WO2021189955A1 - 待检查项确定方法、装置、设备及计算机可读存储介质 - Google Patents

待检查项确定方法、装置、设备及计算机可读存储介质 Download PDF

Info

Publication number
WO2021189955A1
WO2021189955A1 PCT/CN2020/135256 CN2020135256W WO2021189955A1 WO 2021189955 A1 WO2021189955 A1 WO 2021189955A1 CN 2020135256 W CN2020135256 W CN 2020135256W WO 2021189955 A1 WO2021189955 A1 WO 2021189955A1
Authority
WO
WIPO (PCT)
Prior art keywords
probability
check
items
feature vector
check item
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.)
Ceased
Application number
PCT/CN2020/135256
Other languages
English (en)
French (fr)
Inventor
李彦轩
刘卓
孙行智
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ping An Technology Shenzhen Co Ltd
Original Assignee
Ping An Technology Shenzhen Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Ping An Technology Shenzhen Co Ltd filed Critical Ping An Technology Shenzhen Co Ltd
Publication of WO2021189955A1 publication Critical patent/WO2021189955A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

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
    • 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
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
    • 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
    • G16H70/00—ICT specially adapted for the handling or processing of medical references
    • G16H70/60—ICT specially adapted for the handling or processing of medical references relating to pathologies
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Definitions

