CN117038039A - Deep learning-based intelligent ward resource scheduling method and system - Google Patents
Deep learning-based intelligent ward resource scheduling method and system Download PDFInfo
- Publication number
- CN117038039A CN117038039A CN202311047105.8A CN202311047105A CN117038039A CN 117038039 A CN117038039 A CN 117038039A CN 202311047105 A CN202311047105 A CN 202311047105A CN 117038039 A CN117038039 A CN 117038039A
- Authority
- CN
- China
- Prior art keywords
- processing result
- transducer model
- data
- decoder
- ward
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000013135 deep learning Methods 0.000 title claims abstract description 43
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000012549 training Methods 0.000 claims abstract description 48
- 238000000605 extraction Methods 0.000 claims abstract description 17
- 238000012545 processing Methods 0.000 claims description 119
- 230000007246 mechanism Effects 0.000 claims description 24
- 238000013528 artificial neural network Methods 0.000 claims description 21
- 238000010606 normalization Methods 0.000 claims description 17
- 238000010276 construction Methods 0.000 claims description 8
- 238000013507 mapping Methods 0.000 claims description 8
- 230000009466 transformation Effects 0.000 claims description 8
- 238000012795 verification Methods 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 2
- 238000013468 resource allocation Methods 0.000 abstract description 4
- 230000000474 nursing effect Effects 0.000 abstract description 3
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000004891 communication Methods 0.000 description 7
- 208000028659 discharge Diseases 0.000 description 7
- 238000003860 storage Methods 0.000 description 7
- 230000008901 benefit Effects 0.000 description 6
- 238000004140 cleaning Methods 0.000 description 4
- 238000004590 computer program Methods 0.000 description 4
- 238000007726 management method Methods 0.000 description 4
- 230000005236 sound signal Effects 0.000 description 4
- 238000013136 deep learning model Methods 0.000 description 3
- 238000003745 diagnosis Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000001932 seasonal effect Effects 0.000 description 3
- 238000001356 surgical procedure Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 229940079593 drug Drugs 0.000 description 2
- 239000003814 drug Substances 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000003862 health status Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000000737 periodic effect Effects 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- KLDZYURQCUYZBL-UHFFFAOYSA-N 2-[3-[(2-hydroxyphenyl)methylideneamino]propyliminomethyl]phenol Chemical compound OC1=CC=CC=C1C=NCCCN=CC1=CC=CC=C1O KLDZYURQCUYZBL-UHFFFAOYSA-N 0.000 description 1
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 201000001098 delayed sleep phase syndrome Diseases 0.000 description 1
- 208000033921 delayed sleep phase type circadian rhythm sleep disease Diseases 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 208000019622 heart disease Diseases 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 206010022000 influenza Diseases 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
- 230000003442 weekly effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
- G16H40/20—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the management or administration of healthcare resources or facilities, e.g. managing hospital staff or surgery rooms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0499—Feedforward networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Primary Health Care (AREA)
- Business, Economics & Management (AREA)
- Epidemiology (AREA)
- General Business, Economics & Management (AREA)
- Pathology (AREA)
- Databases & Information Systems (AREA)
- Image Analysis (AREA)
Abstract
The application provides an intelligent ward resource scheduling method and system based on deep learning, wherein the method comprises the following steps: acquiring historical ward resource data; extracting the characteristics of each piece of history ward resource data to obtain first characteristic data, and constructing a transducer model, wherein the transducer model comprises an encoder and a decoder; training the transducer model by utilizing the first characteristic data to obtain a trained transducer model; and carrying out feature extraction on the ward resource data acquired in real time to obtain second feature data, and inputting the second feature data into the trained transducer model to obtain a resource prediction result corresponding to the ward resource data acquired in real time. The application compares and analyzes the real-time monitored data with the predicted result, and can more accurately predict the future resource demand, thereby providing more reliable basis for resource allocation and scheduling decision, being beneficial to the hospitals to optimize the resource utilization to the greatest extent while providing high-quality nursing, reducing the cost and improving the efficiency.
Description
Technical Field
The application relates to the technical field of intelligent medical treatment, in particular to an intelligent ward resource scheduling method and system based on deep learning.
Background
Currently, in terms of scheduling ward resources, there are the following problems:
the ward resource utilization rate is low: traditional ward resource scheduling generally relies on manual experience and simple rules, and cannot accurately predict the patient's need for a visit and the usage of a hospital bed.
Patient latency is long: because the conventional scheduling method cannot accurately predict the time to visit and the need to visit the patient, the residence of the ward bed and the extension of the waiting time may be caused.
