all data.

[0097] In this embodiment, the number of data points requiring interpolation is determined by acquiring stage node values ​​and data references. Then, uniform interpolation, segmented interval division, and spline interpolation are performed to expand the patient behavior dataset. This approach allows for refined supplementation and optimization of the original data based on data references, fully uncovering the potential information and patterns within the data. The expanded dataset reflects patient behavior more comprehensively and meticulously, providing richer and more accurate data support for the subsequent construction of time-series prediction models. This helps improve the accuracy of bed occupancy prediction, enabling hospitals to plan bed resources more rationally, prepare bed allocations in advance, reduce patient waiting times, improve hospital bed utilization efficiency and overall operational efficiency, and simultaneously enhance the patient's medical experience.

[0098] In some implementations, constructing a time-series prediction model based on the expanded patient behavior dataset includes: Use the expanded patient behavior dataset as the model input data; The model input data is trained using a time-series prediction algorithm. During training, the goal is to minimize the occupancy time of a single bed control phase sequence. Prediction weight parameters are constructed, and the time-series prediction model is obtained by adjusting the prediction weight parameters.

[0099] Here, model input data refers to the dataset provided to the model for training and learning. In this embodiment, the model input data is an expanded patient behavior dataset, which contains richer and more detailed information on patient behavior during the bed allocation control phase. This provides basic data support for model training, enabling the model to learn the patterns and rules of bed use based on this data.

[0100] Time series forecasting algorithms are a class of algorithms specifically designed to predict future values ​​based on time series data. Time series data is a sequence of data points arranged in chronological order. These algorithms predict data at future points in time by analyzing the time dependencies and patterns in historical data.

[0101] Training refers to the process of inputting a large amount of data to allow the model to learn patterns, rules, and relationships between variables in the data, thereby adjusting the model's internal parameters to optimize its performance. In this embodiment, an expanded patient behavior manipulation dataset is used to train the model with a time-series prediction algorithm, enabling the model to gradually adapt to the characteristics of the data and improve the accuracy of predictions.

[0102] Among them, the shortest occupancy time of a single bed control phase sequence refers to the state in which the time experienced by each independent bed control phase from start to finish is the shortest during the entire bed use process. This is used as one of the goals of model training, aiming to optimize the model to make the flow of beds in each control phase more efficient, reduce unnecessary waiting time, and improve the overall utilization efficiency of beds.

[0103] The prediction weight parameters are a set of parameters generated during model training, which determine the importance of each input variable or feature when the model makes predictions. Different combinations of weight parameters affect the model's prediction results. By setting the weight parameters appropriately, the model can better fit the patterns and rules in the training data. Adjusting the prediction weight parameters refers to modifying and optimizing the prediction weight parameters during model training based on the difference between the model's prediction results and the actual data. By continuously adjusting the weight parameters, the model's prediction results gradually approach the true values, improving the model's prediction accuracy and performance. In this embodiment, by continuously adjusting the prediction weight parameters, the time-series prediction model built based on the expanded patient behavior operation dataset can better achieve the goal of minimizing the occupancy time of a single bed control phase sequence.

[0104] For example, in the neurology department of a hospital, after expanding the patient behavior dataset, rich data on various stages of bed occupancy were obtained. For instance, in the inpatient treatment bed control stage, detailed behavioral data on patients' daily examinations, treatments, and nursing care were included. This expanded data was used as input data for the model, and a time-series prediction algorithm, such as ARIMA (Autoregressive Integral Moving Average), was selected. During training, the loss function was constructed with the objective of minimizing the occupancy time of a single bed control stage sequence (e.g., the inpatient treatment stage from patient admission to discharge). When the model started training, a set of initial prediction weight parameters was randomly generated. During training, the model made predictions based on the input patient behavior data according to the rules of the ARIMA algorithm, and compared the prediction results with actual bed occupancy times and other data. The loss function was optimized based on the comparison results. If there was a significant difference between the prediction results and the actual data, the prediction weight parameters were adjusted according to a certain optimization algorithm. For example, if it was found that the predicted occupancy time was generally longer than the actual time, the weight parameters related to the occupancy time would be adjusted to make the prediction results closer to the actual values. After multiple iterations and adjustments to the prediction weight parameters, a time series prediction model was finally obtained that can better achieve the goal of minimizing the occupancy time of a single bed control phase sequence.