  • This application relates to the technical field of intelligent decision-making, and in particular to a method, device, equipment, and computer-readable storage medium for determining items to be checked.
  • Laboratory examination is an important process in the patient's treatment, and the results have a high reference value for the patient's final diagnosis. Inspection items are generally given by doctors participating in the consultation, and are used to determine the final diagnosis from several similar suspected diagnoses based on known information. However, due to the differences in the clinical experience and medical level of doctors, the test items given by them are also uneven in rationality.
  • the inventor realizes that the existing artificial intelligence technology is applied to the medical field, and machine learning and deep learning algorithms are used to recommend inspection items based on the collected high-quality medical record data, but the results of such methods are ranked according to the degree of recommendation.
  • the actual number of items to be inspected cannot be determined.
  • the associated probability of the case and the inspection item is output. The greater the probability value of a certain inspection item, the more suitable the case is for the inspection item.
  • the case and all inspection items can be obtained, and it is not easy to determine which inspection items should be recommended to the patient. Therefore, how to accurately determine which inspection and inspection items a patient needs to be inspected is a problem that needs to be solved urgently.
  • This application provides a method for determining an item to be inspected, and the method for determining an item to be inspected includes the following steps:
  • the target check item of the patient to be checked is determined from a preset check item library.
  • This application also provides a device for determining an item to be inspected, the device for determining an item to be inspected includes:
  • the obtaining module is used to obtain the symptom description information of the patient to be checked;
  • the determining module is configured to determine the recommendation probability of each check item of the patient to be checked according to the symptom description information and a preset check item recommendation model;
  • the determining module is further configured to determine the number of target check items for the patient to be checked according to the recommendation probability of each check item and a preset check item number recommendation model;
  • the selection module is used to determine the target check item of the patient to be checked from the preset check item library according to the recommended probability of each check item and the number of target check items.
  • the present application also provides a computer device that includes a processor, a memory, and a computer program stored on the memory and executable by the processor, wherein when the computer program is executed by the processor , To implement the steps of the method for determining the items to be checked as follows:
  • the target check item of the patient to be checked is determined from a preset check item library.
  • the present application also provides a computer-readable storage medium having a computer program stored on the computer-readable storage medium, wherein when the computer program is executed by a processor, the steps of the method for determining items to be checked are implemented as follows:
  • the target check item of the patient to be checked is determined from a preset check item library.
  • FIG. 1 is a schematic flow chart of the steps of a method for determining an item to be inspected according to an embodiment of the application;
  • Fig. 2 is a schematic flowchart of sub-steps of the method for determining an item to be checked in Fig. 1;
  • FIG. 3 is a schematic diagram of a hierarchical structure of a recommended model for the number of check items in an embodiment of the application
  • FIG. 4 is a schematic block diagram of a device for determining an item to be checked provided in an embodiment of the application
  • FIG. 5 is a schematic block diagram of sub-modules of the device for determining an item to be checked in FIG. 4;
  • FIG. 6 is a schematic block diagram of the structure of a computer device provided by an embodiment of this application.
  • the embodiments of the present application provide a method, device, equipment, and computer-readable storage medium for determining an item to be checked.
  • the method for determining the items to be checked can be applied to terminal devices, which can be electronic devices such as mobile phones, tablet computers, notebook computers, desktop computers, and personal digital assistants.
  • FIG. 1 is a schematic flowchart of steps of a method for determining an item to be checked provided by an embodiment of the application.
  • the method for determining an item to be checked includes steps S101 to S103.
  • Step S101 Obtain symptom description information of the patient to be checked, and determine the recommendation probability of each check item of the patient to be checked according to the symptom description information and a preset check item recommendation model.
  • the symptom description information includes the uncomfortable parts of the patient under examination, symptoms and the doctor’s preliminary diagnosis results.
  • the uncomfortable parts of the patient during the examination include the head, waist and legs, etc.
  • Symptoms can include dizziness, backache and leg numbness, etc.
  • the symptoms can also be a series of symptoms such as fever and lack of energy.
  • the doctor’s preliminary diagnosis is the result of the doctor’s preliminary diagnosis based on the patient’s symptoms. For example, if the doctor is based on the patient’s stomachache, the preliminary examination result is gastric disease. For example, if the doctor is based on the patient’s chest tightness, the preliminary examination result is trachea. Illness or heart disease.
  • the inspection item recommendation model is obtained by training the machine learning model based on a plurality of sample data, the sample data including the patient's symptoms and the recommended probability of each inspection item corresponding to the marked patient's symptoms.
  • the machine learning model can be selected according to the situation, and this application does not specifically limit it.
  • the machine learning model can be a convolutional neural network model, a recurrent neural network model, and a long and short-term memory network model.
  • the method for establishing the test item recommendation model may be: acquiring the patient's symptoms, and marking the recommended probability of each test item corresponding to the patient's symptoms to construct sample data, and based on the sample data, the convolutional neural network The network model undergoes iterative training to update the model parameters of the convolutional neural network model until the convolutional neural network model converges, thereby obtaining a recommended model for inspection items.
  • the symptom description information of the patient to be checked is obtained, and the symptom description information of the patient to be checked is input into the check item recommendation model to obtain the recommended probability of each check item of the patient to be checked.
  • the recommendation probability of each check item of the patient to be checked can be accurately determined. For example, obtain the symptom description X of the patient to be checked, and input the symptom description X into the check item recommendation model. The probability of getting check item A of the patient to be checked is 15%, the probability of check item B is 20%, and the probability of check item B is 20%. The probability is 30%, and the probability of D is 17%.
  • the patient to be examined inputs the symptom description information of the patient to be examined in the terminal device, and the terminal device further obtains the symptom description information of the patient to be examined.
  • the doctor performs a preliminary examination of the patient to be examined, obtains the symptom description information of the patient to be examined, and inputs the symptom description information of the patient to be examined into the terminal device, so that the terminal device can obtain the symptom description of the patient to be examined Information to determine the items to be checked.
  • Step S102 Determine the number of target examination items for the patient to be examined according to the recommendation probability of each examination item and the preset number of examination items recommendation model.
  • the recommended model for the number of inspection items is obtained by training the preset neural network model according to the recommended probability of each inspection item and the number of labeled inspection items.
  • the recommended model for the number of inspection items includes a feature vector extraction layer and a number of inspection items.
  • the preset neural network model can be selected according to the actual situation, and this application does not specifically limit it.
  • the neural network model can be a convolutional neural network model (Convolutional Neural Networks, CNN), a recurrent neural network model (Recurrent Neural Network, RNN). ), deep neural networks (DeepNeuralNetworks, DNN) and long-short-term memory network models (LSTM, LongShort-TermMemory) and other neural network models.
  • the neural network model can include one of them, or a combination of several neural network models.
  • the method for establishing the recommended model for the number of check items may be: obtaining the recommendation probability of each check item, and labeling the number of check items corresponding to the recommendation probability of each check item to construct a training sample data set ;
  • the preset neural network model is iteratively trained to update the model parameters of the preset neural network model until the preset neural network model converges, thereby obtaining a recommended model for the number of check items.
  • the preset neural network model includes the first LSTM layer, the first Attention layer, the DNN layer, the second LSTM layer, the second Attention layer, the Concatenate (vector stitching layer) and the output layer.
  • the feature vector extraction layer includes the first LSTM layer, the first Attention layer, the DNN layer, the second LSTM layer, and the second Attention layer.
  • the number of check items recommended layer includes Concatenate (vector stitching layer) and the output layer, and the activation of the output layer The function is softmax.
  • step S102 includes sub-step S1021 to sub-step S1022.
  • Sub-step S1021 through the feature vector extraction layer, perform feature extraction processing on the recommendation probability of each check item to obtain multiple feature vectors.
  • the recommended model for the number of check items is obtained by training the preset neural network model according to the recommended probability of each check item, labeled sequence feature vectors, statistical feature vectors, gradient feature vectors, and the number of check items.
  • the training method of the number recommendation model can be: obtain the recommendation probability of each check item, label the sequence feature vector, statistical feature vector, gradient feature vector and the number of check items of the recommendation probability of each check item to obtain the training sample Data set: Based on the training sample data set, iterative training is performed on the preset neural network model to update the model parameters of the preset neural network model until the preset neural network model converges.
  • the preset neural network model includes a feature vector extraction layer and a recommendation layer for the number of check items.
  • the feature vector extraction layer includes a sequence feature vector model, a statistical feature vector model, and a gradient feature vector model.
  • the feature vector model includes a first LSTM layer and a first Attention layer
  • the statistical feature vector model includes a DNN layer
  • the gradient feature vector model includes a second LSTM layer and a second Attention layer
  • a sequence feature vector model, a statistical feature vector model, and a gradient Feature vector models are connected in parallel.
  • the recommended layers for the number of check items include Concatenate (vector stitching layer) and output layer, and the activation function of the output layer is softmax.
  • the iterative training of the preset neural network model includes four simultaneous training tasks.
  • the first training task is based on the recommended probability of each check item and the labeled sequence feature vector.
  • the sequence feature vector model is iteratively trained.
  • the training task The second is the iterative training of the statistical feature vector model based on the recommended probability of each check item and the labeled statistical feature vector.
  • the training task is based on the recommended probability of each check item and the labeled gradient feature vector, and the gradient feature vector model Carry out iterative training.
  • Training task four is to perform iterative training on the recommended layer of the number of check items based on multiple feature vectors and the number of labeled check items.
  • the probability of a training sample data for check item A is 15%
  • the probability of check item B is 20%
  • the probability of C is 30%
  • the probability of D is 17%
  • the labeled sequence feature vector is [a1,a2, a3]
  • the statistical feature vector is [b1,b2,b3]
  • the gradient feature vector is [c1,c2,c3]
  • the number of check items is n
  • the probability of check item A is 15%
  • the probability of check item B is The probability of 20%
  • C is 30%
  • the probability of D is 17%.
  • Input the sequence feature vector model, and the output sequence feature vector is [d1,d2,d3], based on the labeled sequence feature vector and the output sequence feature vector, Update the model parameters of the sequence feature vector model.
  • the probability of a training sample data for check item A is 15%
  • the probability of check item B is 20%
  • the probability of C is 30%
  • the probability of D is 17%
  • the labeled statistical feature vector is [b1,b2, b3]
  • the sequence feature vector is [a1,a2,a3]
  • the gradient feature vector is [c1,c2,c3]
  • the number of check items is n
  • the probability of check item A is 15%
  • the probability of check item B is The probability of 20%
  • C is 30%
  • the probability of D is 17%.
  • Input the statistical feature vector model, and the output statistical feature vector is [D1, D2, D3], based on the labeled statistical feature vector and the output statistical feature vector, Update the model parameters of the statistical feature vector model.
  • the probability of a training sample data for check item A is 15%
  • the probability of check item B is 20%
  • the probability of C is 30%
  • the probability of D is 17%
  • the marked gradient feature vector is [c1,c2, c3]
  • the statistical feature vector is [b1,b2,b3]
  • the sequence feature vector is [a1,a2,a3]
  • the number of check items is n
  • the probability of check item A is 15%
  • the probability of check item B is The probability of 20%
  • C is 30%
  • the probability of D is 17%.
  • Input the gradient feature vector model, and the output gradient feature vector is [e1, e2, e3], based on the labeled gradient feature vector and the output gradient feature vector, Update the model parameters of the gradient feature vector model.
  • the probability of a training sample data for check item A is 15%
  • the probability of check item B is 20%
  • the probability of C is 30%
  • the probability of D is 17%
  • the marked gradient feature vector is [a1,a2, a3]
  • the statistical feature vector is [b1,b2,b3]
  • the sequence feature vector is [c1,c2,c3]
  • the number of check items is n
  • the target feature vector is [a1,a2,a3,b1,b2, b3,c1,c2,c3]
  • the probability of check item A is 15%
  • the probability of check item B is 20%
  • the probability of C is 30%
  • the probability of D is 17%
  • the target feature vector [a1,a2, a3,b1,b2,b3,c1,c2,c3] input to the recommended layer for the number of check items for iterative training to update the model parameters of the recommended layer for the number of check items.
  • the feature vector extraction layer is used to perform sequence feature extraction on the recommendation probability of each check item to obtain the sequence feature vector; the feature vector extraction layer is used to perform statistical feature extraction on the recommendation probability of each check item to obtain the statistical feature Vector: Gradient feature extraction is performed on the recommendation probability of each check item through the feature vector extraction layer to obtain the gradient feature vector.
  • the recommended model for the number of check items is obtained by training the preset neural network model according to the recommendation probability of each check item, the labeled sequence feature vector, the statistical feature vector, the gradient feature vector and the number of check items.