Manual scheduling is complicated: traditional ward resource scheduling generally relies on manual scheduling by medical personnel, and the scheduling process is cumbersome and error-prone.
Medical resource imbalance: in traditional ward resource scheduling, medical resources are often unevenly distributed, resulting in insufficient demand for some departments or patient groups.
The sickroom bed is not enough: in traditional ward resource scheduling, insufficient beds are a common problem, resulting in long patient waiting times and delayed treatment.
In view of the above, there is a need for an intelligent ward resource scheduling method to improve the use efficiency of ward resources and the medical experience of patients.
Disclosure of Invention
The application aims to provide an intelligent ward resource scheduling method and system based on deep learning so as to solve the problems.
In order to achieve the above object, the embodiment of the present application provides the following technical solutions:
in one aspect, an embodiment of the present application provides a deep learning-based smart ward resource scheduling method, where the method includes:
acquiring historical ward resource data;
extracting the characteristics of each piece of history ward resource data to obtain first characteristic data, and constructing a transducer model, wherein the transducer model comprises an encoder and a decoder;
training the transducer model by utilizing the first characteristic data to obtain a trained transducer model;
and carrying out feature extraction on the ward resource data acquired in real time to obtain second feature data, and inputting the second feature data into the trained transducer model to obtain a resource prediction result corresponding to the ward resource data acquired in real time.
In a second aspect, an embodiment of the present application provides an intelligent ward resource scheduling system based on deep learning, where the system includes an acquisition module, a construction module, a training module, and a prediction module.
The acquisition module is used for acquiring the historical ward resource data;
the construction module is used for extracting the characteristics of each piece of history ward resource data to obtain first characteristic data and constructing a transducer model, wherein the transducer model comprises an encoder and a decoder;
the training module is used for training the transducer model by utilizing the first characteristic data to obtain a trained transducer model;
the prediction module is used for extracting the characteristics of the ward resource data acquired in real time to obtain second characteristic data, and inputting the second characteristic data into the trained transducer model to obtain a resource prediction result corresponding to the ward resource data acquired in real time.
In a third aspect, an embodiment of the present application provides an intelligent ward resource scheduling apparatus based on deep learning, where the apparatus includes a memory and a processor. The memory is used for storing a computer program; the processor is used for realizing the intelligent ward resource scheduling method based on the deep learning when executing the computer program.
In a fourth aspect, an embodiment of the present application provides a readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the deep learning-based intelligent ward resource scheduling method described above.
The beneficial effects of the application are as follows:
1. the application can reduce the cost and improve the benefit: by accurate resource scheduling and management, the system can reduce the operating costs of medical institutions. The sickbed and medical care resources are reasonably utilized, the waste and excessive investment of the resources are avoided, and the utilization efficiency of the resources is improved, so that the aims of reducing the cost and improving the benefit are achieved.
2. The application can predict with high precision: the system utilizes the deep learning model to learn and analyze a large amount of historical data, and can accurately predict the requirements of patients on treatment and the service condition of sickbeds. Compared with the traditional statistical method, the deep learning model can capture more complex relevance and regularity, and improves the prediction accuracy of resource requirements.
3. The application can realize real-time resource scheduling: the system can timely find the tension condition of the sickroom bed by monitoring and analyzing the patient's treatment requirements in real time and generate an intelligent scheduling scheme. Medical staff can reasonably arrange sickbed, doctor, nurse and other resources according to the information provided by the system, and the utilization efficiency of ward resources is improved to the greatest extent.
4. The application can realize balanced resource allocation: traditional ward resource scheduling often has the problem of unbalanced medical resource allocation, which results in excessive departments and insufficient supply of other departments. The system comprehensively analyzes the requirements of patients on treatment through the deep learning model, can realize the balanced distribution of medical resources among different departments and patients, and improves the fairness of medical services.
5. The application can improve the experience of patients: by optimizing ward resource scheduling and management, the system can reduce waiting time of patients, improve treatment efficiency and enhance satisfaction of the patients. Meanwhile, balanced resource allocation and efficient scheduling decision can provide better medical services, and the medical experience of patients is improved.
6. The application can realize the self definition of the content of the billboard display module: the visual display large screen and the web end management page support the customization of the analysis interface display template (for example, whether the shift saturation condition of medical staff is displayed, whether the bed usage rate condition and trend of each department are displayed, whether the usage rate and trend of various medical equipment are displayed, whether the usage rate condition and trend of nursing supplies and medicines are displayed, and the like) and the function customization, and a hospital administrator can customize the inquiry and exploration data.
