A regional center AI scheduling management method and system

Abstract

The application discloses a regional center AI scheduling management method, comprising: setting a business emergency degree for each image examination type, and setting a business calculation level for each image AI model; wherein each image examination type corresponds to an image AI model matched therewith; the business emergency degree is one of emergency, sub-urgency and non-urgency; and the business calculation level is one of high, medium and low. A patient DICOM image is received, the image examination type of each patient is judged, each image examination type forms an image queue to be processed, the image queue to be processed is scheduled based on the business emergency degree and the business calculation level, and a scheduling scheme of the image queue to be processed is output. The application further provides a regional center AI scheduling management system. The application provides a scheduling logic for the work of a regional center AI system, so that the business performance of the center AI can meet the importance and rationality of clinical requirements in priority.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical information, in particular to a regional center AI scheduling management method and system. BACKGROUND

[0002] Image AI is increasingly entering the front-line work of imaging diagnosis. For the county medical community, county hospitals, traditional Chinese medicine hospitals, maternal and child health hospitals, and township medical institutions can form a county medical community. At present, on the one hand, the imaging business has the following characteristics: 1. The overall business volume is limited; 2. There is a general demand for mutual application, mutual diagnosis, and mutual diagnosis; 3. The scanning specification is easy to unify. On the other hand, the price of each work module of image AI is still high, reaching more than 500,000 per AI module. The terminal price of modules such as lung nodules, chest X-rays, rib fractures, coronary artery analysis, and bone age calculation is not less than this price. It is not necessary for each county medical institution to purchase image AI modules separately. Only one set of AI modules needs to be purchased for different disease types in the county private cloud center, and this intensive management concept is obvious.

[0003] Using a group of AI modules in the regional center, there is a problem of scheduling the image queue of multiple medical institutions to be processed. Only when this problem is reasonably solved can the regional AI be practical. SUMMARY

[0004] Therefore, the main purpose of the present application is to provide a regional center AI scheduling management method and system. When multiple medical institutions share a group of regional center AI, a scheduling management according to medical institutions and different disease types is needed to achieve fairness and efficiency and to ensure the practicality of regional center AI.

[0005] To achieve the above purpose, the technical scheme of the present application is as follows:

[0006] On the one hand, the present application provides a regional center AI scheduling management method, comprising:

[0007] S101: setting the business emergency degree for each image examination type and the business calculation level for each image AI model; wherein each image examination type corresponds to an image AI model matched therewith; the business emergency degree is one of emergency, sub-urgent, and non-urgent; the business calculation level is one of high, medium, and low;

[0008] S102: receiving patient DICOM images, judging the image examination type of each patient, forming an image queue to be processed for each image examination type, and scheduling the image queue to be processed based on the business emergency degree and the business calculation level;

[0009] S103: output the scheduling scheme of the image queue to be processed.

[0010] Preferably, when the clinic type is general clinic, the rule of the image queue to be processed is first in first out; when the clinic type is emergency clinic, the rule of the image queue to be processed is to form a separate emergency queue and give priority to the emergency queue.

[0011] Preferably, the scheduling rule is to schedule in the order of emergency, sub-emergency and non-emergency according to the business emergency degree; when the DICOM image of the patient with the emergency degree of business is put into the queue, the DICOM image of the patient is given priority.

[0012] Preferably, when the business emergency degrees are the same and the business calculation amount levels are different, the scheduling is performed in the order of low, medium and high according to the business calculation amount level, and the DICOM image of the patient with the low business calculation amount level is given priority.

[0013] Preferably, the scheduling in the order of low, medium and high according to the business calculation amount level comprises: after arranging the DICOM image of the patient with the low business calculation amount level according to the first preset number, arranging the DICOM image of the patient with the medium business calculation amount level; and after arranging the DICOM image of the patient with the medium business calculation amount level according to the second preset number, arranging the DICOM image of the patient with the high business calculation amount level.