[0105] In this embodiment, the expanded patient behavior dataset is used as input data for the model. A time-series prediction algorithm is employed for training, and the prediction weight parameters are constructed and adjusted with the goal of minimizing the occupancy time of a single bed control phase sequence, thus obtaining the time-series prediction model. This approach fully utilizes the rich information of the expanded dataset, enabling the model to deeply learn the time-series characteristics and patterns of bed usage. A loss function is constructed with the goal of minimizing occupancy time, and the model is then trained using this function. This encourages the model to optimize from the perspective of improving bed utilization efficiency. By continuously adjusting the prediction weight parameters, the accuracy of the model's predictions of bed occupancy duration and other information is improved. An accurate time-series prediction model helps hospitals plan bed resources in advance, rationally arrange patient admissions and bed transfers, reduce bed idle time and patient waiting time, optimize medical resource allocation, significantly improve hospital operational efficiency, and provide patients with more efficient and convenient medical services.

[0106] In some implementations, the step of predicting the bed occupancy duration for all currently used beds based on the time-series prediction model includes: Obtain real-time data for all currently used beds, including current bed occupancy status, current patient treatment stage, and duration of occupancy. The expanded patient behavior operation dataset and the real-time data are input into the time series prediction model so that the time series prediction model can output the estimated total occupancy time of each currently used bed based on the characteristics of the expanded patient behavior operation dataset and the real-time data. Based on the estimated total occupancy time and the occupancy time of each currently used bed, the remaining occupancy time of the bed is determined to complete the prediction of the occupancy time of all currently used beds.

[0107] The real-time data for all currently used beds refers to the instantaneous information associated with all beds in the hospital that are currently being used by patients. This information comprehensively reflects the current status of the beds and their corresponding patients, and serves as an immediate basis for predicting bed occupancy duration.

[0108] The current bed occupancy status indicates whether a bed is currently occupied by a patient. It can be simply divided into two basic states: occupied or vacant. It is an important part of real-time data and is used to clarify the immediate usage status of beds.

[0109] The patient's current treatment stage represents their position in the treatment process at any given time, such as preoperative preparation, surgery, or postoperative rehabilitation. This information helps predict bed occupancy time by combining it with the general duration patterns of different treatment stages.

[0110] The occupied time refers to the length of time elapsed from when the patient begins using the bed until the current moment. By knowing the occupied time and combining it with the predicted total occupied time, the remaining usable time can be calculated.

[0111] The characteristics of real-time data refer to the features and attributes of real-time data such as current bed occupancy status, current treatment stage of the patient, and duration of occupancy. These characteristics, combined with the features of the expanded patient behavior and operation dataset, help the time-series prediction model to more accurately analyze and predict bed occupancy, such as the changing trends of bed occupancy duration corresponding to different treatment stages.

[0112] For example, in a large general hospital, there are currently 100 beds in use. Bed number 3 in the surgical ward is currently occupied, the patient is in the postoperative wound healing stage, and has been occupied for 5 days. Real-time data on the current bed occupancy status, patient's current treatment stage, and occupied time for each of the 100 beds are collected and input into a previously constructed time-series prediction model. Based on the different statuses of each bed in the real-time data, such as the characteristics of different treatment stages, and the relationship between different treatment stages and bed occupancy time contained in the expanded patient behavior dataset, the model, after calculation and analysis, outputs that the estimated total occupancy time for bed number 3 is 10 days. Therefore, based on the 5 days already occupied, the remaining occupancy time for bed number 3 can be determined to be 5 days, and so on, to complete the prediction of the occupancy time for all 100 currently occupied beds.

[0113] In this embodiment, real-time data corresponding to all currently used beds is acquired, including key information such as current bed occupancy status, patient's current treatment stage, and occupancy duration. This real-time data is then combined with an expanded patient behavior and operation dataset and input into a time-series prediction model to determine the remaining occupancy time of each bed, thus predicting the occupancy duration for all currently used beds. This method allows for real-time and dynamic monitoring of hospital bed usage, leveraging rich data features to improve prediction accuracy. Accurate bed occupancy prediction enables hospitals to plan bed resources in advance, rationally arrange new patient admissions, avoid wasting or over-congestion of bed resources, optimize hospital operation and management processes, and improve overall operational efficiency. Simultaneously, it provides patients with more accurate bed usage information, indirectly improving their medical experience.

[0114] In some implementations, the step of allocating beds to patients based on the remaining occupancy time includes: Obtain the medical needs information of patients awaiting treatment in each department, including the urgency of the patient's condition, appointment time, and expected length of hospital stay; The priority of bed allocation for the patients to be treated is determined based on the urgency of their condition and the appointment time. The remaining time of bed occupancy is matched with the expected length of hospital stay of the patient awaiting treatment, and a bed is allocated to the patient according to the bed allocation priority.

