Medical task scheduling method and system based on artificial intelligence

Abstract

The invention relates to the technical field of medical task scheduling, and discloses a medical task scheduling method and system based on artificial intelligence, and the method comprises the steps: collecting medical resource information and patient information, and carrying out the preprocessing; performing treatment priority calculation on the patients based on the preprocessed patient information; establishing a medical resource matching objective function; and solving the medical resource matching objective function by adopting an improved empire competition algorithm, scheduling medical resources based on a solving result, and allocating the medical resources to the patient. The medical treatment emergency degree, the vital sign abnormal degree and the special crowd priority dimension are extracted from the patient information for medical treatment priority calculation, medical resources needed by the patient are extracted in an artificial intelligence mode, a medical resource matching target function is established, an improved empire competition algorithm is used for solving, and the medical treatment priority is calculated. The optimal medical resources dispatched and distributed by different patients are obtained, and the utilization rate of the medical resources and the dispatching reasonability are improved.

Description

Technical Field

[0001] The present invention relates to the field of medical task scheduling, and in particular to a medical task scheduling method and system based on artificial intelligence. Background Art

[0002] With the advancement of medical technology and the aging of the population, the demand for medical resources is increasing. When providing medical services, hospitals not only need to face the diverse needs of patients, but also deal with the limited resources. The effective allocation and scheduling of medical resources is an important way to improve medical efficiency and optimize medical quality. The problem of medical task scheduling refers to how to reasonably allocate various resources within the hospital (such as medical staff, equipment, beds, etc.) to ensure that patients can get appropriate medical services in a timely manner. Traditional medical task scheduling mostly relies on manual experience. This method has problems such as low scheduling efficiency, slow response speed, and poor accuracy, and cannot meet the needs of modern hospitals in efficient operation. In recent years, the application of artificial intelligence (AI) technology in the medical field has received widespread attention. In particular, with the support of technologies such as machine learning, deep learning, and optimization algorithms, AI provides a new solution for medical task scheduling. AI can process a large amount of medical data and make fast and accurate decisions based on these data to optimize the allocation of medical resources and improve the efficiency and quality of task scheduling. Summary of the invention

[0003] In view of this, the present invention proposes a medical task scheduling method based on artificial intelligence, which maximizes the matching degree between hospital resources and patient treatment needs, improves the utilization rate of medical resources, shortens waiting time, and alleviates the problem of shortage of medical resources.

[0004] To achieve the above object, the present invention provides a medical task scheduling method based on artificial intelligence, comprising the following steps:

[0005] S1: Collect medical resource information and patient information, and perform preprocessing, wherein the preprocessing method includes information cleaning and information standardization;

[0006] S2: Calculating the patient's priority for medical treatment based on the pre-processed patient information, wherein the priority for medical treatment is calculated using a multi-dimensional weighted scoring method;

[0007] S3: combining the patient's priority for medical treatment, the pre-processed patient information and the medical resource information, establishing a medical resource matching objective function, wherein the medical resource matching objective function aims to maximize the matching degree between the patient and the medical resources, and the solution of the medical resource matching objective function is the medical resource scheduling result;

[0008] S4: Use the improved imperial competition algorithm to solve the medical resource matching objective function, schedule medical resources based on the solution results, and allocate medical resources to patients.

[0009] As a further improvement method of the present invention:

[0010] Optionally, the medical resource information includes bed resource information, medical equipment resource information and medical and nursing human resources information, the bed resource information includes the price and usage of beds, the medical equipment resource information includes the price and usage of different medical equipment, and the medical and nursing human resources information includes the shift schedule of medical and nursing staff in different departments and the titles of doctors;

[0011] The usage of the beds includes whether the beds are vacant and the vacancy rates of different types of beds;

[0012] The usage of the medical device refers to whether the medical device is operating normally, the idle rate of different medical devices, and the time period during which the medical device can be used;

[0013] The doctor's titles include resident physician, attending physician, associate chief physician and chief physician.

[0014] Optionally, preprocessing the medical resource information and the patient information includes:

[0015] The preprocessing process of the medical resource information is as follows:

[0016] Screening the medical resource information for abnormal information, wherein the abnormal information is a value beyond a reasonable range, including the vacancy rate of beds, the vacancy rate of medical equipment, the time period during which medical equipment can be used, and the scheduling of medical staff, and correcting the abnormal information;

[0017] Filling missing values ​​in the medical resource information to obtain cleaned medical resource information, wherein the filling method is to use the average value of similar hospitals or to estimate based on historical data;

[0018] The medical resource information after the information cleaning is standardized by using medical standardization rules to obtain the pre-processed medical resource information, wherein the medical standardization rules include a unified time format, unique hot coding of medical device names, coding of doctor titles, whether beds are idle, whether medical devices are operating normally, time periods when medical devices can be used, and scheduling of medical staff, wherein the coding results of the doctor titles of resident physicians, attending physicians, associate chief physicians, and chief physicians are 1-4 in sequence, and the coding results of whether beds are idle and whether medical devices are operating normally are 0-1, wherein 0 indicates that beds are not idle and medical devices cannot operate normally, and 1 indicates that beds are idle and medical devices can operate normally, the time periods when medical devices can be used are coded as 1, the time periods that cannot be used are coded as 0, and the scheduling dates of medical staff are coded as 1;

[0019] The preprocessing process of the patient information is as follows:

[0020] Deleting patient records that are missing key fields, including patient identification information and symptoms;

[0021] Delete duplicate patient records and obtain cleansed patient information;

[0022] The patient information after the information cleaning is standardized by using the patient standardization rule to obtain the pre-processed patient information, wherein the patient standardization rule includes a unified time format, and the patient's symptoms, the urgency of the disease, vital signs, treatment needs, whether the patient is a special population, whether there is an infectious disease, and medical insurance information are encoded, wherein the encoding result of the urgency of the disease is 0-5, and the higher the encoding value, the higher the urgency of the disease. The encoding method of the symptoms and treatment needs is a one-hot encoding method to obtain a symptom encoding vector and a treatment need encoding vector. The encoding result of the vital signs is a vital sign encoding vector composed of the index values ​​of multiple sign indicators, and whether the patient is a special population, whether there is an infectious disease, and the medical insurance information are all binary encoded;

[0023] The time format in the patient standardization rules is consistent with the time format in the medical standardization rules.