  • Sequence feature extraction is performed on the recommendation probability of each check item through the feature vector extraction layer, and multiple feature vectors of sequence feature vector, statistical feature vector and gradient feature vector are obtained.
  • statistical feature extraction is performed on the recommendation probability of each check item through the feature vector extraction layer, and the method of obtaining the statistical feature vector may be: determining the standard deviation and variance of the recommendation probability according to the recommendation probability of each check item , Range and dispersion coefficient; the standard deviation, variance, range and dispersion coefficient of the recommended probability are processed through the feature vector extraction layer to obtain statistical feature vectors.
  • the standard deviation, variance, range, and dispersion coefficient of the recommended probability are processed by the feature vector extraction layer, and the specific method for obtaining the statistical feature vector may be: standard deviation, variance, range of the recommended probability Annotate the statistical feature vector corresponding to the discrete system to obtain the labeled statistical feature vector sample data set, and perform iterative training on the preset neural network model based on the labeled statistical feature vector sample data set until the preset neural network model converges, Thus, the feature vector extraction layer is obtained.
  • the probability of check item A is 15%
  • the probability of check item B is 20%
  • the probability of C is 30%
  • the probability of D is 25%.
  • the probability of check item A is 15%
  • the probability of check item B is 15%.
  • the probability is 20%
  • the probability of C is 30%
  • the probability of D is 25%.
  • the standard deviation, variance, range and dispersion coefficient are processed, and the standard deviation is 7.8%, the variance is 31.25%, and the range is 15%.
  • the dispersion coefficient is 0.35, the standard deviation is 7.8%, the variance is 31.25%, the range is 15%, and the dispersion coefficient is 0.35 input to the feature vector extraction layer, so that the feature vector extraction layer has a standard deviation of 7.8%, and the variance It is 31.25%, the range is 15%, and the dispersion coefficient is 0.35.
  • Vector extraction is performed to obtain the statistical characteristic vector ⁇ b1, b2, b3, b4 ⁇ .
  • the recommendation probability of each check item is sorted in descending order to obtain the recommendation probability ranking table of the check item; the difference calculation is performed on the recommendation probabilities of two adjacent check items in the recommendation probability ranking table to obtain multiple Probability difference; multiple probability difference values are processed through the feature vector extraction layer to obtain a probability gradient vector.
  • the multiple probability difference values are processed by the feature vector extraction layer to obtain the probability gradient vector.
  • the method may be: label the probability gradient vectors corresponding to the multiple probability difference values to obtain the labeled probability gradient vector samples Data set, based on the labeled probability gradient vector sample data set to iteratively train the preset neural network model until the preset neural network model converges, thereby obtaining the feature vector extraction layer, and inputting multiple probability differences into the feature vector
  • the extraction layer is used to extract the feature vector to perform the extraction of the probability gradient vector for the multiple probability differences to obtain the probability feature vector.
  • the probability of check item A is 15%
  • the probability of check item B is 20%
  • the probability of C is 30%
  • the probability of D is 25%
  • the probability of check item A is 15%
  • the probability of check item B is 20%
  • the probability of check item A is 30%
  • the probability of C is 15%
  • the probability of C is 30%
  • the probability of D is 25%
  • the probability of check item B is 20%
  • the probability of check item A is 15%.
  • the probability of 30% and the probability of D is 25%, the probability of D is 25%, and the probability of check item B is 20%.
  • the probability of 30% and the probability of D is 25%
  • the probability of check item B is 20%.
  • the probability of check item B is 20%, and the probability of check item B is 20%.
  • the probability of A is 15% of the difference 5%; the feature vector extraction layer extracts the feature vector of the difference 5%, the difference 5%, and the difference 5% to obtain the probability gradient vector ⁇ c1, c2, c3 ⁇ .
  • sub-step S1022 the multiple feature vectors are input to the recommendation layer of the number of check items to obtain the number of target check items of the patient to be checked.
  • the recommended layer for the number of check items includes Concatenate (vector stitching layer) and the output layer, and the activation function of the output layer is softmax.
  • the sequence feature vector, the statistical feature vector, and the gradient feature vector are spliced to obtain the target feature vector.
  • the target feature vector can be accurately obtained by splicing sequence feature vector, statistical feature vector and gradient feature vector.
  • the target feature vector is input to the number of inspection items recommendation layer, the probability of the number of inspection items is output, and the number of inspection items with the largest probability is selected as the number of target inspection items. For example, if the target feature vector is input to the recommended layer of the number of check items, the probability that the number of check items is 1 is 20%, the probability that the number of check items is 2 is 25%, and the probability that the number of check items is 3 is 40%, the probability that the number of inspection items is 4 is 30%, and the probability that the number of inspection items is 4 is the greatest, therefore, the number of inspection items is 4 as the target number of inspection items.
  • the trained recommendation model for the number of check items may be stored in the blockchain node.
  • the blockchain is a new application mode of computer technology such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm.
  • Blockchain essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information for verification. The validity of the information (anti-counterfeiting) and the generation of the next block.
  • the blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.
  • Step S103 According to the recommended probability of each check item and the number of target check items, determine the target check item of the patient to be checked from a preset check item library.
  • the preset examination item database is a database composed of each examination item of the patient to be tested.
  • the establishment of the database can be established according to actual conditions, which is not specifically limited in this application.
  • the check items in the preset check item library are sorted in descending order to obtain the check item list; the target check is selected in descending order from the check item list.
  • the number of check items to determine the target check items of the patient to be checked According to the number of target check items and the preset check item library, the target check items of the patient to be checked can be accurately determined.
  • the preset check item library includes check items with a recommendation probability of 20%, check items with a recommendation probability of 14%, check items with a recommendation probability of 30%, 3 check items with a recommendation probability of 40%, 4 , Check item 5 with recommendation probability of 28% and check item 6 with recommendation probability of 22%, according to the recommended probability of check item 1, check item 2, check item 3, check item 4, check item 5 and check item 6 to check
  • the items are sorted in descending order, and the check item list of check item 4, check item 3, check item 5, check item 6, check item 1 and check item 2 is obtained.
  • the number of the target inspection items is 3, and the target inspection items of the patient to be examined are inspection item 4, inspection item 3, and inspection item 5.
  • the method for determining the items to be checked obtained in the above embodiment obtains the symptom description information of the patient to be checked, and determines the recommendation probability of each check item of the patient to be checked according to the symptom description information and the preset check item recommendation model;
  • the recommended probability of each check item and the preset number of check items are recommended models to determine the number of target check items for the patient to be checked; then based on the recommended probability of each check item and the number of target check items, from the preset check item library Determine the target examination items of the patient to be examined.
  • This solution determines the target check items of the patient to be checked by determining the number of target check items for the patient to be checked, and determines the target check item of the patient to be checked from the preset check item library according to the recommended probability of each check item and the number of target check items, which is accurate and fast Determining the items to be checked for patients to be checked greatly improves the efficiency of patient checks.
  • FIG. 4 is a schematic block diagram of an apparatus for determining an item to be checked provided in an embodiment of the present application.
  • the device 200 for determining an item to be checked includes an acquiring module 201, a determining module 202, and a selecting module 203.
  • the obtaining module 201 is used to obtain the symptom description information of the patient to be examined;
  • the determining module 202 is configured to determine the recommendation probability of each check item of the patient to be checked according to the symptom description information and a preset check item recommendation model;
  • the determining module 202 is further configured to determine the number of target check items for the patient to be checked based on the recommendation probability of each check item and a preset check item number recommendation model;
  • the selection module 203 is configured to determine the target check item of the patient to be checked from a preset check item library according to the recommended probability of each check item and the number of target check items.
  • the determining module 202 includes:
  • the feature extraction module 2021 performs feature extraction processing on the recommendation probability of each check item through the feature vector extraction layer to obtain multiple feature vectors.
  • the inspection item number recommendation module 2022 inputs the multiple feature vectors to the inspection item number recommendation layer to obtain the target inspection item number of the patient to be inspected.
  • the feature extraction module 2021 is further configured to:
  • the feature extraction module 2021 is further configured to:
  • the standard deviation, variance, range, and dispersion coefficient of the recommended probability are processed by the feature vector extraction layer to obtain the statistical feature vector.
  • the feature extraction module 2021 is further configured to:
  • the multiple probability difference values are processed by the feature vector extraction layer to obtain the probability gradient vector.
  • the selection module 203 is also used to:
  • FIG. 6 is a schematic block diagram of a structure of a computer device provided by an embodiment of the application.
  • the computer equipment can be a server or a terminal.
  • the computer device includes a processor, a memory, and a network interface connected through a system bus, where the memory may include a non-volatile storage medium and an internal memory.
  • the non-volatile storage medium can store an operating system and a computer program.
  • the computer program includes program instructions, and when the program instructions are executed, the processor can execute any method for determining items to be checked.
  • the processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.
  • the internal memory provides an environment for the operation of the computer program in the non-volatile storage medium.
  • the processor can execute any method for determining the item to be checked.
  • This network interface is used for communication.
  • FIG. 6 is only a block diagram of part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied.
  • the specific computer device may Including more or fewer parts than shown in the figure, or combining some parts, or having a different arrangement of parts.
  • the bus is, for example, an I2C (Inter-integrated Circuit) bus
  • the memory may be a Flash chip, a read-only memory (ROM, Read-OnlyMemory) disk, an optical disk, a U disk or a mobile hard disk, etc.
  • the processor may be a central processing unit. Unit (Central Processing Unit, CPU)
  • the processor can also be other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), field programmable gate array (Field-Programmable Gate Array, FPGA) or Other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
  • the general-purpose processor may be a microprocessor or the processor may also be any conventional processor.
  • the processor is used to run a computer program stored in a memory to implement the following steps:
  • the target check item of the patient to be checked is determined from a preset check item library.
  • the inspection item recommendation model is obtained by training a machine learning model based on a plurality of sample data, and the sample data includes the patient's symptoms and the recommended probability of each inspection item corresponding to the marked patient's symptoms.
  • the processor when the processor implements the recommended model for the number of check items, it is obtained by training a preset neural network model according to the recommendation probability of each check item and the number of marked check items, and the check
  • the item number recommendation model includes a feature vector extraction layer and a check item number recommendation layer; the recommended model is determined according to the recommendation probability of each check item and the preset check item number to determine the target check of the patient to be checked
  • the number of items it is used to achieve:
  • the multiple feature vectors are input to the recommendation layer of the number of examination items to obtain the number of target examination items for the patient to be examined.
  • the processor implements the recommended model for the number of check items based on the recommendation probability of each check item, the labeled sequence feature vector, the statistical feature vector, the gradient feature vector, and the number of check items. Assuming that the neural network model is obtained by training, the feature extraction process is performed on the recommendation probability of each check item through the feature vector extraction layer, and when multiple feature vectors are obtained, it is used to realize:
  • the processor when the processor implements the statistical feature extraction of the recommended probability of each check item through the feature vector extraction layer to obtain a statistical feature vector, it is used to implement:
  • the standard deviation, variance, range, and dispersion coefficient of the recommended probability are processed by the feature vector extraction layer to obtain the statistical feature vector.
  • the processor when the processor implements the gradient feature extraction of the recommended probability of each check item through the feature vector extraction layer to obtain a gradient feature vector, it is used to implement:
  • the multiple probability difference values are processed by the feature vector extraction layer to obtain the probability gradient vector.
  • the processor realizes the determination of the target check item of the patient to be checked from a preset check item library according to the recommended probability of each check item and the number of target check items, Used to achieve:
  • the embodiments of the present application also provide a computer-readable storage medium, and the computer-readable storage medium may be volatile or non-volatile.
  • a computer program is stored on the computer-readable storage medium, and the computer program includes program instructions. The method implemented when the program instructions are executed can refer to the various embodiments of the method for determining items to be checked in this application.
  • the computer-readable storage medium may be the internal storage unit of the computer device described in the foregoing embodiment, for example, the hard disk or memory of the computer device.
  • the computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart memory card (SmartMediaCard, SMC), or a Secure Digital (SD) card equipped on the computer device, Flash Card (FlashCard), etc.
  • a plug-in hard disk such as a smart memory card (SmartMediaCard, SMC), or a Secure Digital (SD) card equipped on the computer device, Flash Card (FlashCard), etc.
  • SD Secure Digital

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Medical Informatics (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Physics & Mathematics (AREA)
  • Primary Health Care (AREA)
  • Epidemiology (AREA)
  • Biophysics (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Pathology (AREA)
  • Medical Treatment And Welfare Office Work (AREA)

Abstract

一种待检查项确定方法、装置、设备及计算机可读存储介质，本申请涉及智能决策技术领域。该方法包括：获取待检查患者的症状描述信息，并根据症状描述信息和预设的检查项推荐模型，确定待检查患者的每个检查项的推荐概率（S101）；根据每个检查项的推荐概率和预设的检查项个数推荐模型，确定待检查患者的目标检查项个数（S102）；根据每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定待检查患者的目标检查项（S103）。该方法通过确定待检查患者的目标检查项个数，并根据每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定待检查患者的目标检查项，准确且快捷的确定待检查患者需要检查项，极大地提高了患者的检查效率。

Description

待检查项确定方法、装置、设备及计算机可读存储介质
本申请要求于2020年10月21日提交中国专利局、申请号为2020111356852，发明名称为“待检查项确定方法、装置、设备及计算机可读存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及智能决策技术领域，尤其涉及一种待检查项确定方法、装置、设备及计算机可读存储介质。
 背景技术
检验检查是患者就诊中的一项重要过程，其结果对患者最终的诊断具有较高的参考价值。检验检查项目一般由参与问诊的医生给出，用于基于已知的信息从几个相近的疑似诊断中确定最终诊断。然而，由于医生在临床经验和医学水平上的差异，其给出的检验检查项目在合理性上也参差不齐。
发明人意识到，现有的将人工智能技术应用到医疗领域，使用机器学习和深度学习算法基于收集到的高质量病历数据进行检验检查项目推荐，但是这类方法的结果是按推荐程度排序的检验检查项目列表，实际需要检验检查项目的个数也不能很好的确定。例如，根据患者的症状描述信息，输出病例和检验检查项目的关联概率，某检验检查项目对应的概率值越大说明该病例更适合做该检验检查项目。但是，这样只能得到该病例和所有检验检查项目的关系，并不好确定应该推荐哪些检验检查项目给患者。因此，如何准确的确定患者需要检验检查哪几项检验检查项目是目前亟待解决的问题。
 技术解决方案
本申请提供一种待检查项确定方法，所述待检查项确定方法包括以下步骤：
获取待检查患者的症状描述信息，并根据所述症状描述信息和预设的检查项推荐模型，确定所述待检查患者的每个检查项的推荐概率；
根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数；
根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项。
本申请还提供一种待检查项确定装置，所述待检查项确定装置包括：
获取模块，用于获取待检查患者的症状描述信息；
确定模块，用于根据所述症状描述信息和预设的检查项推荐模型，确定所述待检查患者的每个检查项的推荐概率；
所述确定模块，还用于根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数；
选取模块，用于根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项。
本申请还提供一种计算机设备，所述计算机设备包括处理器、存储器、以及存储在所述存储器上并可被所述处理器执行的计算机程序，其中所述计算机程序被所述处理器执行时，实现如下述的待检查项确定方法的步骤：
获取待检查患者的症状描述信息，并根据所述症状描述信息和预设的检查项推荐模型，确定所述待检查患者的每个检查项的推荐概率；
根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数；
根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项。
本申请还提供一种计算机可读存储介质，所述计算机可读存储介质上存储有计算机程序，其中所述计算机程序被处理器执行时，实现如下述的待检查项确定方法的步骤：
获取待检查患者的症状描述信息，并根据所述症状描述信息和预设的检查项推荐模型，确定所述待检查患者的每个检查项的推荐概率；
根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数；
根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项。
 附图说明
为了更清楚地说明本申请实施例技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本申请实施例提供的一种待检查项确定方法的步骤流程示意图；
图2为图1中的待检查项确定方法的子步骤流程示意图；
图3为本申请实施例中的检查项个数推荐模型的一层级结构示意图；
图4为本申请实施例提供的一种待检查项确定装置的示意性框图；
图5为图4中的待检查项确定装置的子模块的示意性框图；
图6为本申请实施例提供的一种计算机设备的结构示意性框图。
本申请目的的实现、功能特点及优点将结合实施例，参照附图做进一步说明。
 本发明的实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
附图中所示的流程图仅是示例说明，不是必须包括所有的内容和操作/步骤，也不是必须按所描述的顺序执行。例如，有的操作/步骤还可以分解、组合或部分合并，因此实际执行的顺序有可能根据实际情况改变。
本申请实施例提供一种待检查项确定方法、装置、设备及计算机可读存储介质。其中，该待检查项确定方法可应用于终端设备中，该终端设备可以为手机、平板电脑、笔记本电脑、台式电脑和个人数字助理等电子设备。
下面结合附图，对本申请的一些实施方式作详细说明。在不冲突的情况下，下述的实施例及实施例中的特征可以相互组合。
请参照图1，图1为本申请的实施例提供的一种待检查项确定方法的步骤流程示意图。
如图1所示，该待检查项确定方法包括步骤S101至步骤S103。
步骤S101、获取待检查患者的症状描述信息，并根据所述症状描述信息和预设的检查项推荐模型，确定所述待检查患者的每个检查项的推荐概率。
其中，症状描述信息包括待检查患者身体不舒适的部位、症状和医生初步诊断结果，该检查患者身体不舒服的部位包括头部、腰部和腿部等，症状可以包括头晕、腰酸和腿麻等，症状还可以是发热和没有精神等一系列症状。该医生初步诊断结果为医生根据待检查患者的症状初步诊断的结果，例如，医生根据待检查患者胃痛，则初步检查结果为胃病，又例如，医生根据待检查患者胸闷，则初步检查结果为气管病症或心脏病症。
在一实施例中，检查项推荐模型是根据多个样本数据对机器学习模型进行训练得到的，该样本数据包括患者症状和标注的患者症状对应的每个检查项的推荐概率。其中，机器学习模型可以根据情况进行选择，本申请对此不做具体限定，例如，该机器学习模型可以为卷积神经网络模型、循环神经网络模型和长短时记忆网络模型等机器学习模型。
在一实施例中，检查项推荐模型的建立的方式可以为：获取患者症状，并根据患者症状对应的每个检测项的推荐概率进行标注，以构建样本数据，基于该样本数据对卷积神经网络模型进行迭代训练，以更新卷积神经网络模型的模型参数，直到卷积神经网络模型收敛，从而得到检查项推荐模型。
在一实施例中，获取待检查患者的症状描述信息，将待检查患者的症状描述信息输入至检查项推荐模型中，得到待检查患者的每个检查项的推荐概率。通过将待检查患者的症状描述信息输入至检查项推荐模型中，可以准确的确定待检查患者的每个检查项的推荐概率。例如，获取待检查患者的症状描述X，将该症状描述X输入至检查项推荐模型中，得到该待检查患者的检查项A的概率为15%、检查项B的概率为20%、C的概率为30%、D的概率为17%。
在一实施例中，待检查患者在终端设备中输入待检查患者的症状描述信息，终端设备进而得到待检查患者的症状描述信息。在另一些实施例中，医生对待检查患者进行初步检查，得到待检查患者的症状描述信息，并将该待检查患者的症状描述信息输入至终端设备，以使终端设备获得待检查患者的症状描述信息，以进行待检查项的确定。
步骤S102、根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数。
其中，检查项个数推荐模型是根据每个检查项的推荐概率和标注的检查项个数对预设神经网络模型进行训练得到的，检查项个数推荐模型包括特征向量提取层和检查项个数推荐层。其中，预设神经网络模型可以根据实际情况进行选择，本申请对此不做具体限定，例如，该神经网络模型可以为卷积神经网络模型（ConvolutionalNeuralNetworks, CNN）、循环神经网络模型（RecurrentNeuralNetwork, RNN）、深度神经网络（DeepNeuralNetworks，DNN）和长短时记忆网络模型（LSTM，LongShort-TermMemory）等神经网络模型，该神经网络模型可以包括其中一种，或者几种神经网络模型的组合。
在一实施例中，检查项个数推荐模型的建立方式可以为：获取每个检查项的推荐概率，并标注与每个检查项的推荐概率对应的检查项个数，以构建训练样本数据集；根据训练样本数据集，对预设神经网络模型进行迭代训练，以更新预设神经网络模型的模型参数，直到预设神经网络模型收敛，从而得到检查项个数推荐模型。其中，预设神经网络模型包括第一LSTM层、第一Attention层、DNN层、第二LSTM层、第二Attention层、Concatenate（向量拼接层）和输出层，在预设神经网络模型收敛后，特征向量提取层包括第一LSTM层、第一Attention层、DNN层、第二LSTM层和第二Attention层，检查项个数推荐层包括Concatenate（向量拼接层）和输出层，且输出层的激活函数为softmax。