[0115] Among them, the medical treatment demand information refers to the collection of various information related to bed allocation exhibited by patients during their medical treatment process. In this embodiment, the medical treatment demand information mainly includes key information such as the urgency of the patient's condition, appointment time, and expected length of hospital stay. This information comprehensively reflects the patient's demand for bed use and is an important basis for reasonable bed allocation.

[0116] The urgency level of a patient's condition measures the severity of their illness and reflects their urgent need for a bed and treatment. It is generally categorized into different levels such as critical, urgent, and moderate. The more critical the condition, the more urgently a bed needs to be allocated for treatment. It is a crucial factor in determining the priority of bed allocation during the allocation process.

[0117] The bed allocation priority is determined based on a combination of factors, including the urgency of the patient's condition and appointment time, and is used to indicate the order in which different patients awaiting treatment are allocated beds. Patients with higher priority will be allocated beds first, thereby ensuring that medical resources are prioritized for patients with the most urgent needs, reflecting the rationality and fairness of medical resource allocation.

[0118] This process involves matching the remaining occupancy time of existing beds with the expected length of hospital stay for the patients awaiting treatment, and allocating beds to these patients according to bed allocation priorities. This means comparing and matching the predicted remaining availability of currently used beds with the expected length of hospital stay for the patients, while simultaneously incorporating pre-determined bed allocation priorities to rationally allocate beds for them. This method ensures full utilization of bed resources, improves bed utilization efficiency, and meets the needs of diverse patients.

[0119] The estimated length of hospital stay refers to the standard length of time required for a patient to be discharged from admission, which is quantified by combining the department's clinical diagnosis and treatment guidelines, the complexity of the patient's condition, and the length of hospital stay data of similar cases in the past.

[0120] This can be achieved by referencing hospital clinical diagnosis and treatment grading standards, classifying the urgency of patients' conditions into four levels and assigning corresponding scores. The specific grading and scoring rules are as follows: Critical (e.g., severe shock, acute myocardial infarction, requiring immediate resuscitation) is assigned 4 points; Urgent (e.g., acute appendicitis, severe fractures, requiring treatment within 24 hours) is assigned 3 points; Sub-urgent (e.g., acute exacerbation of chronic diseases, moderate infections, requiring treatment within 48 hours) is assigned 2 points; and Ordinary (e.g., routine physical examinations, chronic disease follow-ups, requiring elective treatment) is assigned 1 point. Higher scores indicate greater urgency and require priority bed allocation. The time difference between the patient's appointment time and the current allocation time is calculated based on the current bed allocation time. Considering the need to prioritize the treatment of critically ill patients, weighting coefficients are set for both the urgency of the condition and the appointment time. The scores corresponding to the urgency of the illness are weighted according to the weighting coefficient of the urgency of the illness; the time difference between the patient's appointment time and the current allocation time is weighted according to the weighting coefficient of the appointment time; the results of the two weighting processes are summed to obtain the bed allocation priority.

[0121] For example, in a hospital, there are 10 patients awaiting treatment in departments such as emergency, internal medicine, and surgery. Patient A's condition is classified as critical, with an appointment time of 9:00 AM that day and an estimated hospital stay of 3 days. Patient B's condition is classified as moderate, with an appointment time of 3:00 PM yesterday and an estimated hospital stay of 5 days. After obtaining the treatment needs information of these patients, the hospital determines the priority of bed allocation based on the urgency of their condition and the appointment time. Patient A, due to their critical condition, has a higher priority than Patient B. Simultaneously, the hospital predicts the remaining occupancy time of a surgical bed, finding that one surgical bed has 4 days of remaining occupancy. Matching this remaining occupancy time with the estimated hospital stays of the patients reveals that Patient A's estimated hospital stay of 3 days matches the remaining occupancy time of this bed well, and Patient A's bed allocation priority is high; therefore, this bed is allocated to Patient A. This process is repeated for the other patients awaiting treatment.

[0122] In this embodiment, by acquiring the medical needs information of patients awaiting treatment, the priority of bed allocation is determined by comprehensively considering the urgency of the condition and the appointment time. Then, the remaining time of bed occupancy is matched with the patient's expected length of hospital stay, and beds are allocated to patients according to the priority. This method fully considers the actual needs of patients and the effective utilization of bed resources. It ensures that critically ill patients can receive timely bed treatment, reflecting the rationality of medical resource allocation, and improves bed utilization efficiency through time-matching, reducing bed idling and waste. At the same time, orderly bed allocation also helps optimize hospital operation and management, improve the patient's medical experience, ensure that the hospital can provide services to patients more efficiently, and achieve the rational allocation of medical resources.