[0024] Optionally, the scoring dimensions of the priority of medical treatment include the urgency of medical treatment, the degree of abnormality of vital signs, and the priority of special populations;

[0025] Scoring the patient records in the preprocessed patient information based on the scoring dimension to obtain a scoring result of the patient in the scoring dimension;

[0026] The urgency of the medical consultation is scored as follows:

[0027]

[0028] in:

[0029] Y 1 Indicates the score result of the urgency of medical treatment, x 1 Indicates the consultation time, time indicates the current time, Indicates the preset time difference conversion coefficient, x 2 Indicates the urgency of the condition;

[0030] The scoring method for the abnormality of the vital signs is:

[0031]

[0032] in:

[0033] Y 2 Indicates the score result of the abnormality of vital signs, x 3 represents the vital signs encoding vector, represents the preset normal vital sign coding vector, weight represents the vital sign coding coefficient, ||·|| 2 represents the L2 norm, and the preset normal vital sign encoding vector is the mean of the vital sign encoding vectors of non-patients;

[0034] The scoring method for the priority dimension of special populations is as follows:

[0035] Y 3 =x 4

[0036] in:

[0037] Y 3 represents the scoring result of the priority dimension of special populations, x 4 Indicates whether it is the coding result of special population, x 4 ∈{0,1};

[0038] The scoring results of the urgency of medical treatment, the degree of abnormal vital signs, and the priority of special populations are weighted, and the weighted result is used as the medical treatment priority Y: w 1 ,w 2 ,w 3 They are the urgency of medical treatment, the degree of abnormal vital signs, and the dimension weights of priority for special populations.

[0039] Optionally, the dimension weight of the scoring dimension is calculated as follows:

[0040] Obtain the scoring results of all patients in the preprocessed patient information in terms of the urgency of medical treatment, the degree of abnormal vital signs, and the priority of special populations, calculate the information entropy of different scoring dimensions, and calculate the objective dimension weight of each scoring dimension based on the information entropy. The information entropy of the urgency of medical treatment, the degree of abnormal vital signs, and the priority of special populations are e 1 ,e 2 ,e 3 , the objective dimension weight is calculated as follows:

[0041]

[0042] in:

[0043] The weights of objective dimensions are, in order, the urgency of medical treatment, the degree of abnormal vital signs, and the priority dimension of special populations;

[0044] The judgment matrix was constructed by using the hierarchical analysis method, and consistency test and normalization were performed to obtain the subjective dimension weights of the urgency of the medical treatment, the degree of abnormal vital signs, and the priority dimension of special populations.

[0045] Based on the objective dimension weight and subjective dimension weight of the scoring dimension, the dimension weight of the scoring dimension is calculated: i∈[1,3], where α i It represents the dynamic weighting parameter of the i-th scoring dimension, and the 1st to 3rd scoring dimensions are respectively the urgency of medical treatment, the abnormality of vital signs and the priority dimension of special populations.

[0046] Optionally, the construction process of the medical resource matching objective function is:

[0047] Based on whether the patient has infectious disease coding results, medical insurance information, symptom coding vectors and treatment demand coding vectors in the pre-processed patient information, the medical resources required by the patient are extracted; the symptom coding vector is the coding result of the patient's current symptoms, a convolutional neural network model is constructed, and the convolutional neural network model is used to receive the patient's symptom coding vector, and output whether the patient needs hospitalization and the doctor's title of the required doctor. If the patient needs hospitalization, a bed and medical staff are allocated to the patient. If the patient has an infectious disease, a bed type with fewer people in the ward is tended to be allocated to the patient, otherwise a bed type with a higher vacancy rate is allocated. The doctor's title of the required doctor, whether a bed is allocated, the bed type and whether medical staff are allocated are added to the medical resources required by the patient;

[0048] The treatment demand coding vector is the coding result of the medical devices and drugs required by the patient, combined with the coding result of the medical insurance information. If the patient pays for medical insurance, the coding results of the medical devices and drugs covered by the medical insurance are selected and added to the medical resources required by the patient. Otherwise, the coding results of medical devices and drugs with the same functions as the medical devices required by the patient and with a higher idle rate are selected and added to the medical resources required by the patient.

[0049] Based on the medical resources required by patients and the medical resources dispatched to patients, a medical resource matching objective function is constructed.

[0050] Optionally, the expression of the medical resource matching objective function is:

[0051]

[0052] θ=(θ 1 ,θ 2 ,...,θ n ,...,θ N )

[0053] in:

[0054] Y(n) is the priority of the nth patient in the preprocessed patient information, x 1 (n) represents the consultation time of the nth patient, θ n To dispatch medical resources to the nth patient, β n represents the medical resources required by the nth patient, TIME(θ n ) indicates dispatching medical resources θ to the nth patient n Scheduling time, n∈[1,N], N represents the number of patients in the preprocessed patient information, and θ represents the solution result of the medical resource matching objective function.