[0014] Preferably, when the image examination types are different, the business emergency degrees are the same, and the business calculation amount levels are the same, the DICOM images of the patients of each image examination type are arranged in a cycle.

[0015] Preferably, when the image examination types are different, the business emergency degrees are the same, and the business calculation amount levels are the same, the scheduling rule is first in first out according to the examination time.

[0016] In another aspect, the present application also provides a regional center AI scheduling management system, comprising: a setting module, a scheduling module and an output module.

[0017] The setting module, the scheduling module and the output module are sequentially connected in order.

[0018] The setting module, the scheduling module and the output module are sequentially connected in order.

[0019] The setting module is used for setting the business emergency degree of each image examination type and the business calculation amount level of each image AI model; wherein each image examination type corresponds to an image AI model matched therewith; the business emergency degree is one of emergency, sub-emergency and non-emergency; and the business calculation amount level is one of high, medium and low.

[0020] The scheduling module is configured to receive the patient DICOM images, determine the image examination types of each patient, form an image queue to be processed for each image examination type, and schedule the image queue to be processed based on the business urgency and the business calculation amount level.

[0021] The output module is configured to output the scheduling scheme of the image queue to be processed.

[0022] Preferably, when the clinic type is ordinary clinic, the rule of the image queue to be processed is first-in first-out; and when the clinic type is emergency clinic, the rule of the image queue to be processed is to form a separate emergency queue and to arrange the emergency queue preferentially.

[0023] Preferably, the scheduling rule is to schedule in the order of emergency, sub-emergency and non-emergency according to the business urgency; and when the patient DICOM image with the emergency business urgency enters the queue, the patient DICOM image is preferentially queued.

[0024] Preferably, when the business urgency is the same and the business calculation amount level is different, the scheduling is performed in the order of low, medium and high according to the business calculation amount level, and the patient DICOM image with the low business calculation amount level is preferentially arranged.

[0025] Preferably, the scheduling in the order of low, medium and high according to the business calculation amount level comprises: arranging a patient DICOM image with the medium business calculation amount level after arranging the patient DICOM images with the low business calculation amount level according to the first preset number; and arranging a patient DICOM image with the high business calculation amount level after arranging the patient DICOM images with the medium business calculation amount level according to the second preset number.

[0026] Preferably, when the image examination types are different, the business urgencies are the same, and the business calculation amount levels are the same, the patient DICOM images of each image examination type are circularly arranged.

[0027] Preferably, when the image examination types are different, the business urgencies are the same, and the business calculation amount levels are the same, the scheduling rule is to arrange in the order of examination time first-in first-out.

[0028] Technical effects of the present application:

[0029] The application provides a regional center AI scheduling management method, which sets a business emergency degree for each image examination type and sets a business calculation level for each image AI model, judges the image examination type of each patient after receiving a patient DICOM image, forms an image queue to be processed for each image examination type, schedules the image queue to be processed based on the business emergency degree and the business calculation level, and outputs a scheduling scheme; when multiple medical institutions share a set of regional center AI, fairness and efficiency can be achieved, and the problem that each county medical institution needs to separately purchase an image AI module is solved, that is, only one set of AI module needs to be purchased for each disease in the private cloud center of the county, medical costs are saved, and the practicality of the regional center AI is ensured. The business performance of the center AI can meet the importance and rationality of clinical demand in priority. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 It is a regional center AI scheduling management method flowchart of embodiment 1.

[0031] Figure 2 It is a scheduling diagram of independent computing power of embodiment 1.

[0032] Figure 3 It is a unified computing power scheduling diagram of embodiment 1.

[0033] Figure 4 It is a regional center AI scheduling management system flowchart of embodiment 2.

[0034] Figure 5 It is a scheduling diagram of independent computing power of embodiment 2.

[0035] Figure 6 It is a unified computing power scheduling diagram of embodiment 2. DETAILED DESCRIPTION

[0036] The application provides a regional center AI scheduling management method, image examinations performed by patients in county hospitals, traditional Chinese medicine hospitals, maternal and child health hospitals and township hospitals are uploaded to a regional center for image AI processing according to diseases, and due to the limited processing capacity of the image AI module, queue management is needed, so AI scheduling management logic of different business types is needed in this case.