[0123] In one embodiment of this application, a method for dynamic prediction and intelligent allocation of hospital beds based on big data is also provided. Figure 2 This is a flowchart illustrating a method for dynamic prediction and intelligent allocation of hospital beds based on big data, provided in one embodiment of the present invention. (See attached diagram.) Figure 2 The method includes the following steps: S210. Obtain big data on bed availability in all departments of the hospital; S220. Based on the routine treatment process and the unconventional treatment process, determine the acquireability of the bed big data, wherein the acquireability is used to indicate the degree to which the bed control phase is affected by the unconventional treatment process; S230. Based on the acquired performance, analyze the data reference value of the bed big data, wherein the data reference value is used to indicate the credibility of the bed big data in predicting the actual bed turnover demand; S240. Based on the data reference, the patient behavior operation dataset is expanded, and a time series prediction model is constructed based on the expanded patient behavior operation dataset. S250. Based on the time-series prediction model, predict the bed occupancy time of all currently used beds to obtain the remaining bed occupancy time, and allocate beds to patients to be treated according to the remaining bed occupancy time.

[0124] Specific embodiments of the method for dynamic prediction and intelligent allocation of hospital beds based on big data in this application can be found in the examples shown in the above-described system for dynamic prediction and intelligent allocation of hospital beds based on big data, and will not be repeated here.

[0125] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0126] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A hospital bed dynamic prediction and intelligent allocation system based on big data, characterized in that, The system includes: The data acquisition unit is used to acquire big data on the number of beds in all departments of the hospital. The first analysis unit is used to determine the acquireability of the bed big data based on routine treatment processes and unconventional treatment processes, wherein the acquireability is used to indicate the degree to which the bed control phase is affected by the unconventional treatment processes; The second analysis unit is used to analyze the data reference value of the bed big data based on the acquisition performance, wherein the data reference value is used to indicate the credibility of the bed big data in predicting the actual bed turnover demand. The model building unit is used to expand the patient behavior operation dataset based on the data reference and to build a time series prediction model based on the expanded patient behavior operation dataset. The prediction unit is used to predict the bed occupancy time of all currently used beds based on the time-series prediction model, obtain the remaining bed occupancy time, and allocate beds to patients to be treated according to the remaining bed occupancy time.

2. The hospital bed dynamic prediction and intelligent allocation system based on big data as described in claim 1, characterized in that, The acquisition of big data on bed availability across all departments within the hospital includes: Vital sign data of patients corresponding to beds in each of the aforementioned departments are collected using pre-set bed monitoring equipment. By using pre-set bed sensors, the occupancy status data of beds in each department is collected, and each bed operation and location change is timestamped to obtain bed operation records and timestamp information; The vital signs data, the occupancy status data, the bed operation records, and the timestamp information are aggregated into the bed big data.

3. The hospital bed dynamic prediction and intelligent allocation system based on big data as described in claim 1, characterized in that, The determination of the acquireability of the bed big data based on routine and unconventional treatment processes includes: Obtain all current operation records for a single bed, including all occupancy control behaviors for that single bed, such as admission procedures, inpatient treatment, surgery, and discharge. Based on all the operation records, the occupancy control behavior with stage characteristics is taken as a bed control stage. All the bed control stages are numbered in the order of occurrence to obtain a stage vector. Construct a multidimensional vector space by taking the stage type of each element of the stage vector as a single dimension; Using N as the initial clustering number, clustering operations are performed on all stage vectors in the multidimensional vector space to obtain k clusters. When k>N, the processing procedure corresponding to the largest cluster is taken as the department's routine treatment procedure, and the processing procedures corresponding to the other clusters besides the largest cluster are taken as the department's non-routine treatment procedures. The acquisition performance of the bed big data is obtained by analyzing the stage vectors of the conventional treatment process and the unconventional treatment process.

4. The hospital bed dynamic prediction and intelligent allocation system based on big data as described in claim 3, characterized in that, The analysis of the stage vectors in the conventional and unconventional treatment processes yields the identifiable characteristics of the bed-based big data, including: Obtain the number of stage vectors in the target treatment process that include the i-th bed control stage, and the total number of stage vectors in the target treatment process, wherein the target treatment process is the conventional treatment process or the unconventional treatment process; Based on the ratio of the number of stage vectors to the total number of stage vectors, determine the maximum percentage performance of the i-th bed control stage under different clusters; The maximum value of the percentage is taken as the acquisition performance of the i-th bed control stage.