[0055] Optionally, an improved imperial competition algorithm is used to solve the medical resource matching objective function, including:

[0056] Extract all medical resources based on the medical resource information, and reorganize the medical resources H times to obtain H groups of reorganized medical resources, each group of reorganized medical resources is a country, and the reorganized medical resources divide the extracted medical resources into N resources as medical resources dispatched to N patients;

[0057] Substitute the reorganized medical resources into the medical resource matching objective function to obtain the fitness function value of the country, select the top 20% countries with the highest fitness function value as empires, and the other countries as colonies;

[0058] Calculate the number of colonies each empire has colonized. The formula for calculating the number of colonies is: Where F is the fitness function value of the empire, and Sum is the sum of the fitness function values ​​of all empires;

[0059] The colonies are assigned to the nearest empire until the number of colonies colonized by the empire is reached. The distance between the countries is the similarity between the reorganized medical resources, and the similarity is calculated by cosine similarity.

[0060] The colonies converge to the empire to which they belong, and the convergence formula is:

[0061]

[0062] in:

[0063] g stands for colony, It indicates the result of the colony moving closer to the empire to which it belongs. G indicates the empire to which the colony g belongs, and F g represents the fitness function value of colony g, F G represents the fitness function value of Empire G, b represents the control parameter, and rand(0,1) represents a random number between 0 and 1;

[0064] After the convergence is completed, the fitness function value of the colonies colonized by the empire is calculated. If the fitness value of a colony is better than that of the empire to which it belongs, the colony and the empire will exchange positions and the colony will become the new empire;

[0065] Conduct imperial competition for all empires. The imperial competition process is as follows:

[0066] Calculate the comprehensive national strength of the empire, where the comprehensive national strength is calculated as the weighted sum of the fitness function value of the empire and the average fitness function value of the colonies it colonizes, where the weight of the fitness function value of the empire is 1 and the weight of the average fitness function value of the colonies is ε;

[0067] The empire with high comprehensive national strength selects the colonies of the empire with low comprehensive national strength, and the selected colonies move closer to the empire with high comprehensive national strength, completing an imperial competition. The probability of each colony of the empire with low comprehensive national strength being selected is the difference in comprehensive national strength / the comprehensive national strength of the empire with low comprehensive national strength;

[0068] Using the improved reform probability, reform all colonies in the empire after imperial competition, where the reform probability of a colony is P:

[0069]

[0070] in:

[0071] P 0represents the probability of basic reform, k represents the number of imperial competitions currently completed, Max represents the preset maximum number of imperial competitions, f represents the fitness function value of the colony, and max represents the maximum fitness function value of all colonies under the empire to which the colony belongs;

[0072] The reform method of the colony is to randomly perturb the corresponding reorganized medical resources of the colony;

[0073] The imperial competition and colonial reform are repeated until the preset maximum number of imperial competitions is reached, and the empire with the largest fitness function value at this time is selected as the solution result of the medical resource matching objective function.

[0074] In order to solve the above problems, the present invention provides a medical task scheduling system based on artificial intelligence, characterized in that the medical task scheduling system based on artificial intelligence includes a server and a data storage device, and the server includes a medical priority evaluation module and a medical task scheduling module:

[0075] The medical treatment priority evaluation module is used to pre-process the medical resource information and patient information, and calculate the medical treatment priority of the patient based on the pre-processed patient information;

[0076] The medical task scheduling module is used to establish a medical resource matching objective function based on the patient's priority of medical treatment, pre-processed patient information and medical resource information, use the improved imperial competition algorithm to solve the medical resource matching objective function, perform medical resource scheduling based on the solution result, and allocate medical resources to patients;

[0077] The data storage device is used to store medical resource information and patient information.

[0078] In order to solve the above problem, the present invention further provides an electronic device, the electronic device comprising:

[0079] A memory storing at least one instruction;

[0080] Communication interface, enabling electronic equipment to communicate; and

[0081] A processor executes instructions stored in the memory to implement the above-mentioned artificial intelligence-based medical task scheduling method.

[0082] In order to solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored, and the at least one instruction is executed by a processor in an electronic device to implement the above-mentioned artificial intelligence-based medical task scheduling method.

[0083] Compared with the prior art, the present invention proposes a medical task scheduling method based on artificial intelligence, which has the following advantages:

[0084] First, this scheme proposes a method for calculating the priority of medical treatment, standardizes the medical resource information and patient information, establishes the relationship between the two by coding, and extracts the urgency of medical treatment, the degree of abnormal vital signs and the priority dimensions of special populations from the patient information as the scoring dimensions of the priority of medical treatment, and introduces dynamic weighting parameters in the multi-dimensional weighted evaluation process. When the information entropy of the scoring dimension is small, it means that the difference in the scoring dimension is large. The dynamic weighting parameter tends to reduce the proportion of the subjective dimension weight and increase the proportion of the objective dimension weight. When the information entropy of the scoring dimension is large, it means that the difference in the scoring dimension is small. The dynamic weighting parameter tends to reduce the proportion of the objective dimension weight and increase the proportion of the subjective dimension weight, thereby improving the scientificity and effectiveness of the priority of medical treatment.