Additional features and advantages of the application will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the embodiments of the application. The objectives and other advantages of the application will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments will be briefly described below, it being understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and other related drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of an intelligent ward resource scheduling method based on deep learning according to an embodiment of the application;
FIG. 2 is a schematic diagram of a deep learning-based intelligent ward resource scheduling system according to an embodiment of the present application;
fig. 3 is a schematic structural diagram of an intelligent ward resource scheduling apparatus based on deep learning according to an embodiment of the present application.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present application more apparent, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments of the present application. The components of the embodiments of the present application generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the application, as presented in the figures, is not intended to limit the scope of the application, as claimed, but is merely representative of selected embodiments of the application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
It should be noted that: like reference numerals or letters denote like items in the following figures, and thus once an item is defined in one figure, no further definition or explanation thereof is necessary in the following figures. Meanwhile, in the description of the present application, the terms "first", "second", and the like are used only to distinguish the description, and are not to be construed as indicating or implying relative importance.
Example 1
As shown in fig. 1, the present embodiment provides an intelligent ward resource scheduling method based on deep learning, which includes step S1, step S2, step S3 and step S4.
Step S1, acquiring historical ward resource data;
step S2, extracting the characteristics of each piece of history ward resource data to obtain first characteristic data, and constructing a transducer model, wherein the transducer model comprises an encoder and a decoder;
in this step, different types of ward resource data have different feature extraction methods, which may specifically include the following:
(1) Ward resource data: bed occupation of each ward recorded per hour, for example, 12 beds at 8:00am, for a total of 20 beds;
the bed occupation condition feature extraction method comprises the following steps: for each ward, the average occupancy, the highest occupancy, the lowest occupancy, etc. characteristics of the bed can be calculated. This can help identify which time periods the ward is more crowded, requiring more beds.
Feature example: ward a, average bed occupancy of 0.6 (average bed occupancy of 12/20=0.6);
(2) Ward resource data: the bed usage of each ward is recorded daily, and the bed occupation is displayed in a time sequence mode.
The bed use trend feature extraction method comprises the following steps: the trend of use of each ward bed is extracted using time series analysis methods such as moving average, exponential smoothing, ARIMA (autoregressive moving average model), etc. This may help predict future bed demands in order to make scheduling preparations in advance.
Feature example: based on the past month data, the bed usage of ward B is trending upward, possibly requiring an increase in the number of beds.
(3) Ward resource data: each time a patient is admitted, his personal information (age, sex), diagnostic information (disease type, severity) etc. are recorded.
Patient admission feature extraction method: for each patient admission, features such as the patient's disease type, severity, treatment expectations, etc. may be extracted. These features can be used to predict the type of bed and the level of care that a patient may need.
Feature example: patient C, age 45 years, sex men, was admitted for diagnosis of heart disease with high severity.
(4) Ward resource data: every time a patient is discharged, his hospitalization time, discharge diagnosis and health status are recorded.
Patient discharge feature extraction method: for each patient discharge, features such as the time the patient stays in the ward, the health status at discharge, etc. may be extracted. These features can be used to predict when there will be a bed empty.
Feature example: patient D, hospitalized for 4 days, and the discharge diagnosis is cold, and the health condition is stable when discharged.
(5) Ward resource data: the time required for cleaning was recorded after each bed cleaning was completed.
The bed cleaning time characteristic extraction method comprises the following steps: every time the bed is empty, a certain time is required for cleaning and preparation. Thus, it can be calculated how long it takes for each bed to be re-usable after discharge.
Feature example: bed E requires 1 hour to clean and prepare after the patient is discharged.
(6) Ward resource data: machine learning algorithms are used to classify each patient, classifying the patient into different categories (severe, general, day-to-day surgery, etc.).
The method for extracting the classification characteristics of the patient comprises the following steps: patients can be classified into different categories such as severe, general, daytime surgery, etc., according to their characteristics and care needs. This can be used to optimize the scheduling of beds and healthcare workers.
Feature example: patient F was classified as severe and patient G was classified as daytime surgery.
(7) Ward resource data: daily or weekly hospital admission and discharge data are recorded to capture seasonal and periodic changes.
Seasonal and periodic feature extraction methods: for some hospitals, bed requirements may vary from season to season or periodically. These features can be extracted to make adjustments when needed.
Feature example: the number of admissions to ward H increases in the summer each year, possibly associated with an increase in seasonal influenza cases.