5. The hospital bed dynamic prediction and intelligent allocation system based on big data as described in claim 1, characterized in that, The analysis of the data reference value of the bed big data based on the acquired performance includes: Obtain the bed control stage sequence corresponding to a single bed, wherein the bed control stage sequence is composed of the numbers of each bed control stage in chronological order; Obtain the first patient operation stage node value and the second patient operation stage node value of the i-th bed control stage in the bed control stage sequence, wherein the second patient operation stage node value is the patient operation stage node value corresponding to the time point before the first patient operation stage node value, and the patient operation stage node value is a value that quantitatively represents the patient operation progress at different time points within the bed control stage. Based on the absolute value of the difference between the node values ​​of the first patient operation stage and the node values ​​of the second patient operation stage, normalization processing is performed to obtain the significance of the operation amplitude of the patient operation in the i-th bed control stage. The significance of the operation amplitude is used to indicate the degree of change of the patient operation at adjacent time points. The data reference value of the bed big data is determined based on the significance of the operational amplitude and the acquisition performance.

6. The hospital bed dynamic prediction and intelligent allocation system based on big data as described in claim 5, characterized in that, The determination of the data reference value of the bed big data based on the significance of the operational amplitude and the acquisition performance includes: Obtain the acquisitive performance, operational significance, and mean vector of the bed control stage sequence for the i-th bed control stage. Determine the product of the acquired performance and the significance of the operational amplitude, and then determine the cosine similarity between the bed control stage sequence and the mean vector; The ratio of the product to the cosine similarity is normalized to obtain the data reference value of the bed big data.

7. The hospital bed dynamic prediction and intelligent allocation system based on big data as described in claim 1, characterized in that, The expansion of the patient behavior operation dataset based on the aforementioned data reference includes: For the patient behavior operation data in the i-th bed control stage, obtain the stage node value and data reference value corresponding to any two adjacent operation data; By comparing the stage node values ​​corresponding to the two adjacent operation data, the stage node value difference is obtained; By comparing the data reference values ​​corresponding to the two adjacent operation data, the data reference value difference is obtained; Based on the difference between the stage node values ​​and the difference in data reference, the number of data points that need to be interpolated is obtained; Based on the number of data points to be interpolated, the two adjacent operation data are uniformly interpolated to generate supplementary operation data, thus obtaining the first expanded dataset. The first expanded dataset is divided into segments to obtain multiple segments, and spline interpolation is performed on the multiple segments to obtain a second expanded dataset, thereby expanding the patient behavior operation dataset.

8. The hospital bed dynamic prediction and intelligent allocation system based on big data as described in claim 7, characterized in that, The step of constructing a time-series prediction model based on the expanded patient behavior dataset includes: Use the expanded patient behavior dataset as the model input data; The model input data is trained using a time-series prediction algorithm. During training, the goal is to minimize the occupancy time of a single bed control phase sequence. Prediction weight parameters are constructed, and the time-series prediction model is obtained by adjusting the prediction weight parameters.

9. The hospital bed dynamic prediction and intelligent allocation system based on big data as described in claim 1, characterized in that, The step of predicting the occupancy duration of all currently used beds based on the time-series prediction model includes: Obtain real-time data for all currently used beds, including current bed occupancy status, current patient treatment stage, and duration of occupancy. The expanded patient behavior operation dataset and the real-time data are input into the time series prediction model so that the time series prediction model can output the estimated total occupancy time of each currently used bed based on the characteristics of the expanded patient behavior operation dataset and the real-time data. Based on the estimated total occupancy time and the occupancy time of each currently used bed, the remaining occupancy time of the bed is determined to complete the prediction of the occupancy time of all currently used beds.

10. The hospital bed dynamic prediction and intelligent allocation system based on big data as described in claim 1, characterized in that, The process of allocating beds to patients based on the remaining occupancy time includes: Obtain the medical needs information of patients awaiting treatment in each department, including the urgency of the patient's condition, appointment time, and expected length of hospital stay; The priority of bed allocation for the patients to be treated is determined based on the urgency of their condition and the appointment time. The remaining time of bed occupancy is matched with the expected length of hospital stay of the patient awaiting treatment, and a bed is allocated to the patient according to the bed allocation priority.

Images