在一实施例中，如图2所示，步骤S102包括子步骤S1021至子步骤S1022。
子步骤S1021、通过所述特征向量提取层对所述每个检查项的推荐概率进行特征提取处理，得到多个特征向量。
其中，检查项个数推荐模型是根据每个检查项的推荐概率、标注的序列特征向量、统计特征向量、梯度特征向量和检查项个数对预设神经网络模型进行训练得到的，该检查项个数推荐模型的训练方式可以为：获取每个检查项的推荐概率，对每个检查项的推荐概率的序列特征向量、统计特征向量、梯度特征向量和检查项个数进行标注，得到训练样本数据集；基于训练样本数据集，对预设神经网络模型进行迭代训练，以更新预设神经网络模型的模型参数，直到预设神经网络模型收敛。
示例性的，如图3所示，预设神经网络模型包括特征向量提取层和检查项个数推荐层，特征向量提取层包括序列特征向量模型、统计特征向量模型和梯度特征向量模型，该序列特征向量模型包括第一LSTM层和第一Attention层，该统计特征向量模型包括DNN层，该梯度特征向量模型包括第二LSTM层和第二Attention层，序列特征向量模型、统计特征向量模型和梯度特征向量模型并联，检查项个数推荐层包括Concatenate（向量拼接层）和输出层，且输出层的激活函数为softmax。
其中，对预设神经网络模型进行迭代训练包括四个同步进行的训练任务，训练任务一为基于每个检查项的推荐概率和标注的序列特征向量，对序列特征向量模型进行迭代训练，训练任务二为基于每个检查项的推荐概率和标注的统计特征向量，对统计特征向量模型进行迭代训练，训练任务三为基于每个检查项的推荐概率和标注的梯度特征向量，对梯度特征向量模型进行迭代训练，训练任务四为基于多个特征向量和标注的检查项个数，对检查项个数推荐层进行迭代训练。
例如，一个训练样本数据为检查项A的概率为15%、检查项B的概率为20%、C的概率为30%、D的概率为17%、标注的序列特征向量为[a1,a2,a3]，统计特征向量为[b1,b2,b3]，梯度特征向量为[c1,c2,c3]，检查项个数为n，将检查项A的概率为15%、检查项B的概率为20%、C的概率为30%、D的概率为17%输入序列特征向量模型，得到输出的序列特征向量为[d1,d2,d3]，基于标注的序列特征向量和输出的序列特征向量，更新序列特征向量模型的模型参数。
例如，一个训练样本数据为检查项A的概率为15%、检查项B的概率为20%、C的概率为30%、D的概率为17%、标注的统计特征向量为[b1,b2,b3]，序列特征向量为[a1,a2,a3]，梯度特征向量为[c1,c2,c3]，检查项个数为n，将检查项A的概率为15%、检查项B的概率为20%、C的概率为30%、D的概率为17%输入统计特征向量模型，得到输出的统计特征向量为[D1,D2,D3]，基于标注的统计特征向量和输出的统计特征向量，更新统计特征向量模型的模型参数。
例如，一个训练样本数据为检查项A的概率为15%、检查项B的概率为20%、C的概率为30%、D的概率为17%、标注的梯度特征向量为 [c1,c2,c3]，统计特征向量为[b1,b2,b3]，序列特征向量为[a1,a2,a3]，检查项个数为n，将检查项A的概率为15%、检查项B的概率为20%、C的概率为30%、D的概率为17%输入梯度特征向量模型，得到输出的梯度特征向量为[e1,e2,e3]，基于标注的梯度特征向量和输出的梯度特征向量，更新梯度特征向量模型的模型参数。
例如，个训练样本数据为检查项A的概率为15%、检查项B的概率为20%、C的概率为30%、D的概率为17%、标注的梯度特征向量为[a1,a2,a3]，统计特征向量为[b1,b2,b3]，序列特征向量为[c1,c2,c3]，检查项个数为n，则目标特征向量为[a1,a2,a3,b1,b2,b3,c1,c2,c3]，将检查项A的概率为15%、检查项B的概率为20%、C的概率为30%、D的概率为17%和目标特征向量[a1,a2,a3,b1,b2,b3,c1,c2,c3]输入至检查项个数推荐层进行迭代训练更新检查项个数推荐层的模型参数。
在一实施例中，通过特征向量提取层对每个检查项的推荐概率进行序列特征提取，得到序列特征向量；通过特征向量提取层对每个检查项的推荐概率进行统计特征提取，得到统计特征向量；通过特征向量提取层对每个检查项的推荐概率进行梯度特征提取，得到梯度特征向量。其中，检查项个数推荐模型是根据每个检查项的推荐概率、标注的序列特征向量、统计特征向量、梯度特征向量和检查项个数对预设神经网络模型进行训练得到的。通过特征向量提取层对每个检查项的推荐概率进行序列特征提取，得到序列特征向量、统计特征向量和梯度特征向量多个特征向量。
在一实施例中，通过特征向量提取层对每个检查项的推荐概率进行统计特征提取，得到统计特征向量的方式可以为：根据每个检查项的推荐概率，确定推荐概率的标准差、方差、极差和离散系数；通过特征向量提取层对该推荐概率的标准差、方差、极差和离散系数进行处理，得到统计特征向量。
在一实施例中，通过特征向量提取层对该推荐概率的标准差、方差、极差和离散系数进行处理，得到统计特征向量的具体方式可以为：对推荐概率的标准差、方差、极差和离散系对应的统计特征向量进行标注，得到标注的统计特征向量样本数据集，基于该标注的统计特征向量样本数据集对预设神经网络模型进行迭代训练，直到该预设神经网络模型收敛，从而得到特征向量提取层。将推荐概率的标准差、方差、极差和离散系数输入至该特征向量提取层中，以使该特征向量提取层对该推荐概率的标准差、方差、极差和离散系进行统计特征向量提取，得到统计特征向量。通过对每个检查项的推荐概率进行标准差、方差、极差和离散系数计算提高了统计特征向量训练的准确性。
示例性的，检查项A的概率为15%、检查项B的概率为20%、C的概率为30%、D的概率为25%，对检查项A的概率为15%、检查项B的概率为20%、C的概率为30%、D的概率为25%进行标准差、方差、极差和离散系数进行处理，得到标准差为7.8%，方差为31.25%，极差为15%，离散系数为0.35，将标准差为7.8%，方差为31.25%，极差为15%，离散系数为0.35输入至特性向量提取层，以使该特征向量提取层对该标准差为7.8%，方差为31.25%，极差为15%，离散系数为0.35进行向量提取，得到统计特性向量{b1，b2，b3，b4}。
在一实施例中，对每个检查项的推荐概率进行降序排序，得到检查项的推荐概率排序表；对推荐概率排序表中相邻两个检查项的推荐概率进行差值计算，得到多个概率差值；通过特征向量提取层对多个概率差值进行处理，得到概率梯度向量。
在一实施例中，通过特征向量提取层对多个概率差值进行处理，得到概率梯度向量的方式可以为：对多个概率差值对应的概率梯度向量进行标注，得到标注的概率梯度向量样本数据集，基于该标注的概率梯度向量样本数据集对预设神经网络模型进行迭代训练，直到该预设神经网络模型收敛，从而得到特性向量提取层，将多个概率差值输入值该特征向量提取层，以使该特征向量提取对该多个概率差值进行概率梯度向量的提取，得到概率特征向量。
示例性的，检查项A的概率为15%、检查项B的概率为20%、C的概率为30%、D的概率为25%，对该检查项A的概率为15%、检查项B的概率为20%、C的概率为30%、D的概率为25%进行降序排序，得到C的概率为30%、D的概率为25%、检查项B的概率为20%和检查项A的概率为15%，对C的概率为30%、D的概率为25%、检查项B的概率为20%和检查项A的概率为15%进行相邻两项进行差值计算，得到C的概率为30%与D的概率为25%的差值5%、D的概率为25%与检查项B的概率为20%的差值5%、检查项B的概率为20%与检查项A的概率为15%的差值5%；特征向量提取层对差值5%、差值5%和差值5%进行特征向量提取，得到概率梯度向量{c1，c2，c3}。
子步骤S1022、将所述多个特征向量输入至所述检查项个数推荐层，得到所述待检查患者的目标检查项个数。
其中，检查项个数推荐层包括Concatenate（向量拼接层）和输出层，且输出层的激活函数为softmax。
在一实施例中，在得到序列特征向量、统计特征向量和梯度特征向量多个特征向量之后，对序列特征向量、统计特征向量和梯度特征向量进行拼接，得到目标特征向量。通过对序列特征向量、统计特征向量和梯度特征向量进行拼接可以准确的得到目标特征向量。示例性的，序列特征向量{a1，a2，a3，a4}、统计特性向量{b1，b2，b3，b4}，概率梯度向量{c1，c2，c3}，对序列特征向量{a1，a2，a3，a4}、统计特性向量{b1，b2，b3，b4}，概率梯度向量{c1，c2，c3}进行向量拼接，得到目标特征向量{a1，a2，a3，a4，b1，b2，b3，b4，c1，c2，c3}。
在一实施例中，将目标特征向量输入至检查项个数推荐层，输出检查项个数的概率，选取检查项个数概率最大的作为目标检查项个数。例如，将目标特征向量输入至该检查项个数推荐层，得到检查项个数为1的概率为20%、检查项个数为2的概率为25%、检查项个数为3的概率为40%、检查项个数为4的概率为30%，检查项个数为4的概率最大，因此，检查项个数为4为目标检查项个数。
在一实施例中，可以将训练后的检查项个数推荐模型存储于区块链节点中。其中，区块链是分布式数据存储、点对点传输、共识机制、加密算法等计算机技术的新型应用模式。区块链(Blockchain)，本质上是一个去中心化的数据库，是一串使用密码学方法相关联产生的数据块，每一个数据块中包含了一批次网络交易的信息，用于验证其信息的有效性(防伪)和生成下一个区块。区块链可以包括区块链底层平台、平台产品服务层以及应用服务层等。
步骤S103、根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项。
其中，预设检查项库为待检测患者每个检查项组成的数据库。该数据库的建立可以根据实际情况进行建立，本申请对此不做具体限定。
在一实施例中，根据每个检查项的推荐概率，对预设检查项库中的检查项进行降序排序，得到检查项列表；从检查项列表中依次按照从大到小的顺序选取目标检查项个数的检查项，确定待检查患者的目标检查项。根据目标检查项个数和预设检查项库，可以准确确定待检查患者的目标检查项。
示例性的，预设检查项库中包括推荐概率为20%的检查项1、推荐概率为14%的检查项2、推荐概率为30%的检查项3、推荐概率为40%的检查项4、推荐概率为28%的检查项5和推荐概率为22%的检查项6，根据检查项1、检查项2、检查项3、检查项4、检查项5和检查项6的推荐概率对检查项进行降序排序，得到检查项4、检查项3、检查项5、检查项6、检查项1和检查项2的检查项列表。该目标检查项个数为3，则待检查患者的目标检查项为检查项4、检查项3、检查项5。