In this step, the specific method for constructing the transducer model includes:
and S21, constructing a transducer model, wherein the transducer model comprises an encoder and a decoder, the encoder is used for learning the relation and characteristic representation between each ward resource, and the decoder is used for outputting a resource prediction result. In this step, the processing procedure of the encoder includes step S211;
step S211, the encoder is composed of a plurality of identical encoder layers, and each encoder layer sequentially comprises a first self-attention mechanism, residual connection, layer normalization and a first feedforward neural network; when training is carried out, inputting each piece of first characteristic data into the encoder, carrying out context sensing and characteristic extraction through a first self-attention mechanism in the encoder, and then carrying out residual connection and layer normalization processing to obtain a first processing result; inputting the first processing result into the first feedforward neural network, performing nonlinear transformation and feature mapping through the first feedforward neural network to obtain feature representations corresponding to each ward resource data, integrating the feature representations with the first processing result to obtain a second processing result, and outputting the feature representations and the second processing result to the decoder, wherein when the next piece of first feature data is input for training, integrating the second processing result corresponding to the next piece of feature data with all previous second processing results to obtain a new second processing result, and outputting the new second processing result to the decoder.
In this step, when the next piece of the feature data is input for training, the second processing result corresponding to the next piece of the feature data is integrated with all previous second processing results to obtain a new second processing result, and then the new second processing result is output to the decoder, which can be understood as:
for example, after training the first piece of characteristic data, a corresponding second processing result is obtained, and then the second processing result is input into a decoder; after training the second piece of characteristic data, obtaining a second processing result corresponding to the second piece of characteristic data, integrating the second processing result corresponding to the second piece of characteristic data with the second processing result corresponding to the first piece of characteristic data to obtain a new second processing result, and inputting the new second processing result into the decoder; after training the third piece of characteristic data, obtaining a second processing result corresponding to the third piece of characteristic data, integrating the second processing result corresponding to the third piece of characteristic data with a new second processing result, obtaining a new second processing result again, and inputting the new second processing result into a decoder;
in this step, the processing procedure of the decoder includes step S212;
the decoder is composed of a plurality of identical decoder layers, each decoder layer comprising in sequence a second self-attention mechanism, an encoder-decoder attention mechanism, a residual connection, layer normalization and a second feedforward neural network; when training is carried out, inputting the second processing result and the characteristic representation into the second self-attention mechanism, calculating a first correlation between the second processing result and the characteristic representation, calculating a second correlation between adjacent second processing results, and integrating the first correlation and the second correlation to obtain a third processing result; inputting the third processing result into the attention mechanism of the encoder-decoder, and calculating the relevance between the third processing result and the characteristic representation to obtain a fourth processing result; and sequentially carrying out residual connection and layer normalization processing on the fourth processing result to obtain a fifth processing result, inputting the fifth processing result into the second feedforward neural network, carrying out nonlinear transformation and feature mapping processing on the fifth processing result, and outputting a resource prediction result.
In order to stabilize model training and avoid gradient vanishing problems, introducing residual connection and layer normalization in front of the second feedforward neural network;
s3, training the transducer model by utilizing the first characteristic data to obtain a trained transducer model;
when training is performed, training the transducer model by using a plurality of ward resource data of the same type to obtain a trained transducer model corresponding to the ward resource data of the same type, namely, a trained transducer model corresponding to the ward resource data of different types; the specific training step comprises the step S31;
and S31, dividing the characteristic data into a training set and a verification set, training the transducer model by using the training set, adjusting model parameters by using a back propagation and optimization algorithm to obtain a preliminary transducer model, and adjusting parameters of the preliminary transducer model by using the verification set to obtain the trained transducer model.
In this step, the optimization algorithm may be an Adam optimizer;
and S4, performing feature extraction on the ward resource data acquired in real time to obtain second feature data, and inputting the second feature data into the trained transducer model to obtain a resource prediction result corresponding to the ward resource data acquired in real time.
In this step, besides inputting the features corresponding to the ward resource data acquired in real time into the trained transducer model for prediction, the ward resource time sequence data can be input into the trained transducer model for prediction at a period of time before the current moment, and after the resource prediction result corresponding to the ward resource data acquired in real time is obtained, the resource prediction result can be visually displayed in an intuitive and easy-to-understand manner. For example:
the content custom setting of the billboard display module can be carried out: the visual display large screen and the web end management page support the customization of the analysis interface display template (for example, whether the shift saturation condition of medical staff is displayed, whether the bed usage rate condition and trend of each department are displayed, whether the usage rate and trend of various medical equipment are displayed, whether the usage rate condition and trend of nursing supplies and medicines are displayed, and the like) and the function customization, and through the setting, a hospital administrator can search and explore data in a self-defined mode.