[0085] At the same time, this scheme proposes a medical resource scheduling method, which combines the patient's treatment priority, pre-processed patient information and medical resource information to establish a medical resource matching objective function. The medical resource matching objective function fully considers the patient's treatment priority and medical insurance information, and tends to provide patients with medical resources covered by medical insurance. The colonial convergence and imperial reform process in the imperial competition algorithm are improved, and the difference between the fitness function values ​​is used as the convergence step size, so that the reform probability changes adaptively with the number of imperial competitions to improve the diversity of colonies. The improved imperial competition algorithm is used to solve the medical resource matching objective function, and the medical resources scheduled and allocated to different patients are obtained, thereby improving the utilization rate of medical resources and the completion of medical tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 A flowchart of a medical task scheduling method based on artificial intelligence provided by one embodiment of the present invention;

[0087] Figure 2 A functional module diagram of a medical task scheduling system based on artificial intelligence provided by an embodiment of the present invention;

[0088] Figure 2 In: 100 artificial intelligence-based medical task scheduling system, 101 medical priority assessment module, 102 medical task scheduling module, 103 data storage device;

[0089] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0090] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0091] The embodiment of the present application provides a medical task scheduling method based on artificial intelligence. The execution subject of the medical task scheduling method based on artificial intelligence includes but is not limited to at least one of the electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the medical task scheduling method based on artificial intelligence can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0092] Reference Figure 1 , Embodiment 1 of the present invention is:

[0093] S1: Collect medical resource information and patient information and perform preprocessing.

[0094] The medical resource information includes bed resource information, medical equipment resource information and medical and nursing human resources information. The bed resource information includes the price and usage of beds, the medical equipment resource information includes the price and usage of different medical equipment, and the medical and nursing human resources information includes the shift schedule of medical and nursing staff in different departments and the titles of doctors.

[0095] The bed usage status includes whether the bed is vacant and the vacancy rate of different types of beds. The types of beds include ICU beds, general ward beds and specialist ward beds;

[0096] The usage of the medical device refers to whether the medical device is operating normally, the idle rate of different medical devices, and the time period during which the medical device can be used;

[0097] The doctor's titles include resident physician, attending physician, associate chief physician and chief physician.

[0098] Preprocessing the medical resource information and patient information includes:

[0099] The preprocessing process of the medical resource information is as follows:

[0100] Screening the medical resource information for abnormal information, wherein the abnormal information is a value beyond a reasonable range, including the vacancy rate of beds, the vacancy rate of medical equipment, the time period during which medical equipment can be used, and the scheduling of medical staff, and correcting the abnormal information;

[0101] Filling missing values ​​in the medical resource information to obtain cleaned medical resource information, wherein the filling method is to use the average value of similar hospitals or to estimate based on historical data;

[0102] The medical resource information after the information cleaning is standardized by using medical standardization rules to obtain the pre-processed medical resource information, wherein the medical standardization rules include a unified time format, a unique hot encoding of the medical device name, and encoding of the doctor's title, whether the bed is idle, whether the medical device is operating normally, the time period in which the medical device can be used, and the scheduling of medical staff. The encoding results of the resident physician, attending physician, deputy chief physician, and chief physician in the doctor's title are 1-4, respectively, and the encoding results of whether the bed is idle and whether the medical device is operating normally are 0-1, wherein 0 indicates that the bed is not idle and the medical device cannot operate normally, and 1 indicates that the bed is idle and the medical device can operate normally, the time period in which the medical device can be used is encoded as 1, the time period that cannot be used is encoded as 0, and the scheduling date of the medical staff is encoded as 1; specifically, the unified time format is a timestamp method;

[0103] The preprocessing process of the patient information is as follows:

[0104] Deleting patient records that are missing key fields, including patient identification information and symptoms;

[0105] Delete duplicate patient records and obtain cleansed patient information;

[0106] The patient information after the information cleaning is standardized by using the patient standardization rules to obtain the pre-processed patient information, wherein the patient standardization rules include a unified time format, and the patient's symptoms, the urgency of the disease, vital signs, treatment needs, whether the patient is a special population, whether there is an infectious disease, and medical insurance information are encoded, wherein the encoding result of the urgency of the disease is 0-5, and the higher the encoding value, the higher the urgency of the disease. The encoding method of the symptoms and treatment needs is a one-hot encoding method to obtain a symptom encoding vector and a treatment need encoding vector. The encoding result of the vital signs is a vital sign encoding vector composed of the indicator values ​​of multiple vital sign indicators. Whether the patient is a special population, whether there is an infectious disease, and medical insurance information are all binary encoded, wherein 1 represents a special population, the presence of an infectious disease, and medical insurance payment, and 0 represents not a special population, the absence of an infectious disease, and non-medical insurance payment; specifically, the vital sign indicators include blood pressure, heart rate, urine routine indicators, and blood routine indicators; the special population includes the elderly, the young, pregnant women, and the disabled;

[0107] The time format in the patient standardization rules is consistent with the time format in the medical standardization rules.

[0108] S2: Calculate the patient's priority for medical treatment based on the pre-processed patient information.

[0109] The scoring dimensions of the priority of medical treatment include the urgency of medical treatment, the degree of abnormality of vital signs and the priority of special populations;

[0110] Scoring the patient records in the preprocessed patient information based on the scoring dimension to obtain a scoring result of the patient in the scoring dimension;

[0111] The urgency of the medical consultation is scored as follows:

[0112]

[0113] in:

[0114] Y 1 Indicates the score result of the urgency of medical treatment, x 1 Indicates the consultation time, time indicates the current time, Indicates the preset time difference conversion coefficient, x 2 Indicates the urgency of the condition;

[0115] The scoring method for the abnormality of the vital signs is:

[0116]

[0117] in:

[0118] Y 2 Indicates the score result of the abnormality of vital signs, x 3 represents the vital signs encoding vector, represents the preset normal vital sign coding vector, weight represents the vital sign coding coefficient, ||·|| 2 represents the L2 norm, the preset normal vital sign encoding vector is the mean of the vital sign encoding vectors of non-patients; the higher the score result of the degree of abnormality of the vital signs, the greater the deviation between the patient's vital sign encoding vector and the normal value, and the higher the degree of abnormality of the patient's vital signs;

[0119] The scoring method for the priority dimension of special populations is as follows:

[0120] Y 3 =x 4

[0121] in:

[0122] Y 3 represents the scoring result of the priority dimension of special populations, x 4 Indicates whether it is the coding result of special population, x 4∈{0,1};

[0123] The scoring results of the urgency of medical treatment, the degree of abnormal vital signs, and the priority of special populations are weighted, and the weighted result is used as the medical treatment priority Y: w 1 ,w 2 ,w 3 They are the urgency of medical treatment, the degree of abnormal vital signs, and the dimension weights of priority for special populations.