[0037] At present, the more practical image AI modules in the image business are chest DR, head CT, chest CT, CT rib fracture, coronary CT, DR bone age, breast molybdenum target and prostate MR.

[0038] The queuing logic is divided into two kinds according to the computing power architecture method: each different AI module has independent computing power; and multiple AI modules use unified GPU computing power.

[0039] The method specifically comprises:

[0040] S101: setting a business emergency degree for each image examination type and setting a business calculation level for each image AI model; wherein each image examination type corresponds to an image AI model matched therewith; the business emergency degree is one of emergency, sub-emergency and non-emergency; and the business calculation level is one of high, medium and low.

[0041] S102: receiving patient DICOM images, judging the image examination type of each patient, forming an image queue to be processed for each image examination type, and scheduling the image queue to be processed based on the business emergency degree and the business calculation level.

[0042] S103: outputting a scheduling scheme of the image queue to be processed.

[0043] In one embodiment provided by the application, when the clinic type is ordinary clinic, the rule of the image queue to be processed is first-in first-out; and when the clinic type is emergency, the rule of the image queue to be processed is to form a separate emergency queue and to arrange the emergency queue preferentially.

[0044] In one embodiment provided by the application, the scheduling rule is to schedule in the order of the business emergency degree being emergency, sub-emergency and non-emergency; and when the patient DICOM image with the business emergency degree of emergency enters the queue, the patient DICOM image is preferentially queued.

[0045] In one embodiment provided by the application, when the business emergency degree is the same and the business calculation level is different, the scheduling is performed in the order of the business calculation level being low, medium and high, and the patient DICOM image with the business calculation level of low is preferentially arranged.

[0046] In one embodiment provided by the application, the scheduling in the order of the business calculation level being low, medium and high comprises: after arranging the patient DICOM image with the business calculation level of low according to a first preset number, arranging a patient DICOM image with the business calculation level of medium; and after arranging the patient DICOM image with the business calculation level of medium according to a second preset number, arranging a patient DICOM image with the business calculation level of high.

[0047] In one embodiment provided by the application, when the image examination types are different, the business emergency degrees are the same, and the business calculation levels are the same, the patient DICOM images of each image examination type are circularly arranged.

[0048] In one embodiment of the present application, when the image examination types are different, the business emergency levels are the same, and the business calculation levels are the same, the scheduling rule is: first in, first out according to the examination time.

[0049] The logic of the overall scheduling is to output a scheduling scheme in combination of unified computing power and independent computing power. Factors affecting the queue order include: the order of examinations, image examination types, whether it is an emergency, single image AI module GPU occupancy, etc. The default order is the first in, first out order logic. The scheduling logic rules are shown in Table 1.

[0050] Table 1: Scheduling logic rules

[0051]

[0052] It should be noted that the rules in Table 1 are not fixed, and the rules shown in Table 1 are only one of the more reasonable rules, and can be adjusted flexibly according to the actual situation.

[0053] In one embodiment of the present application, a regional center AI scheduling management system includes a setting module, a scheduling module, and an output module. Wherein,

[0054] The setting module, the scheduling module, and the output module are sequentially connected in order.

[0055] The setting module is used to set the business emergency level for each image examination type and the business calculation level for each image AI model; wherein each image examination type corresponds to an image AI model matched therewith; the business emergency level is one of emergency, sub-emergency, and non-emergency; and the business calculation level is one of high, medium, and low.

[0056] The scheduling module is used to receive patient DICOM images, determine the image examination type of each patient, form an image queue to be processed for each image examination type, and schedule the image queue to be processed based on the business emergency level and the business calculation level.

[0057] The output module is used to output the scheduling scheme of the image queue to be processed.