Meanwhile, according to the resource prediction result, the resources such as sickbeds, medical staff, medical equipment and the like can be intelligently scheduled and optimized. For example:
(1) Scheduling sickbed resources: according to the prediction result and the actual condition of the sickbed, the system can intelligently arrange the time of entering and exiting the hospital of the patient and the distribution and release of the sickbed, and the idle time and the waiting time of the sickbed are reduced to the greatest extent.
(2) Scheduling medical staff: according to medical staff prediction results and real-time monitoring, the system can reasonably arrange working time and task allocation of medical staff, and ensure that the number and technical level of the medical staff in a ward can meet the requirements of patients.
(3) Medical device scheduling: according to the state of the medical equipment and the prediction result, the system can intelligently schedule and arrange the use of the medical equipment, avoid the conditions of idle use and excessive use of the equipment, and improve the utilization rate and benefit of the equipment.
Example 2
As shown in fig. 2, the present embodiment provides an intelligent ward resource scheduling system based on deep learning, which includes an acquisition module 701, a construction module 702, a training module 703, and a prediction module 704.
An acquisition module 701, configured to acquire historic ward resource data;
the construction module 702 is configured to perform feature extraction on each piece of history ward resource data to obtain first feature data, and construct a transducer model, where the transducer model includes an encoder and a decoder;
the training module 703 is configured to train the transducer model by using the first feature data, so as to obtain a trained transducer model;
and the prediction module 704 is configured to perform feature extraction on ward resource data acquired in real time to obtain second feature data, and input the second feature data into the trained transducer model to obtain a resource prediction result corresponding to the ward resource data acquired in real time.
In one embodiment of the disclosure, the building block 702 further comprises a building unit 7021.
A construction unit 7021, configured to construct a transducer model, where the transducer model includes an encoder for learning a relationship and a feature representation between each of the ward resources, and a decoder for outputting a resource prediction result.
In a specific embodiment of the disclosure, the constructing unit 7021 further includes a first processing unit 70211.
A first processing unit 70211 for said encoder being composed of a plurality of identical encoder layers, each of said encoder layers comprising in sequence a first self-attention mechanism, a residual connection, a layer normalization and a first feed forward neural network; when training is carried out, inputting each piece of first characteristic data into the encoder, carrying out context sensing and characteristic extraction through a first self-attention mechanism in the encoder, and then carrying out residual connection and layer normalization processing to obtain a first processing result; inputting the first processing result into the first feedforward neural network, performing nonlinear transformation and feature mapping through the first feedforward neural network to obtain feature representations corresponding to each ward resource data, integrating the feature representations with the first processing result to obtain a second processing result, and outputting the feature representations and the second processing result to the decoder, wherein when the next first feature data is input for training, integrating the second processing result corresponding to the next first feature data with all previous second processing results to obtain a new second processing result, and outputting the new second processing result to the decoder.
In one embodiment of the present disclosure, the constructing unit 7021 further includes a second processing unit 70212.
A second processing unit 70212 for the decoder to be composed of a plurality of identical decoder layers, each decoder layer comprising in turn a second self-attention mechanism, an encoder-decoder attention mechanism, a residual connection, a layer normalization and a second feedforward neural network; when training is carried out, inputting the second processing result and the characteristic representation into the second self-attention mechanism, calculating a first correlation between the second processing result and the characteristic representation, calculating a second correlation between adjacent second processing results, and integrating the first correlation and the second correlation to obtain a third processing result; inputting the third processing result into the attention mechanism of the encoder-decoder, and calculating the relevance between the third processing result and the characteristic representation to obtain a fourth processing result; and sequentially carrying out residual connection and layer normalization processing on the fourth processing result to obtain a fifth processing result, inputting the fifth processing result into the second feedforward neural network, carrying out nonlinear transformation and feature mapping processing on the fifth processing result, and outputting a resource prediction result.
In a specific embodiment of the present disclosure, the training module 703 further includes a training unit 7031.
The training unit 7031 is configured to divide the feature data into a training set and a verification set, train the transducer model by using the training set, adjust model parameters through a back propagation and optimization algorithm to obtain a preliminary transducer model, and tune the preliminary transducer model by using the verification set to obtain the trained transducer model.
It should be noted that, regarding the system in the above embodiment, the specific manner in which the respective modules perform the operations has been described in detail in the embodiment regarding the method, and will not be described in detail herein.
Example 3
Corresponding to the above method embodiments, the present disclosure further provides a deep learning-based smart ward resource scheduling apparatus, and the deep learning-based smart ward resource scheduling apparatus described below and the deep learning-based smart ward resource scheduling method described above may be referred to correspondingly to each other.