上述实施例提供的待检查项确定方法，获取待检查患者的症状描述信息，并根据症状描述信息和预设的检查项推荐模型，确定待检查患者的每个检查项的推荐概率；然后根据每个检查项的推荐概率和预设的检查项个数推荐模型，确定待检查患者的目标检查项个数；之后根据每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定待检查患者的目标检查项。本方案通过确定待检查患者的目标检查项个数，并根据每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定待检查患者的目标检查项，准确且快捷的确定待检查患者需要检查项，极大地提高了患者的检查效率。
请参照图4，图4是本申请实施例提供的一种待检查项确定装置的示意性框图。
如图4所示，该待检查项确定装置200包括获取模块201、确定模块202和选取模块203。
获取模块201，用于获取待检查患者的症状描述信息；
确定模块202，用于根据所述症状描述信息和预设的检查项推荐模型，确定所述待检查患者的每个检查项的推荐概率；
所述确定模块202，还用于根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数；
选取模块203，用于根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项。
在一实施例中，如图5所示，所述确定模块202包括：
特征提取模块2021，通过所述特征向量提取层对所述每个检查项的推荐概率进行特征提取处理，得到多个特征向量。
检查项个数推荐模块2022，将所述多个特征向量输入至所述检查项个数推荐层，得到所述待检查患者的目标检查项个数。
在一实施例中，所述特征提取模块2021还用于：
通过所述特征向量提取层对所述每个检查项的推荐概率进行序列特征提取，得到序列特征向量；
通过所述特征向量提取层对所述每个检查项的推荐概率进行统计特征提取，得到统计特征向量；
通过所述特征向量提取层对所述每个检查项的推荐概率进行梯度特征提取，得到梯度特征向量。
在一实施例中，所述特征提取模块2021还用于：
根据所述每个检查项的推荐概率，确定所述推荐概率的标准差、方差、极差和离散系数；
通过所述特征向量提取层对所述推荐概率的标准差、方差、极差和离散系数进行处理，得到所述统计特征向量。
在一实施例中，所述特征提取模块2021还用于：
对每个检查项的推荐概率进行降序排序，得到检查项的推荐概率排序表；
对所述推荐概率排序表中相邻两个检查项的推荐概率进行差值计算，得到多个概率差值；
通过所述特征向量提取层对多个概率差值进行处理，得到所述概率梯度向量。
在一实施例中，所述选取模块203还用于，
根据所述每个检查项的推荐概率，对预设检查项库中的检查项进行降序排序，得到检查项列表；
从所述检查项列表中依次按照从大到小的顺序选取目标检查项个数的检查项，以确定所述待检查患者的目标检查项。
需要说明的是，所属领域的技术人员可以清楚地了解到，为了描述的方便和简洁，上述描述待检查项确定装置的具体工作过程，可以参考前述待检查项确定方法实施例中的对应过程，在此不再赘述。
请参阅图6，图6为本申请实施例提供的一种计算机设备的结构示意性框图。该计算机设备可以为服务器或终端。
如图6所示，该计算机设备包括通过系统总线连接的处理器、存储器和网络接口，其中，存储器可以包括非易失性存储介质和内存储器。
非易失性存储介质可存储操作系统和计算机程序。该计算机程序包括程序指令，该程序指令被执行时，可使得处理器执行任意一种待检查项确定方法。
处理器用于提供计算和控制能力，支撑整个计算机设备的运行。
内存储器为非易失性存储介质中的计算机程序的运行提供环境，该计算机程序被处理器执行时，可使得处理器执行任意一种待检查项确定方法。
该网络接口用于通信。本领域技术人员可以理解，图6中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的计算机设备的限定，具体的计算机设备可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
应当理解的是，该总线比如为I2C（Inter-integratedCircuit）总线，存储器可以是Flash芯片、只读存储器 (ROM，Read-OnlyMemory)磁盘、光盘、U盘或移动硬盘等，处理器可以是中央处理单元 (CentralProcessingUnit，CPU)，该处理器还可以是其他通用处理器、数字信号处理器 (DigitalSignalProcessor，DSP)、专用集成电路 (ApplicationSpecificIntegratedCircuit，ASIC)、现场可编程门阵列 (Field-ProgrammableGateArray，FPGA) 或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。其中，通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。
其中，在一个实施例中，所述处理器用于运行存储在存储器中的计算机程序，以实现如下步骤：
获取待检查患者的症状描述信息，并根据所述症状描述信息和预设的检查项推荐模型，确定所述待检查患者的每个检查项的推荐概率；
根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数；
根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项。
在一个实施例中，所述检查项推荐模型是根据多个样本数据对机器学习模型进行训练得到的，所述样本数据包括患者症状和标注的患者症状对应的每个检查项的推荐概率。
在一个实施例中，所述处理器在实现所述检查项个数推荐模型是根据每个检查项的推荐概率和标注的检查项个数对预设神经网络模型进行训练得到的，所述检查项个数推荐模型包括特征向量提取层和检查项个数推荐层；所述根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数时，用于实现：
通过所述特征向量提取层对所述每个检查项的推荐概率进行特征提取处理，得到多个特征向量；
将所述多个特征向量输入至所述检查项个数推荐层，得到所述待检查患者的目标检查项个数。
在一个实施例中，所述处理器在实现所述检查项个数推荐模型是根据每个检查项的推荐概率、标注的序列特征向量、统计特征向量、梯度特征向量和检查项个数对预设神经网络模型进行训练得到的，所述通过所述特征向量提取层对所述每个检查项的推荐概率进行特征提取处理，得到多个特征向量时，用于实现：
通过所述特征向量提取层对所述每个检查项的推荐概率进行序列特征提取，得到序列特征向量；
通过所述特征向量提取层对所述每个检查项的推荐概率进行统计特征提取，得到统计特征向量；
通过所述特征向量提取层对所述每个检查项的推荐概率进行梯度特征提取，得到梯度特征向量。
在一个实施例中，所述处理器在实现所述通过所述特征向量提取层对所述每个检查项的推荐概率进行统计特征提取，得到统计特征向量时，用于实现：
根据所述每个检查项的推荐概率，确定所述推荐概率的标准差、方差、极差和离散系数；
通过所述特征向量提取层对所述推荐概率的标准差、方差、极差和离散系数进行处理，得到所述统计特征向量。
在一个实施例中，所述处理器在实现所述通过所述特征向量提取层对所述每个检查项的推荐概率进行梯度特征提取，得到梯度特征向量时，用于实现：
对每个检查项的推荐概率进行降序排序，得到检查项的推荐概率排序表；
对所述推荐概率排序表中相邻两个检查项的推荐概率进行差值计算，得到多个概率差值；
通过所述特征向量提取层对多个概率差值进行处理，得到所述概率梯度向量。
在一个实施例中，所述处理器在实现所述根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项时，用于实现：
根据所述每个检查项的推荐概率，对预设检查项库中的检查项进行降序排序，得到检查项列表；
从所述检查项列表中依次按照从大到小的顺序选取目标检查项个数的检查项，以确定所述待检查患者的目标检查项。
需要说明的是，所属领域的技术人员可以清楚地了解到，为了描述的方便和简洁，上述描述计算机设备的具体工作过程，可以参考前述待检查项确定方法实施例中的对应过程，在此不再赘述。
本申请实施例还提供一种计算机可读存储介质，所述计算机可读存储介质可以是易失性的，也可以是非易失性的。所述计算机可读存储介质上存储有计算机程序，所述计算机程序中包括程序指令，所述程序指令被执行时所实现的方法可参照本申请待检查项确定方法的各个实施例。
其中，所述计算机可读存储介质可以是前述实施例所述的计算机设备的内部存储单元，例如所述计算机设备的硬盘或内存。所述计算机可读存储介质也可以是所述计算机设备的外部存储设备，例如所述计算机设备上配备的插接式硬盘，智能存储卡(SmartMediaCard，SMC)，安全数字(SecureDigital，SD)卡，闪存卡(FlashCard)等。
应当理解，在此本申请说明书中所使用的术语仅仅是出于描述特定实施例的目的而并不意在限制本申请。如在本申请说明书和所附权利要求书中所使用的那样，除非上下文清楚地指明其它情况，否则单数形式的“一”、“一个”及“该”意在包括复数形式。
还应当理解，在本申请说明书和所附权利要求书中使用的术语“和/ 或”是指相关联列出的项中的一个或多个的任何组合以及所有可能组合，并且包括这些组合。需要说明的是，在本文中，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者系统不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者系统所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、方法、物品或者系统中还存在另外的相同要素。
上述本申请实施例序号仅仅为了描述，不代表实施例的优劣。以上所述，仅为本申请的具体实施方式，但本申请的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本申请揭露的技术范围内，可轻易想到各种等效的修改或替换，这些修改或替换都应涵盖在本申请的保护范围之内。因此，本申请的保护范围应以权利要求的保护范围为准。