[0058] In one embodiment of the present application, when the outpatient type is ordinary outpatient, the rule of the image queue to be processed is first in, first out; and when the outpatient type is emergency, the rule of the image queue to be processed is to form a separate emergency queue and to prioritize the emergency queue.

[0059] In one embodiment of the present application, the scheduling rule is: scheduling according to the order of emergency, sub-emergency, and non-emergency; and when a patient DICOM image with an emergency level enters the queue, the patient DICOM image is given priority in the queue.

[0060] In one embodiment of the present application, when the service emergency degree is the same and the service calculation amount level is different, the scheduling is performed in the order of low, medium and high service calculation amount level, and the patient DICOM image with low service calculation amount level is preferentially arranged.

[0061] In one embodiment of the present application, the scheduling in the order of low, medium and high service calculation amount level comprises: arranging a patient DICOM image with medium service calculation amount level after arranging the patient DICOM image with low service calculation amount level according to the first preset number; and arranging a patient DICOM image with high service calculation amount level after arranging the patient DICOM image with medium service calculation amount level according to the second preset number.

[0062] In one embodiment of the present application, when the image examination types are different, the service emergency degrees are the same, and the service calculation amount levels are the same, the patient DICOM images of each image examination type are arranged in a cycle.

[0063] In one embodiment of the present application, when the image examination types are different, the service emergency degrees are the same, and the service calculation amount levels are the same, the scheduling rule is: first in, first out according to the examination time.

[0064] The present application provides a regional center AI scheduling management method and system, and the protection scope of the present application is not limited to the embodiments given, and all the deformations or variants meeting the spirit of the present application belong to the protection scope of the present application.

[0065] Embodiment 1: a regional center AI scheduling management method

[0066] The regional center AI scheduling management method of the present embodiment. As shown in the figure, it comprises: Figure 1

[0067] S101: setting the service emergency degree for each image examination type, and setting the service calculation amount level for each image AI model; wherein each image examination type corresponds to an image AI model matched therewith; the service emergency degree is one of emergency, sub-urgent and non-urgent; and the service calculation amount level is one of high, medium and low.

[0068] ​S102: receiving a patient DICOM image, determining the image examination type of each patient, each image examination type forming an image queue to be processed, based on the business emergency level, the business calculation level, using unified computing power combined with independent computing power to schedule the image queue to be processed; this embodiment is divided into eight examination types: chest DR, head CT, chest CT, CT rib fracture, coronary CT, DR bone age, mammography, and prostate MR.

[0069] S103: output the scheduling scheme of the image queue to be processed. The scheduling is as shown in Figures 2-3 .

[0070] Embodiment 2: a regional center AI scheduling management system

[0071] The regional center AI scheduling management system of this embodiment is as shown in Figure 4 , which comprises a setting module 10, a scheduling module 20, and an output module 30. Among them,

[0072] The setting module 10, the scheduling module 20, and the output module 30 are sequentially connected in order.

[0073] The setting module 10 sets the business emergency level for each image examination type, and sets the business calculation level for each image AI model; wherein each image examination type corresponds to an image AI model matched therewith; the business emergency level is one of urgent, sub-urgent, and not urgent; the business calculation level is one of high, medium, and low; the scheduling module 20 receives a patient DICOM image, determines the image examination type of each patient, each image examination type forms an image queue to be processed, based on the business emergency level, the business calculation level, using unified computing power combined with independent computing power to schedule the image queue to be processed; this embodiment is divided into eight image examination types: chest DR, head CT, chest CT, CT rib fracture, coronary CT, DR bone age, mammography, and prostate MR. The output module 30 outputs the scheduling scheme of the image queue to be processed. The scheduling is as shown in Figures 5-6 .

[0074] The above only describes the preferred embodiments of the present disclosure, and cannot be understood as a limitation on the scope of the present disclosure. It should be noted that, for those skilled in the art, without departing from the concept of the present disclosure, a number of modifications and improvements can be made, which are within the protection scope of the present disclosure.