[0124] The dimension weight calculation method of the scoring dimension is:

[0125] Obtain the scoring results of all patients in the preprocessed patient information in terms of the urgency of medical treatment, the degree of abnormal vital signs, and the priority of special populations, calculate the information entropy of different scoring dimensions, and calculate the objective dimension weight of each scoring dimension based on the information entropy. The information entropy of the urgency of medical treatment, the degree of abnormal vital signs, and the priority of special populations are e 1 ,e 2 ,e 3 , the objective dimension weight is calculated as follows:

[0126]

[0127] in:

[0128] The weights of objective dimensions are, in order, the urgency of medical treatment, the degree of abnormal vital signs, and the priority dimension of special populations;

[0129] The judgment matrix was constructed by using the hierarchical analysis method, and consistency test and normalization were performed to obtain the subjective dimension weights of the urgency of the medical treatment, the degree of abnormal vital signs, and the priority dimension of special populations.

[0130] Based on the objective dimension weight and subjective dimension weight of the scoring dimension, the dimension weight of the scoring dimension is calculated: i∈[1,3], where α i Represents the dynamic weighting parameter of the i-th scoring dimension, wherein the 1st to 3rd scoring dimensions are the urgency of medical treatment, the degree of abnormal vital signs, and the priority dimension of special populations. As a preferred embodiment of the present invention, when the information entropy of the scoring dimension is small, it means that the difference of the scoring dimension is large, and the dynamic weighting parameter tends to reduce the proportion of the subjective dimension weight and increase the proportion of the objective dimension weight. When the information entropy of the scoring dimension is large, it means that the difference of the scoring dimension is small, and the dynamic weighting parameter tends to reduce the proportion of the objective dimension weight and increase the proportion of the subjective dimension weight.

[0131] S3: Based on the patient's priority of medical treatment, pre-processed patient information and medical resource information, a medical resource matching objective function is established.

[0132] The construction process of the medical resource matching objective function is as follows:

[0133] Based on whether the patient has infectious disease coding results, medical insurance information, symptom coding vectors and treatment demand coding vectors in the pre-processed patient information, the medical resources required by the patient are extracted; the symptom coding vector is the coding result of the patient's current symptoms, a convolutional neural network model is constructed, and the convolutional neural network model is used to receive the patient's symptom coding vector, and output whether the patient needs hospitalization and the doctor's title of the required doctor. If the patient needs hospitalization, a bed and medical staff are allocated to the patient. If the patient has an infectious disease, a bed type with fewer people in the ward is tended to be allocated to the patient, otherwise a bed type with a higher vacancy rate is allocated. The doctor's title of the required doctor, whether a bed is allocated, the bed type and whether medical staff are allocated are added to the medical resources required by the patient;

[0134] The treatment demand coding vector is the coding result of the medical devices and drugs required by the patient, combined with the coding result of the medical insurance information. If the patient pays for medical insurance, the coding results of the medical devices and drugs covered by the medical insurance are selected and added to the medical resources required by the patient. Otherwise, the coding results of medical devices and drugs with the same functions as the medical devices required by the patient and with a higher idle rate are selected and added to the medical resources required by the patient.

[0135] Based on the medical resources required by patients and the medical resources dispatched to patients, a medical resource matching objective function is constructed.

[0136] The expression of the medical resource matching objective function is:

[0137]

[0138] θ=(θ 1 ,θ 2 ,...,θ n ,...,θ N )

[0139] in:

[0140] Y(n) is the priority of the nth patient in the preprocessed patient information, x 1 (n) represents the consultation time of the nth patient, θ n To dispatch medical resources to the nth patient, β n represents the medical resources required by the nth patient, TIME(θ n ) indicates dispatching medical resources θ to the nth patient nScheduling time, n∈[1,N], N represents the number of patients in the preprocessed patient information, and θ represents the solution result of the medical resource matching objective function.

[0141] S4: Use the improved imperial competition algorithm to solve the medical resource matching objective function, schedule medical resources based on the solution results, and allocate medical resources to patients.

[0142] The improved imperial competition algorithm is used to solve the medical resource matching objective function, including:

[0143] Extract all medical resources based on the medical resource information, and reorganize the medical resources H times to obtain H groups of reorganized medical resources, each group of reorganized medical resources is a country, and the reorganized medical resources divide the extracted medical resources into N resources as medical resources dispatched to N patients;

[0144] Substitute the reorganized medical resources into the medical resource matching objective function to obtain the fitness function value of the country, select the top 20% countries with the highest fitness function value as empires, and the other countries as colonies;

[0145] Calculate the number of colonies each empire has colonized. The formula for calculating the number of colonies is: Where F is the fitness function value of the empire, and Sum is the sum of the fitness function values ​​of all empires;

[0146] The colonies are assigned to the nearest empire until the number of colonies colonized by the empire is reached. The distance between the countries is the similarity between the reorganized medical resources, and the similarity is calculated by cosine similarity.