Fig. 3 is a block diagram illustrating an intelligent ward resource scheduling apparatus 800 based on deep learning, according to an exemplary embodiment. As shown in fig. 3, the deep learning-based smart ward resource scheduling device 800 may include: a processor 801, a memory 802. The deep learning-based smart ward resource scheduling device 800 may also include one or more of a multimedia component 803, an input/output (I/O) interface 804, and a communication component 805.
The processor 801 is configured to control the overall operation of the deep learning-based intelligent ward resource scheduling apparatus 800 to perform all or part of the steps of the deep learning-based intelligent ward resource scheduling method. The memory 802 is used to store various types of data to support operation at the deep learning-based smart ward resource scheduling device 800, which may include, for example, instructions for any application or method operating on the deep learning-based smart ward resource scheduling device 800, as well as application-related data, such as contact data, messages, pictures, audio, video, and the like. The Memory 802 may be implemented by any type or combination of volatile or non-volatile Memory devices, such as static random access Memory (Static Random Access Memory, SRAM for short), electrically erasable programmable Read-Only Memory (Electrically Erasable Programmable Read-Only Memory, EEPROM for short), erasable programmable Read-Only Memory (Erasable Programmable Read-Only Memory, EPROM for short), programmable Read-Only Memory (Programmable Read-Only Memory, PROM for short), read-Only Memory (ROM for short), magnetic Memory, flash Memory, magnetic disk, or optical disk. The multimedia component 803 may include a screen and an audio component. Wherein the screen may be, for example, a touch screen, the audio component being for outputting and/or inputting audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signals may be further stored in the memory 802 or transmitted through the communication component 805. The audio assembly further comprises at least one speaker for outputting audio signals. The I/O interface 804 provides an interface between the processor 801 and other interface modules, which may be a keyboard, mouse, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 805 is configured to perform wired or wireless communication between the deep learning-based smart ward resource scheduling device 800 and other devices. Wireless communication, such as Wi-Fi, bluetooth, near field communication (Near FieldCommunication, NFC for short), 2G, 3G or 4G, or a combination of one or more thereof, the respective communication component 805 may thus comprise: wi-Fi module, bluetooth module, NFC module.
In an exemplary embodiment, the deep learning-based smart ward resource scheduling device 800 may be implemented by one or more application specific integrated circuits (Application Specific Integrated Circuit, abbreviated as ASIC), digital signal processors (DigitalSignal Processor, abbreviated as DSP), digital signal processing devices (Digital Signal Processing Device, abbreviated as DSPD), programmable logic devices (Programmable Logic Device, abbreviated as PLD), field programmable gate arrays (Field Programmable Gate Array, abbreviated as FPGA), controllers, microcontrollers, microprocessors, or other electronic components for performing the deep learning-based smart ward resource scheduling method described above.
In another exemplary embodiment, a computer readable storage medium is also provided that includes program instructions that when executed by a processor implement the steps of the deep learning-based intelligent ward resource scheduling method described above. For example, the computer readable storage medium may be the memory 802 including program instructions described above that are executable by the processor 801 of the deep learning-based smart ward resource scheduling device 800 to perform the deep learning-based smart ward resource scheduling method described above.
Example 4
Corresponding to the above method embodiments, the present disclosure further provides a readable storage medium, and a readable storage medium described below and an intelligent ward resource scheduling method based on deep learning described above may be referred to correspondingly.
A readable storage medium having stored thereon a computer program which when executed by a processor implements the steps of the deep learning based intelligent ward resource scheduling method of the above method embodiments.
The readable storage medium may be a usb disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a magnetic disk, or an optical disk, and the like.
The above description is only of the preferred embodiments of the present application and is not intended to limit the present application, but various modifications and variations can be made to the present application by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.
Claims (10)
1. An intelligent ward resource scheduling method based on deep learning is characterized by comprising the following steps:
acquiring historical ward resource data;
extracting the characteristics of each piece of history ward resource data to obtain first characteristic data, and constructing a transducer model, wherein the transducer model comprises an encoder and a decoder;
training the transducer model by utilizing the first characteristic data to obtain a trained transducer model;
and carrying out feature extraction on the ward resource data acquired in real time to obtain second feature data, and inputting the second feature data into the trained transducer model to obtain a resource prediction result corresponding to the ward resource data acquired in real time.
2. The deep learning-based intelligent ward resource scheduling method of claim 1, wherein constructing a transducer model comprises:
a transducer model is constructed, and the transducer model comprises an encoder and a decoder, wherein the encoder is used for learning the relation and characteristic representation between each ward resource, and the decoder is used for outputting a resource prediction result.