Claims

1. A regional hub AI scheduling management method, characterized by, The method comprises the steps of: S101: setting a business emergency degree for each image examination type and a business calculation level for each image AI model; wherein each image examination type corresponds to an image AI model matched therewith; the business emergency degree is one of emergency, sub-emergency, and non-emergency; and the business calculation level is one of high, medium, and low; S102: receiving patient DICOM images, determining the image examination type of each patient, forming an image queue to be processed for each image examination type, and scheduling the image queue to be processed based on the business emergency degree and the business calculation level; wherein the rule of the image queue to be processed is first-in first-out; when the out-patient type is emergency, the rule of the image queue to be processed is to form a separate emergency queue and to prioritize the arrangement of the emergency queue; the rule of the scheduling is to arrange in the order of emergency, sub-emergency, and non-emergency; when the patient DICOM image with the emergency degree is in the queue, the patient DICOM image is given priority; when the business emergency degrees are the same and the business calculation levels are different, the arrangement is performed in the order of low, medium, and high business calculation levels, and the patient DICOM image with the low business calculation level is given priority; the arrangement in the order of low, medium, and high business calculation levels comprises: after arranging the patient DICOM images with the low business calculation level in a first preset number, arranging a patient DICOM image with the medium business calculation level; and after arranging the patient DICOM images with the medium business calculation level in a second preset number, arranging a patient DICOM image with the high business calculation level; S103: outputting the scheduling scheme of the image queue to be processed. When the image examination types are different, the business emergency degrees are the same, and the business calculation levels are the same, the patient DICOM images of each image examination type are arranged in a cycle.

2. The zone-centric Al scheduling management method of claim 1, wherein, When the image examination types are different, the business emergency degrees are the same, and the business calculation levels are the same, the scheduling rule is to arrange in the order of examination time first-in first-out.

3. The zone-centric Al scheduling management method of claim 2, wherein, The method comprises the steps of:

4. A regional hub AI scheduling management system, characterized by, The setting module, the scheduling module, and the output module are sequentially connected in order. The setting module is configured to set a business emergency degree for each image examination type and a business calculation level for each image AI model; wherein each image examination type corresponds to an image AI model matched therewith; the business emergency degree is one of emergency, sub-emergency, and non-emergency; and the business calculation level is one of high, medium, and low. The setting module is configured to set a business emergency degree for each image examination type and a business calculation level for each image AI model; wherein each image examination type corresponds to an image AI model matched therewith; the business emergency degree is one of emergency, sub-emergency, and non-emergency; and the business calculation level is one of high, medium, and low. ​ ​ The scheduling module is configured to receive patient DICOM images, determine the image examination type of each patient, form an image queue to be processed for each image examination type, and schedule the image queue to be processed based on the business emergency level and the business calculation level; wherein the rule of the image queue to be processed is first in first out; when the clinic type is emergency, the rule of the image queue to be processed is to form a separate emergency queue and to prioritize the emergency queue; and the rule of the scheduling is to schedule in the order of emergency, sub-emergency, and non-emergency. When the patient DICOM image with the emergency level of business is in the queue, the patient DICOM image is given priority. When the emergency level of business is the same, and the business calculation level is different, the scheduling is performed in the order of low, medium, and high business calculation levels, and the patient DICOM image with the low business calculation level is prioritized; the scheduling in the order of low, medium, and high business calculation levels includes: after arranging the patient DICOM images with the low business calculation level according to a first preset number, arranging a patient DICOM image with the medium business calculation level; and after arranging the patient DICOM images with the medium business calculation level according to a second preset number, arranging a patient DICOM image with the high business calculation level. The output module is configured to output the scheduling scheme of the image queue to be processed.

5. The zone-centric Al scheduling management system of claim 4, wherein, When the image examination types are different, the emergency levels of business are the same, and the business calculation levels are the same, the patient DICOM images of each image examination type are arranged in a cycle.

6. The zone-centric Al scheduling management system of claim 5, wherein, When the image examination types are different, the emergency levels of business are the same, and the business calculation levels are the same, the scheduling rule is first in first out according to the examination time.

Images