[0147] The colonies converge to the empire to which they belong, and the convergence formula is:

[0148]

[0149] in:

[0150] g stands for colony, It indicates the result of the colony moving closer to the empire to which it belongs. G indicates the empire to which the colony g belongs, and F g represents the fitness function value of colony g, F G represents the fitness function value of Empire G, b represents the control parameter, and rand(0,1) represents a random number between 0 and 1;

[0151] After the convergence is completed, the fitness function value of the colonies colonized by the empire is calculated. If the fitness value of a colony is better than that of the empire to which it belongs, the colony and the empire will exchange positions and the colony will become the new empire;

[0152] Conduct imperial competition for all empires. The imperial competition process is as follows:

[0153] Calculate the comprehensive national strength of the empire, where the comprehensive national strength is calculated as the weighted sum of the fitness function value of the empire and the average fitness function value of the colonies it colonizes, where the weight of the fitness function value of the empire is 1, and the weight of the average fitness function value of the colonies is ε; specifically, ε is set to 0.3;

[0154] The empire with high comprehensive national strength selects the colonies of the empire with low comprehensive national strength, and the selected colonies move closer to the empire with high comprehensive national strength, completing an imperial competition, wherein the probability of each colony of the empire with low comprehensive national strength being selected is the difference in comprehensive national strength / the comprehensive national strength of the empire with low comprehensive national strength; as an embodiment of the present invention, when the empire has no colonies, the empire perishes;

[0155] Using the improved reform probability, reform all colonies in the empire after imperial competition, where the reform probability of a colony is P:

[0156]

[0157] in:

[0158] P 0 represents the probability of basic reform, k represents the number of imperial competitions currently completed, Max represents the preset maximum number of imperial competitions, f represents the fitness function value of the colony, and max represents the maximum fitness function value of all colonies under the empire to which the colony belongs;

[0159] The reform method of the colony is to randomly perturb the reorganized medical resources corresponding to the colony; as an embodiment of the present invention, the reorganized medical resources corresponding to the colony are checked after the approach or random perturbation to avoid the reorganized medical resources exceeding all medical resources in the medical resource information;

[0160] The imperial competition and colonial reform are repeated until the preset maximum number of imperial competitions is reached, and the empire with the largest fitness function value at this time is selected as the solution result of the medical resource matching objective function.

[0161] Embodiment 2:

[0162] This solution conducts comparative experiments on the medical task scheduling method based on artificial intelligence, the first-come-first-served medical task scheduling method, and the priority-based medical task scheduling method, wherein the comparative experimental data is real medical resource information and patient information, and the comparative experimental results are shown in Table 1:

[0163] Table 1

[0164]

[0165] As shown in Table 1, the medical task completion rate is the completion ratio of medical tasks within the specified time, each medical task corresponds to one patient, the medical resource utilization rate is the utilization rate of bed resources, medical equipment resources and medical human resources, and the emergency task is the emergency. Overall, the artificial intelligence-based medical task scheduling method can effectively improve the completion rate of medical tasks and the utilization rate of medical resources, and avoid a large number of medical resources being idle.

[0166] Embodiment 3:

[0167] like Figure 2 , is a functional module diagram of an artificial intelligence-based medical task scheduling system 100 provided in one embodiment of the present invention, which can implement the artificial intelligence-based medical task scheduling method in Example 1.

[0168] According to the functions implemented, the medical task scheduling system 100 based on artificial intelligence may include a visit priority evaluation module 101, a medical task scheduling module 102 and a data storage device 103. The module described in the present invention may also be referred to as a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, and is stored in the memory of the electronic device.

[0169] The medical treatment priority evaluation module 101 is used to pre-process the medical resource information and the patient information, and calculate the medical treatment priority of the patient based on the pre-processed patient information;

[0170] The medical task scheduling module 102 is used to establish a medical resource matching objective function based on the patient's priority of medical treatment, pre-processed patient information and medical resource information, use an improved imperial competition algorithm to solve the medical resource matching objective function, perform medical resource scheduling based on the solution result, and allocate medical resources to patients;

[0171] The data storage device 103 is used to store medical resource information and patient information.

[0172] In detail, each module in the medical task scheduling system 100 based on artificial intelligence in the embodiment of the present invention is used in the same manner as described above. Figure 1 The same technical means as the artificial intelligence-based medical task scheduling method described in the text and can produce the same technical effects will not be repeated here.

[0173] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0174] It should be noted that the serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. And the terms "including", "comprising" or any other variants thereof in this article are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "including a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0175] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0176] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A medical task scheduling method based on artificial intelligence, characterized in that: The method comprises: S1: Collect medical resource information and patient information and perform preprocessing, wherein the preprocessing method includes information cleaning and information standardization, wherein the medical resources include bed resources, medical equipment resources and medical and nursing human resources, and the medical resource information is the usage of different medical resources, and the patient information is composed of patient records of multiple patients, wherein the patient records include the patient's identity information, consultation time, symptoms, urgency of the condition, vital signs, treatment needs, whether the patient is a special population, whether the patient has an infectious disease and medical insurance information; S2: Calculating the patient's priority for medical treatment based on the pre-processed patient information, wherein the priority for medical treatment is calculated using a multi-dimensional weighted scoring method; S3: combining the patient's priority for medical treatment, the pre-processed patient information and the medical resource information, establishing a medical resource matching objective function, wherein the medical resource matching objective function aims to maximize the matching degree between the patient and the medical resources, and the solution of the medical resource matching objective function is the medical resource scheduling result; S4: Use the improved imperial competition algorithm to solve the medical resource matching objective function, schedule medical resources based on the solution results, and allocate medical resources to patients.

2. The medical task scheduling method based on artificial intelligence according to claim 1, characterized in that: The medical resource information includes bed resource information, medical equipment resource information and medical and nursing human resources information. The bed resource information includes the price and usage of beds, the medical equipment resource information includes the price and usage of different medical equipment, and the medical and nursing human resources information includes the shift schedule of medical and nursing staff in different departments and the titles of doctors. The usage of the beds includes whether the beds are vacant and the vacancy rates of different types of beds; The usage of the medical device refers to whether the medical device is operating normally, the idle rate of different medical devices, and the time period during which the medical device can be used; The doctor's titles include resident physician, attending physician, associate chief physician and chief physician.

3. The medical task scheduling method based on artificial intelligence as claimed in claim 2, characterized in that: Preprocessing the medical resource information and patient information includes: The preprocessing process of the medical resource information is as follows: Screening the medical resource information for abnormal information, wherein the abnormal information is a value beyond a reasonable range, including the vacancy rate of beds, the vacancy rate of medical equipment, the time period during which medical equipment can be used, and the scheduling of medical staff, and correcting the abnormal information; Filling missing values ​​in the medical resource information to obtain cleaned medical resource information, wherein the filling method is to use the average value of similar hospitals or to estimate based on historical data; The medical resource information after the information cleaning is standardized by using medical standardization rules to obtain the pre-processed medical resource information, wherein the medical standardization rules include a unified time format, unique hot coding of medical device names, coding of doctor titles, whether beds are idle, whether medical devices are operating normally, time periods when medical devices can be used, and scheduling of medical staff, wherein the coding results of the doctor titles of resident physicians, attending physicians, associate chief physicians, and chief physicians are 1-4 in sequence, and the coding results of whether beds are idle and whether medical devices are operating normally are 0-1, wherein 0 indicates that beds are not idle and medical devices cannot operate normally, and 1 indicates that beds are idle and medical devices can operate normally, the time periods when medical devices can be used are coded as 1, the time periods that cannot be used are coded as 0, and the scheduling dates of medical staff are coded as 1; The preprocessing process of the patient information is as follows: Deleting patient records that are missing key fields, including patient identification information and symptoms; Delete duplicate patient records and obtain cleansed patient information; The patient information after the information cleaning is standardized by using the patient standardization rule to obtain the pre-processed patient information, wherein the patient standardization rule includes a unified time format, and the patient's symptoms, the urgency of the disease, vital signs, treatment needs, whether the patient is a special population, whether there is an infectious disease, and medical insurance information are encoded, wherein the encoding result of the urgency of the disease is 0-5, and the higher the encoding value, the higher the urgency of the disease. The encoding method of the symptoms and treatment needs is a one-hot encoding method to obtain a symptom encoding vector and a treatment need encoding vector. The encoding result of the vital signs is a vital sign encoding vector composed of the index values ​​of multiple sign indicators, and whether the patient is a special population, whether there is an infectious disease, and the medical insurance information are all binary encoded; The time format in the patient standardization rules is consistent with the time format in the medical standardization rules.

4. The medical task scheduling method based on artificial intelligence according to claim 1, characterized in that: The scoring dimensions of the priority of medical treatment include the urgency of medical treatment, the degree of abnormality of vital signs and the priority of special populations; Scoring the patient records in the preprocessed patient information based on the scoring dimension to obtain a scoring result of the patient in the scoring dimension; The urgency of the medical consultation is scored as follows: in: Y1 represents the score result of the urgency of the visit, x1 represents the visit time, and time represents the current time. It represents the preset time difference conversion coefficient, and x2 represents the urgency of the disease; The scoring method for the abnormality of the vital signs is: in: Y2 represents the score result of the abnormal degree of vital signs, x3 represents the encoding vector of vital signs, represents a preset normal vital sign coding vector, weight represents a vital sign coding coefficient, ||·||2 represents an L2 norm, and the preset normal vital sign coding vector is the mean of the vital sign coding vectors of non-patients; The scoring method for the priority dimension of special populations is as follows: Y3=x4 in: Y3 represents the scoring result of the priority dimension of special populations, and x4 represents the coding result of whether it is a special population. x4∈{0,1}； The scoring results of the urgency of medical treatment, the degree of abnormal vital signs, and the priority of special populations are weighted, and the weighted result is used as the medical treatment priority Y: w1, w2, and w3 are the dimension weights of the urgency of medical treatment, the degree of abnormal vital signs, and the priority of special populations, respectively.

5. The medical task scheduling method based on artificial intelligence as claimed in claim 4, characterized in that: The dimension weight calculation method of the scoring dimension is: The scoring results of all patients in the pre-processed patient information in terms of the urgency of medical treatment, the degree of abnormal vital signs, and the priority of special populations are obtained, and the information entropy of different scoring dimensions is calculated. The objective dimension weight of each scoring dimension is calculated based on the information entropy. The information entropy of the urgency of medical treatment, the degree of abnormal vital signs, and the priority of special populations are e1, e2, and e3 respectively. The objective dimension weight is calculated as follows: in: The weights of objective dimensions are, in order, the urgency of medical treatment, the degree of abnormal vital signs, and the priority dimension of special populations; The judgment matrix was constructed by using the hierarchical analysis method, and consistency test and normalization were performed to obtain the subjective dimension weights of the urgency of the medical treatment, the degree of abnormal vital signs, and the priority dimension of special populations. Based on the objective dimension weight and subjective dimension weight of the scoring dimension, the dimension weight of the scoring dimension is calculated: i∈[1,3], where α i It represents the dynamic weighting parameter of the i-th scoring dimension, and the 1st to 3rd scoring dimensions are respectively the urgency of medical treatment, the abnormality of vital signs and the priority dimension of special populations.

6. The medical task scheduling method based on artificial intelligence according to claim 1, characterized in that: The construction process of the medical resource matching objective function is as follows: Extract the medical resources required by the patient based on whether the patient has infectious disease coding results, medical insurance information, symptom coding vectors, and treatment demand coding vectors in the pre-processed patient information; the symptom coding vector is the coding result of the patient's current symptoms, and a convolutional neural network model is constructed. The convolutional neural network model is used to receive the patient's symptom coding vector, and output whether the patient needs hospitalization and the doctor's title of the required doctor, and the doctor's title of the required doctor, whether a bed is allocated, the bed type, and whether medical staff are allocated are added to the medical resources required by the patient; The treatment demand coding vector is the coding result of the medical devices and drugs required by the patient, combined with the coding result of the medical insurance information. If the patient pays for medical insurance, the coding results of the medical devices and drugs covered by the medical insurance are selected and added to the medical resources required by the patient. Otherwise, the coding results of medical devices and drugs with the same functions as the medical devices required by the patient and with a higher idle rate are selected and added to the medical resources required by the patient. Based on the medical resources required by patients and the medical resources dispatched to patients, a medical resource matching objective function is constructed.

7. The medical task scheduling method based on artificial intelligence as claimed in claim 6, characterized in that: The expression of the medical resource matching objective function is: θ=(θ1,θ2,...,θ n ,...,θ N ) in: Y(n) is the priority of the nth patient in the preprocessed patient information, x1(n) represents the time of the nth patient’s visit, θ n To dispatch medical resources to the nth patient, β n represents the medical resources required by the nth patient, TIME(θ n ) indicates dispatching medical resources θ to the nth patient n Scheduling time, n∈[1,N], N represents the number of patients in the preprocessed patient information, and θ represents the solution result of the medical resource matching objective function.

8. The medical task scheduling method based on artificial intelligence as claimed in claim 7, characterized in that: The improved imperial competition algorithm is used to solve the medical resource matching objective function, including: Extract all medical resources based on the medical resource information, and reorganize the medical resources H times to obtain H groups of reorganized medical resources, each group of reorganized medical resources is a country, and the reorganized medical resources divide the extracted medical resources into N resources as medical resources dispatched to N patients; Substitute the reorganized medical resources into the medical resource matching objective function to obtain the fitness function value of the country, select the top 20% countries with the highest fitness function value as empires, and the other countries as colonies; Calculate the number of colonies each empire has colonized. The formula for calculating the number of colonies is: Where F is the fitness function value of the empire, and Sum is the sum of the fitness function values ​​of all empires; The colonies are assigned to the nearest empire until the number of colonies colonized by the empire is reached. The distance between the countries is the similarity between the reorganized medical resources, and the similarity is calculated by cosine similarity. The colonies converge to the empire to which they belong, and the convergence formula is: in: g stands for colony, It indicates the result of the colony moving closer to the empire to which it belongs. G indicates the empire to which the colony g belongs, and F g represents the fitness function value of colony g, F G represents the fitness function value of Empire G, b represents the control parameter, and rand(0,1) represents a random number between 0 and 1; After the convergence is completed, the fitness function value of the colonies colonized by the empire is calculated. If the fitness value of a colony is better than that of the empire to which it belongs, the colony and the empire will exchange positions and the colony will become the new empire; Conduct imperial competition for all empires. The imperial competition process is as follows: Calculate the comprehensive national strength of the empire, where the comprehensive national strength is calculated as the weighted sum of the fitness function value of the empire and the average fitness function value of the colonies it colonizes, where the weight of the fitness function value of the empire is 1 and the weight of the average fitness function value of the colonies is ε; The empire with high comprehensive national strength selects the colonies of the empire with low comprehensive national strength, and the selected colonies move closer to the empire with high comprehensive national strength, completing an imperial competition. The probability of each colony of the empire with low comprehensive national strength being selected is the difference in comprehensive national strength / the comprehensive national strength of the empire with low comprehensive national strength; Using improved reform probability, reform all colonies in the empire after imperial competition, where the reform probability of a colony is P: in: P0 represents the basic reform probability, k represents the number of imperial competitions currently completed, Max represents the preset maximum number of imperial competitions, f represents the fitness function value of the colony, and max represents the maximum fitness function value of all colonies under the empire to which the colony belongs; The reform method of the colony is to randomly perturb the corresponding reorganized medical resources of the colony; The imperial competition and colonial reform are repeated until the preset maximum number of imperial competitions is reached, and the empire with the largest fitness function value at this time is selected as the solution result of the medical resource matching objective function.

9. A medical task scheduling system based on artificial intelligence, characterized in that: The medical task scheduling system based on artificial intelligence includes a server and a data storage device, and the server includes a medical priority evaluation module and a medical task scheduling module: The medical treatment priority evaluation module is used to pre-process the medical resource information and patient information, and calculate the medical treatment priority of the patient based on the pre-processed patient information; The medical task scheduling module is used to establish a medical resource matching objective function based on the patient's priority of medical treatment, pre-processed patient information and medical resource information, use the improved imperial competition algorithm to solve the medical resource matching objective function, perform medical resource scheduling based on the solution result, and allocate medical resources to patients; The data storage device is used to store medical resource information and patient information; To implement an artificial intelligence-based medical task scheduling method as described in any one of claims 1-9.

Images