3. The deep learning-based intelligent ward resource scheduling method of claim 2, wherein the encoder process comprises:
the encoder is composed of a plurality of identical encoder layers, each of which sequentially comprises a first self-attention mechanism, a residual connection, layer normalization and a first feedforward neural network; when training is carried out, inputting each piece of first characteristic data into the encoder, carrying out context sensing and characteristic extraction through a first self-attention mechanism in the encoder, and then carrying out residual connection and layer normalization processing to obtain a first processing result; inputting the first processing result into the first feedforward neural network, performing nonlinear transformation and feature mapping through the first feedforward neural network to obtain feature representations corresponding to each ward resource data, integrating the feature representations with the first processing result to obtain a second processing result, and outputting the feature representations and the second processing result to the decoder, wherein when the next first feature data is input for training, integrating the second processing result corresponding to the next first feature data with all previous second processing results to obtain a new second processing result, and outputting the new second processing result to the decoder.
4. The deep learning-based intelligent ward resource scheduling method of claim 3, wherein the decoder processing comprises:
the decoder is composed of a plurality of identical decoder layers, each decoder layer comprising in sequence a second self-attention mechanism, an encoder-decoder attention mechanism, a residual connection, layer normalization and a second feedforward neural network; when training is carried out, inputting the second processing result and the characteristic representation into the second self-attention mechanism, calculating a first correlation between the second processing result and the characteristic representation, calculating a second correlation between adjacent second processing results, and integrating the first correlation and the second correlation to obtain a third processing result; inputting the third processing result into the attention mechanism of the encoder-decoder, and calculating the relevance between the third processing result and the characteristic representation to obtain a fourth processing result; and sequentially carrying out residual connection and layer normalization processing on the fourth processing result to obtain a fifth processing result, inputting the fifth processing result into the second feedforward neural network, carrying out nonlinear transformation and feature mapping processing on the fifth processing result, and outputting a resource prediction result.
5. The deep learning-based intelligent ward resource scheduling method of claim 1, wherein training the transducer model using the first feature data to obtain a trained transducer model comprises:
dividing the characteristic data into a training set and a verification set, training the transducer model by using the training set, adjusting model parameters by using a back propagation and optimization algorithm to obtain a preliminary transducer model, and adjusting parameters of the preliminary transducer model by using the verification set to obtain the trained transducer model.
6. An intelligent ward resource scheduling system based on deep learning, which is characterized by comprising:
the acquisition module is used for acquiring the historical ward resource data;
the construction module is used for extracting the characteristics of each piece of history ward resource data to obtain first characteristic data and constructing a transducer model, wherein the transducer model comprises an encoder and a decoder;
the training module is used for training the transducer model by utilizing the first characteristic data to obtain a trained transducer model;
the prediction module is used for extracting the characteristics of the ward resource data acquired in real time to obtain second characteristic data, and inputting the second characteristic data into the trained transducer model to obtain a resource prediction result corresponding to the ward resource data acquired in real time.
7. The deep learning based intelligent ward resource scheduling system of claim 6, wherein the building module comprises:
and the construction unit is used for constructing a transducer model, and the transducer model comprises an encoder and a decoder, wherein the encoder is used for learning the relation and characteristic representation between each ward resource, and the decoder is used for outputting a resource prediction result.
8. The deep learning-based intelligent ward resource scheduling system of claim 7, wherein the construction unit comprises:
a first processing unit for the encoder to be composed of a plurality of identical encoder layers, each of the encoder layers comprising in sequence a first self-attention mechanism, a residual connection, a layer normalization and a first feed forward neural network; when training is carried out, inputting each piece of first characteristic data into the encoder, carrying out context sensing and characteristic extraction through a first self-attention mechanism in the encoder, and then carrying out residual connection and layer normalization processing to obtain a first processing result; inputting the first processing result into the first feedforward neural network, performing nonlinear transformation and feature mapping through the first feedforward neural network to obtain feature representations corresponding to each ward resource data, integrating the feature representations with the first processing result to obtain a second processing result, and outputting the feature representations and the second processing result to the decoder, wherein when the next first feature data is input for training, integrating the second processing result corresponding to the next first feature data with all previous second processing results to obtain a new second processing result, and outputting the new second processing result to the decoder.