Claims (20)

  1. 一种待检查项确定方法，其中，包括：
    获取待检查患者的症状描述信息，并根据所述症状描述信息和预设的检查项推荐模型，确定所述待检查患者的每个检查项的推荐概率；
    根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数；
    根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项。
  2. 如权利要求1所述的待检查项确定方法，其中，所述检查项推荐模型是根据多个样本数据对机器学习模型进行训练得到的，所述样本数据包括患者症状和标注的患者症状对应的每个检查项的推荐概率。
  3. 如权利要求1所述的待检查项确定方法，其中，所述检查项个数推荐模型是根据每个检查项的推荐概率和标注的检查项个数对预设神经网络模型进行训练得到的，所述检查项个数推荐模型包括特征向量提取层和检查项个数推荐层；所述根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数，包括：
    通过所述特征向量提取层对所述每个检查项的推荐概率进行特征提取处理，得到多个特征向量；
    将所述多个特征向量输入至所述检查项个数推荐层，得到所述待检查患者的目标检查项个数。
  4. 如权利要求3所述的待检查项确定方法，其中，所述检查项个数推荐模型是根据每个检查项的推荐概率、标注的序列特征向量、统计特征向量、梯度特征向量和检查项个数对预设神经网络模型进行训练得到的，所述通过所述特征向量提取层对所述每个检查项的推荐概率进行特征提取处理，得到多个特征向量，包括：
    通过所述特征向量提取层对所述每个检查项的推荐概率进行序列特征提取，得到序列特征向量；
    通过所述特征向量提取层对所述每个检查项的推荐概率进行统计特征提取，得到统计特征向量；
    通过所述特征向量提取层对所述每个检查项的推荐概率进行梯度特征提取，得到梯度特征向量。
  5. 如权利要求4所述的待检查项确定方法，其中，所述通过所述特征向量提取层对所述每个检查项的推荐概率进行统计特征提取，得到统计特征向量，包括：
    根据所述每个检查项的推荐概率，确定所述推荐概率的标准差、方差、极差和离散系数；
    通过所述特征向量提取层对所述推荐概率的标准差、方差、极差和离散系数进行处理，得到所述统计特征向量。
  6. 如权利要求4所述的待检查项确定方法，其中，所述通过所述特征向量提取层对所述每个检查项的推荐概率进行梯度特征提取，得到梯度特征向量，包括：
    对每个检查项的推荐概率进行降序排序，得到检查项的推荐概率排序表；
    对所述推荐概率排序表中相邻两个检查项的推荐概率进行差值计算，得到多个概率差值；
    通过所述特征向量提取层对多个概率差值进行处理，得到所述概率梯度向量。
  7. 如权利要求1-6中任一项所述的待检查项确定方法，其中，所述根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项，包括：
    根据所述每个检查项的推荐概率，对预设检查项库中的检查项进行降序排序，得到检查项列表；
    从所述检查项列表中依次按照从大到小的顺序选取目标检查项个数的检查项，以确定所述待检查患者的目标检查项。
  8. 一种待检查项确定装置，其中，所述待检查项确定装置包括：
    获取模块，用于获取待检查患者的症状描述信息；
    确定模块，用于根据所述症状描述信息和预设的检查项推荐模型，确定所述待检查患者的每个检查项的推荐概率；
    所述确定模块，还用于根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数；
    选取模块，用于根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项。
  9. 一种计算机设备，其中，所述计算机设备包括处理器、存储器、以及存储在所述存储器上并可被所述处理器执行的计算机程序，其中所述计算机程序被所述处理器执行时，实现如下所述的待检查项确定方法：
    获取待检查患者的症状描述信息，并根据所述症状描述信息和预设的检查项推荐模型，确定所述待检查患者的每个检查项的推荐概率；
    根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数；
    根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项。
  10. 如权利要求1所述的计算机设备，其中，所述检查项个数推荐模型是根据每个检查项的推荐概率和标注的检查项个数对预设神经网络模型进行训练得到的，所述检查项个数推荐模型包括特征向量提取层和检查项个数推荐层；所述根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数，包括：
    通过所述特征向量提取层对所述每个检查项的推荐概率进行特征提取处理，得到多个特征向量；
    将所述多个特征向量输入至所述检查项个数推荐层，得到所述待检查患者的目标检查项个数。
  11. 如权利要求10所述的计算机设备，其中，所述检查项个数推荐模型是根据每个检查项的推荐概率、标注的序列特征向量、统计特征向量、梯度特征向量和检查项个数对预设神经网络模型进行训练得到的，所述通过所述特征向量提取层对所述每个检查项的推荐概率进行特征提取处理，得到多个特征向量，包括：
    通过所述特征向量提取层对所述每个检查项的推荐概率进行序列特征提取，得到序列特征向量；
    通过所述特征向量提取层对所述每个检查项的推荐概率进行统计特征提取，得到统计特征向量；
    通过所述特征向量提取层对所述每个检查项的推荐概率进行梯度特征提取，得到梯度特征向量。
  12. 如权利要求11所述的计算机设备，其中，所述通过所述特征向量提取层对所述每个检查项的推荐概率进行统计特征提取，得到统计特征向量，包括：
    根据所述每个检查项的推荐概率，确定所述推荐概率的标准差、方差、极差和离散系数；
    通过所述特征向量提取层对所述推荐概率的标准差、方差、极差和离散系数进行处理，得到所述统计特征向量。
  13. 如权利要求11所述的计算机设备，其中，所述通过所述特征向量提取层对所述每个检查项的推荐概率进行梯度特征提取，得到梯度特征向量，包括：
    对每个检查项的推荐概率进行降序排序，得到检查项的推荐概率排序表；
    对所述推荐概率排序表中相邻两个检查项的推荐概率进行差值计算，得到多个概率差值；
    通过所述特征向量提取层对多个概率差值进行处理，得到所述概率梯度向量。
  14. 如权利要求9-13中任一项所述的计算机设备，其中，所述根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项，包括：
    根据所述每个检查项的推荐概率，对预设检查项库中的检查项进行降序排序，得到检查项列表；
    从所述检查项列表中依次按照从大到小的顺序选取目标检查项个数的检查项，以确定所述待检查患者的目标检查项。
  15. 一种计算机可读存储介质，其中，所述计算机可读存储介质上存储有计算机程序，其中所述计算机程序被处理器执行时，实现如下所述的待检查项确定方法：
    获取待检查患者的症状描述信息，并根据所述症状描述信息和预设的检查项推荐模型，确定所述待检查患者的每个检查项的推荐概率；
    根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数；
    根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项。
  16. 如权利要求15所述的计算机可读存储介质，其中，所述检查项个数推荐模型是根据每个检查项的推荐概率和标注的检查项个数对预设神经网络模型进行训练得到的，所述检查项个数推荐模型包括特征向量提取层和检查项个数推荐层；所述根据所述每个检查项的推荐概率和预设的检查项个数推荐模型，确定所述待检查患者的目标检查项个数，包括：
    通过所述特征向量提取层对所述每个检查项的推荐概率进行特征提取处理，得到多个特征向量；
    将所述多个特征向量输入至所述检查项个数推荐层，得到所述待检查患者的目标检查项个数。
  17. 如权利要求16所述的计算机可读存储介质，其中，所述检查项个数推荐模型是根据每个检查项的推荐概率、标注的序列特征向量、统计特征向量、梯度特征向量和检查项个数对预设神经网络模型进行训练得到的，所述通过所述特征向量提取层对所述每个检查项的推荐概率进行特征提取处理，得到多个特征向量，包括：
    通过所述特征向量提取层对所述每个检查项的推荐概率进行序列特征提取，得到序列特征向量；
    通过所述特征向量提取层对所述每个检查项的推荐概率进行统计特征提取，得到统计特征向量；
    通过所述特征向量提取层对所述每个检查项的推荐概率进行梯度特征提取，得到梯度特征向量。
  18. 如权利要求17所述的计算机可读存储介质，其中，所述通过所述特征向量提取层对所述每个检查项的推荐概率进行统计特征提取，得到统计特征向量，包括：
    根据所述每个检查项的推荐概率，确定所述推荐概率的标准差、方差、极差和离散系数；
    通过所述特征向量提取层对所述推荐概率的标准差、方差、极差和离散系数进行处理，得到所述统计特征向量。
  19. 如权利要求17所述的计算机可读存储介质，其中，所述通过所述特征向量提取层对所述每个检查项的推荐概率进行梯度特征提取，得到梯度特征向量，包括：
    对每个检查项的推荐概率进行降序排序，得到检查项的推荐概率排序表；
    对所述推荐概率排序表中相邻两个检查项的推荐概率进行差值计算，得到多个概率差值；
    通过所述特征向量提取层对多个概率差值进行处理，得到所述概率梯度向量。
  20. 如权利要求15-19中任一项所述的计算机可读存储介质，其中，所述根据所述每个检查项的推荐概率和目标检查项个数，从预设检查项库中确定所述待检查患者的目标检查项，包括：
    根据所述每个检查项的推荐概率，对预设检查项库中的检查项进行降序排序，得到检查项列表；
    从所述检查项列表中依次按照从大到小的顺序选取目标检查项个数的检查项，以确定所述待检查患者的目标检查项。
PCT/CN2020/135256 2020-10-21 2020-12-10 待检查项确定方法、装置、设备及计算机可读存储介质 Ceased WO2021189955A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011135685.2A CN112259245B (zh) 2020-10-21 2020-10-21 待检查项确定方法、装置、设备及计算机可读存储介质
CN202011135685.2 2020-10-21

Publications (1)

Publication Number Publication Date
WO2021189955A1 true WO2021189955A1 (zh) 2021-09-30

Family

ID=74264626

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/135256 Ceased WO2021189955A1 (zh) 2020-10-21 2020-12-10 待检查项确定方法、装置、设备及计算机可读存储介质

Country Status (2)

Country Link
CN (1) CN112259245B (zh)
WO (1) WO2021189955A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114428959A (zh) * 2021-12-20 2022-05-03 北京网神洞鉴科技有限公司 信息系统修正方法、装置、电子设备、介质及产品
CN116030984A (zh) * 2023-03-31 2023-04-28 武汉携康智能健康设备有限公司 一种基于智能健康站的用户体检系统及体检方法
CN116052887A (zh) * 2023-03-01 2023-05-02 联仁健康医疗大数据科技股份有限公司 一种过度检查的检测方法、装置、电子设备及存储介质
CN116206774A (zh) * 2023-04-27 2023-06-02 深圳市浩然盈科通讯科技有限公司 一种结合大数据自动匹配护理治疗方案的方法及系统
CN119067240A (zh) * 2024-08-30 2024-12-03 首都医科大学宣武医院 一种基于区块链和深度学习的待检患者预约方法及系统

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112927788B (zh) * 2021-03-30 2024-02-27 善诊(上海)信息技术有限公司 一种体检项目推荐方法、装置、设备及存储介质
CN113724860A (zh) * 2021-08-31 2021-11-30 平安国际智慧城市科技股份有限公司 基于人工智能的医疗检查推荐方法、装置、设备及介质
CN113948201A (zh) * 2021-09-29 2022-01-18 新瑞鹏宠物医疗集团有限公司 临床决策辅助方法及系统
CN115497616B (zh) * 2022-10-25 2023-05-23 杭州杏林信息科技有限公司 一种感染性疾病辅助决策的方法、系统、设备及存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100312798A1 (en) * 2007-12-28 2010-12-09 Koninklijke Philips Electronics N.V. Retrieval of similar patient cases based on disease probability vectors
CN108565019A (zh) * 2018-04-13 2018-09-21 合肥工业大学 多学科适用的临床检查组合推荐方法及装置
CN110473615A (zh) * 2019-08-16 2019-11-19 北京声智科技有限公司 一种信息处理方法及装置
CN110598116A (zh) * 2019-09-19 2019-12-20 上海腾程医学科技信息有限公司 检验项目推荐方法及装置、终端设备、存储介质

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111696667A (zh) * 2020-06-11 2020-09-22 吾征智能技术(北京)有限公司 一种常见妇科疾病预测模型构建方法和预测系统

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100312798A1 (en) * 2007-12-28 2010-12-09 Koninklijke Philips Electronics N.V. Retrieval of similar patient cases based on disease probability vectors
CN108565019A (zh) * 2018-04-13 2018-09-21 合肥工业大学 多学科适用的临床检查组合推荐方法及装置
CN110473615A (zh) * 2019-08-16 2019-11-19 北京声智科技有限公司 一种信息处理方法及装置
CN110598116A (zh) * 2019-09-19 2019-12-20 上海腾程医学科技信息有限公司 检验项目推荐方法及装置、终端设备、存储介质

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114428959A (zh) * 2021-12-20 2022-05-03 北京网神洞鉴科技有限公司 信息系统修正方法、装置、电子设备、介质及产品
CN116052887A (zh) * 2023-03-01 2023-05-02 联仁健康医疗大数据科技股份有限公司 一种过度检查的检测方法、装置、电子设备及存储介质
CN116052887B (zh) * 2023-03-01 2023-06-27 联仁健康医疗大数据科技股份有限公司 一种过度检查的检测方法、装置、电子设备及存储介质
CN116030984A (zh) * 2023-03-31 2023-04-28 武汉携康智能健康设备有限公司 一种基于智能健康站的用户体检系统及体检方法
CN116030984B (zh) * 2023-03-31 2023-06-09 武汉携康智能健康设备有限公司 一种基于智能健康站的用户体检系统及体检方法
CN116206774A (zh) * 2023-04-27 2023-06-02 深圳市浩然盈科通讯科技有限公司 一种结合大数据自动匹配护理治疗方案的方法及系统
CN119067240A (zh) * 2024-08-30 2024-12-03 首都医科大学宣武医院 一种基于区块链和深度学习的待检患者预约方法及系统

Also Published As

Publication number Publication date
CN112259245A (zh) 2021-01-22
CN112259245B (zh) 2023-07-25

Similar Documents

Publication Publication Date Title
WO2021189955A1 (zh) 待检查项确定方法、装置、设备及计算机可读存储介质
Ghaderzadeh et al. Deep convolutional neural network–based computer-aided detection system for COVID-19 using multiple lung scans: design and implementation study
CN112037912A (zh) 基于医疗知识图谱的分诊模型训练方法、装置及设备
Osman et al. SOM-LWL method for identification of COVID-19 on chest X-rays
CN110704640A (zh) 一种知识图谱的表示学习方法及装置
CN112016295A (zh) 症状数据处理方法、装置、计算机设备及存储介质
US12182311B1 (en) Apparatus and a method for generating a dictionary data filter for data deidentification
CN112016311B (zh) 基于深度学习模型的实体识别方法、装置、设备及介质
Bakasa et al. Stacked ensemble deep learning for pancreas cancer classification using extreme gradient boosting
Li et al. Using association rule mining for phenotype extraction from electronic health records
Wang et al. Fgnn2: A powerful pretraining framework for learning the logic functionality of circuits
CN116467440B (zh) 基于李氏人工肝病历的多层次语义文本分类方法
Akl et al. A hybrid CNN and ensemble model for COVID-19 lung infection detection on chest CT scans
WO2025024554A9 (en) Systems and methods for phenotyping using large language model prompting
CN111755086A (zh) 数据异常检测方法及装置
WO2022227171A1 (zh) 关键信息提取方法、装置、电子设备及介质
WO2021151355A1 (zh) 基于强化学习模型的疾病排序方法、装置、设备及介质
Ameri et al. Clinical decision support systems (CDSS) in assistance to COVID‐19 diagnosis: A scoping review on types and evaluation methods
WO2021114637A1 (zh) 一种基于深度神经网络的数据处理方法及装置
CN115423120A (zh) 一种业务模型的训练方法、训练装置和计算设备
Nazha et al. How I read an article that uses machine learning methods
CN112394924A (zh) 用于生成提问模型的方法、装置、电子设备和介质
US20250201375A1 (en) Leveraging Large Language Models for Automating Lines of Therapy Adjudication in Cancer Patients
Khozama et al. Study the effect of the risk factors in the estimation of the breast cancer risk score using machine learning
Zhang et al. Semi‐supervised graph convolutional networks for the domain adaptive recognition of thyroid nodules in cross‐device ultrasound images

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20926432

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20926432

Country of ref document: EP

Kind code of ref document: A1