9. The deep learning based intelligent ward resource scheduling system of claim 8, wherein the building unit comprises:
a second processing unit for the decoder to be composed of a plurality of identical decoder layers, each decoder layer comprising in sequence a second self-attention mechanism, an encoder-decoder attention mechanism, a residual connection, a layer normalization and a second feed forward neural network; when training is carried out, inputting the second processing result and the characteristic representation into the second self-attention mechanism, calculating a first correlation between the second processing result and the characteristic representation, calculating a second correlation between adjacent second processing results, and integrating the first correlation and the second correlation to obtain a third processing result; inputting the third processing result into the attention mechanism of the encoder-decoder, and calculating the relevance between the third processing result and the characteristic representation to obtain a fourth processing result; and sequentially carrying out residual connection and layer normalization processing on the fourth processing result to obtain a fifth processing result, inputting the fifth processing result into the second feedforward neural network, carrying out nonlinear transformation and feature mapping processing on the fifth processing result, and outputting a resource prediction result.
10. The deep learning based intelligent ward resource scheduling system of claim 6, wherein the training module comprises:
the training unit is used for dividing the characteristic data into a training set and a verification set, training the transducer model by utilizing the training set, adjusting model parameters through a back propagation and optimization algorithm to obtain a preliminary transducer model, and adjusting parameters of the preliminary transducer model by utilizing the verification set to obtain the trained transducer model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311047105.8A CN117038039A (en) | 2023-08-18 | 2023-08-18 | Deep learning-based intelligent ward resource scheduling method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311047105.8A CN117038039A (en) | 2023-08-18 | 2023-08-18 | Deep learning-based intelligent ward resource scheduling method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117038039A true CN117038039A (en) | 2023-11-10 |
Family
ID=88642842
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311047105.8A Pending CN117038039A (en) | 2023-08-18 | 2023-08-18 | Deep learning-based intelligent ward resource scheduling method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117038039A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117672482A (en) * | 2023-12-19 | 2024-03-08 | 江苏道朴网络科技有限公司 | Intelligent distribution management system for medical resources |
| CN117976174A (en) * | 2024-03-31 | 2024-05-03 | 四川省肿瘤医院 | Self-adaptive scheduling system for intravenous catheter department |
-
2023
- 2023-08-18 CN CN202311047105.8A patent/CN117038039A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117672482A (en) * | 2023-12-19 | 2024-03-08 | 江苏道朴网络科技有限公司 | Intelligent distribution management system for medical resources |
| CN117976174A (en) * | 2024-03-31 | 2024-05-03 | 四川省肿瘤医院 | Self-adaptive scheduling system for intravenous catheter department |
| CN117976174B (en) * | 2024-03-31 | 2024-06-04 | 四川省肿瘤医院 | Self-adaptive scheduling system for intravenous catheter department |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11600390B2 (en) | Machine learning clinical decision support system for risk categorization | |
| CN117038039A (en) | Deep learning-based intelligent ward resource scheduling method and system | |
| US11152119B2 (en) | Care path analysis and management platform | |
| US20190325995A1 (en) | Method and system for predicting patient outcomes using multi-modal input with missing data modalities | |
| Paré et al. | Home telemonitoring for chronic disease management: an economic assessment | |
| Rodrigues et al. | Discrete event simulation model for planning Level 2 “step-down” bed needs using NEMS | |
| Duma et al. | An ad hoc process mining approach to discover patient paths of an Emergency Department | |
| CN112562836A (en) | Doctor recommendation method and device, electronic equipment and storage medium | |
| WO2018160929A1 (en) | Patient predictive admission, discharge, and monitoring tool | |
| CN110033866A (en) | Healthalert method, apparatus, computer equipment and storage medium | |
| CN111667927A (en) | ZigBee network intelligent medical system and acquisition node networking method thereof | |
| US20210151140A1 (en) | Event Data Modelling | |
| CN111667915A (en) | Intelligent medical system with disease reasoning and diagnosis method thereof | |
| US20230011521A1 (en) | System and method for adjusting hospital unit capacity based on patient-specific variables | |
| Fong et al. | Evaluation of stream mining classifiers for real‐time clinical decision support system: a case study of blood glucose prediction in diabetes therapy | |
| CN111540460A (en) | Intelligent medical system with knowledge base and diagnosis method thereof | |
| Saad et al. | A sensing-based patient classification framework for efficient patient-nurse scheduling | |
| US20070055543A1 (en) | Medical resource estimation and simulation system | |
| CN111540461A (en) | Intelligent medical system with problem understanding function and diagnosis method thereof | |
| CN112309519B (en) | Electronic medical record medication structured processing system based on multiple models | |
| WO2018112185A1 (en) | Systems and methods for real-time patient volume prediction | |
| US20240112790A1 (en) | System and method for optimizing resource allocation | |
| Wang et al. | DensityTransfer: A data driven approach for imputing electronic health records | |
| US20230008936A1 (en) | System and method for adaptive learning for hospital census simulation | |
| JP2003526137A (en) | Method and apparatus for automatically allocating staffing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |