A doctor assignment algorithm based on consultation data

Through a deep neural network model based on consultation data, the problem of unreasonable allocation of doctor resources in online consultation platforms is solved, rapid allocation and resource optimization are achieved, and user experience and total platform transaction volume are improved.

CN114334106BActive Publication Date: 2024-11-261DRUG TECH (SHANGHAI) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111510972.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-10
Publication Date
2024-11-26
Estimated Expiration
2041-12-10

AI Technical Summary

Technical Problem

The allocation of doctor resources in the existing online consultation platform is unreasonable, resulting in patients waiting to queue up, poor user experience, and the total transaction volume of the platform is low.

Method used

Using a doctor allocation algorithm based on consultation data, by obtaining historical allocation information and building a deep neural network model, analyzing the processing time of each doctor for each disease, reasonably allocating doctor resources, reducing patient queue time, improving user experience, and optimizing Utilize doctor resources.

Benefits of technology

It has achieved rapid and reasonable allocation of patients and doctors, reduced the queue time of online patients, improved user experience, reasonably planned doctor resources, and increased the total transaction volume of the platform.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114334106B_ABST
    Figure CN114334106B_ABST
Patent Text Reader

Abstract

The present invention discloses a doctor allocation system based on consultation data, which belongs to the medical field and is applied to an online consultation platform; it includes: obtaining consultation data; obtaining the allocation identification information corresponding to each doctor; constructing a deep neural network, training to obtain the allocation neural network model and outputting it; receiving a consultation request; obtaining a list of doctors available for consultation; selecting all doctors whose second average processing time is within a preset average processing time range; obtaining the order time ratio corresponding to each doctor according to the order rate and the second average processing time corresponding to each selected doctor; and allocating the doctor corresponding to the largest order time ratio to the user to be consulted. The beneficial effect of the present invention is that by analyzing the processing time of each doctor for each condition, the corresponding doctor can be reasonably and quickly allocated to each patient for diagnosis and treatment, thereby improving the patient's experience.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the medical field, and in particular to a doctor allocation algorithm based on consultation data. Background Art

[0002] With the popularization of the Internet, the medical industry has also opened online services. The traditional way of consultation is usually to go to the hospital to register or make an appointment for outpatient diagnosis. With the popularization of the Internet, users can also make an appointment online and go to the hospital to receive the doctor's diagnosis during the appointment time. The doctor and the patient communicate face to face for diagnosis, but the current diagnosis method still has great limitations, forcing patients to go to the hospital for diagnosis. In addition, due to the unreasonable allocation of online doctors for diagnosis and treatment of patients, it is difficult to make online appointments or registrations, the use of doctor resources is unreasonable, the waiting time for online consultations is long, and the user experience is poor, which leads to a low total transaction volume (Gross Merchandise Volume, GMV) of the platform. Therefore, in response to the above problems, it is urgent to design a doctor allocation algorithm based on consultation data to meet the needs of actual use. Summary of the invention

[0003] In order to solve the above technical problems, the present invention provides a doctor allocation algorithm based on consultation data.

[0004] The technical problem solved by the present invention can be achieved by adopting the following technical solutions:

[0005] The present invention provides a doctor allocation system based on consultation data, which is applied to an online consultation platform; the system is characterized by comprising:

[0006] an acquisition unit, configured to acquire historical allocation information from the online consultation platform, and acquire corresponding doctor's consultation data according to the historical allocation information, wherein the consultation data includes the first number of consultations being processed by each doctor, the second number of consultations to be processed by each doctor, the total number of consultations within a preset time period, and the corresponding total consultation duration;

[0007] a first processing unit, connected to the acquisition unit, for obtaining the saturation of the number of patients received by each doctor according to the first number of patients received and the second number of patients received, and for normalizing the total number of patients received and the total consultation time corresponding to each doctor in the consultation data to obtain a first quantitative value of the total consultation time corresponding to each doctor and a second quantitative value of the total number of patients received;

[0008] a marking unit, connected to the acquisition unit and the first processing unit, respectively, for marking the corresponding doctor with allocation identification information according to the historical allocation information, and associating the allocation identification information with the patient number saturation corresponding to the doctor, the first quantitative value, and the second quantitative value;

[0009] a training unit connected to the marking unit, used to construct a deep neural network, and train the deep neural network according to the saturation of the number of patients received, the first quantized value, the second quantized value, and the corresponding allocation identification information to obtain an allocation neural network model and output it;

[0010] a receiving unit, configured to receive a consultation request from a user to be consulted, wherein the consultation request includes a symptom corresponding to the user to be consulted, and an estimated average processing time corresponding to the symptom;

[0011] a second processing unit, connected to the receiving unit and the training unit respectively, for obtaining a list of available doctors at the current moment, and the patient saturation, the first quantized value and the second quantized value corresponding to each doctor in the list of available doctors, and inputting the obtained values ​​into the allocation neural network model to obtain an allocatable list and output it;

[0012] a third processing unit, connected to the second processing unit, for obtaining the second average processing time corresponding to each assignable doctor in the assignable list for treating the disease, selecting all the doctors whose second average processing time is within a preset average processing time range, and obtaining an average processing time series output according to the second average processing time of the selected doctors;

[0013] a fourth processing unit, connected to the third processing unit, for processing and obtaining the order rate corresponding to each doctor according to the number of consultations and the number of consultations with orders for each doctor corresponding to the disease, and obtaining the order time ratio corresponding to each doctor according to the order rate corresponding to each selected doctor and the second average processing time in the average processing time series, wherein the number of consultations with orders refers to the total number of users who place orders on the online consultation platform after consultation;

[0014] A matching unit, connected to the fourth processing unit, is used to assign the doctor corresponding to the largest order placement time ratio to the user to be consulted.

[0015] Preferably, the first processing unit obtains the patient reception saturation according to the following formula:

[0016]

[0017] in,

[0018] C d Indicates the number of first admissions that the doctor is currently handling;

[0019] C st Indicates the number of second patients to be treated by the doctor;

[0020] P c Indicates the saturation level of the number of patients received.

[0021] Preferably, the first processing unit performs deviation standardization processing on the total number of patients and the total consultation time, respectively, to obtain the first quantitative value and the second quantitative value corresponding to each doctor.

[0022] Preferably, it further comprises a database, wherein the database is used to store the first average treatment time corresponding to each of the symptoms;

[0023] The database obtains all the diseases handled by all the doctors on the online consultation platform, as well as the time taken for each treatment corresponding to each disease, and obtains the first average treatment time corresponding to each disease by the following process:

[0024]

[0025] in,

[0026] T(DIS k ) represents the treatment time corresponding to the k-th treatment of each specific disease;

[0027] P k represents the frequency corresponding to the k-th treatment time of each of the specific symptoms;

[0028] E(DIS) represents the first average treatment time corresponding to each of the symptoms.

[0029] Preferably, the preset average treatment time range is obtained according to the first average treatment time and the allowable error range corresponding to each of the symptoms, wherein the allowable error range corresponding to each of the symptoms is obtained according to the following formula:

[0030]

[0031] in,

[0032] E(DIS) represents the first average treatment time corresponding to each of the symptoms;

[0033] T(DIS m ) represents the treatment time corresponding to the mth treatment of the disease;

[0034] d represents the allowable error range;

[0035] Then the preset average processing time range is [E(DIS)-d, E(DIS)+d].

[0036] Preferably, the third processing unit uses the following formula to obtain the second average processing time corresponding to each doctor treating the disease:

[0037]

[0038] in,

[0039] a represents the number of times each doctor treated the disease;

[0040] T x Indicates the treatment time corresponding to the xth time each doctor treats the disease;

[0041] The second average processing time of each of the doctors for processing the disease is represented.

[0042] Preferably, the training unit comprises:

[0043] a preprocessing module, used for preprocessing the saturation of the number of patients received, the first quantized value and the second quantized value, and the associated allocation identification information;

[0044] a training data preparation module, connected to the preprocessing module, and used for taking the saturation of the number of patients, the first quantized value, and the second quantized value as inputs of the deep neural network, and taking the allocation identification information as output of the deep neural network;

[0045] A training module, connected to the training data preparation module, is used to train the saturation of the number of patients received, the first quantized value, the second quantized value, and the corresponding allocation identification information to obtain the allocation neural network model.

[0046] Preferably, after the allocation neural network model is trained, an adaptive learning rate adjustment function is used to optimize the obtained allocation neural network model.

[0047] Preferably, the fourth processing unit calculates the order rate corresponding to each doctor according to the number of consultations and the number of consultations and orders of the doctor corresponding to the disease by using the following formula:

[0048]

[0049] in,

[0050] C trtIndicates the number of patients each doctor has received for the disease;

[0051] C or Indicates the number of consultations ordered by each doctor corresponding to the disease;

[0052] P or Indicates the order rate corresponding to each doctor.

[0053] Preferably, the fourth processing unit uses the following formula to process the order placement time ratio corresponding to each of the selected doctors according to the order placement rate and the second average processing time respectively:

[0054]

[0055] in,

[0056] P or represents the order rate corresponding to the qth doctor;

[0057] T(D q ) z represents the second average processing time of the qth doctor corresponding to the disease;

[0058] D q Represents the order time ratio corresponding to the qth doctor.

[0059] The beneficial effects of the present invention are:

[0060] The present invention is based on the platform's early consultation data. By analyzing the processing time of each doctor for each condition, it can reasonably and quickly assign the corresponding doctor to each patient for diagnosis and treatment, greatly reducing the online patient waiting time and improving the patient experience; it solves the situation where doctors have different free time, reasonably plans doctor resources, and at the same time, through doctors diagnosing patients and placing orders for medicines, the total transaction volume of the platform can be further increased. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 The structure diagram of a doctor allocation algorithm based on medical consultation data in the present invention. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0064] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0065] The present invention provides a doctor allocation system based on consultation data, belongs to the medical field, and is applied to an online consultation platform. The online consultation platform is similar to the online consultation software in the prior art, and its specific operation mode is roughly as follows: the patient initiates a consultation request, the platform allocates a doctor to receive the patient, the doctor responds to the patient's consultation, the doctor communicates with the patient, after the consultation, the doctor prescribes medicine for the patient on the platform, the patient pays and ends the consultation request, the doctor invites the patient to evaluate the service, and the consultation ends.

[0066] like Figure 1 As shown, the above doctor allocation system specifically includes two parts: one is the training of the allocation neural network model, and the other is the allocation using the trained allocation neural network model;

[0067] The training process of the distribution neural network model is as follows:

[0068] An acquisition unit 1 is used to acquire historical allocation information from the online consultation platform, and acquire the corresponding doctor's consultation data according to the historical allocation information, wherein the consultation data includes the first number of consultations being processed by each doctor, the second number of consultations to be processed by each doctor, the total number of consultations within a preset time period, and the corresponding total consultation duration;

[0069] Specifically, historical allocation information refers to a patient who initiated a consultation request on the online consultation platform at a certain historical moment, and the platform assigned the patient to a certain doctor, that is, the number of first consultations being processed by the doctor from the beginning of the day to the allocation moment, the number of second consultations to be processed, the total number of consultations, and the corresponding total consultation time are obtained as consultation data;

[0070] The above-mentioned consultation data and historical allocation information correspond one-to-one, that is, the platform assigns patients to a certain doctor and obtains the doctor's current consultation data;

[0071] Among them, the consultation data refers to the valid consultation data, which means the consultation data with more than 4 consultation communication information, that is, the information of the doctor and the patient's question and answer must exceed 4;

[0072] The preset time period can be determined based on the actual consultation situation on the online consultation platform. Generally, the preset time period is less than one day, that is, within 24 hours;

[0073] It should be noted that in the historical allocation process, the platform uses random allocation to assign patients to doctors.

[0074] A first processing unit 2, connected to the acquisition unit 1, for obtaining the saturation of the number of patients received by each doctor according to the first number of patients received and the second number of patients received;

[0075] The first processing unit 2 obtains the saturation of the number of patients received according to the following formula (1):

[0076]

[0077] in,

[0078] C d Indicates the number of first-visit patients the doctor is currently handling;

[0079] C st Indicates the number of second-term patients to be treated by the doctor;

[0080] P c Indicates the saturation of the number of patients received;

[0081] When there are many patients on the platform, a doctor may handle multiple patients at the same time, which is the number of patients that the doctor is currently treating. d ,The second number of patients to be treated by the doctor indicates the number of patients assigned to this doctor by the platform but not treated by the doctor;

[0082] The first processing unit 2 also performs min-max normalization (Min-Max Normalization) on the total number of patients and total consultation time corresponding to each doctor in the historical allocation information, also known as deviation normalization, which is a linear transformation of the original data, to obtain the first quantized value of the total consultation time corresponding to each doctor and the second quantized value of the total number of patients, so that the quantized values ​​are mapped between [0, 1].

[0083] a marking unit 3, connected to the acquisition unit 1 and the first processing unit 2, respectively, for marking the corresponding doctor with the allocation identification information according to the historical allocation information, and associating the allocation identification information with the saturation of the number of patients corresponding to the doctor, the first quantitative value and the second quantitative value;

[0084] Specifically, since the patient admission data and the historical allocation information are in one-to-one correspondence, according to the historical allocation information, the corresponding calculated patient admission number saturation, the first quantitative value and the second quantitative value are marked with allocation identification information, and the allocation identification information is assignable, indicating that the doctor corresponding to the above-mentioned patient admission number saturation, the first quantitative value and the second quantitative value can be assigned;

[0085] A training unit 4, connected to the marking unit 3, is used to construct a deep neural network, and train the deep neural network according to the saturation of the number of patients received, the first quantized value, the second quantized value, and the corresponding allocation identification information to obtain an allocation neural network model and output it;

[0086] In a preferred embodiment, the training unit 4 comprises:

[0087] A preprocessing module, used for preprocessing the saturation of the number of patients, the first quantitative value and the second quantitative value, and the associated allocation identification information;

[0088] Specifically, multiple thresholds are pre-set, and the consultation data and the calculated saturation of the number of consultations, the first quantitative value, and the second quantitative value are compared with the corresponding thresholds respectively, and abnormal consultation data are eliminated. For example, the saturation of the number of consultations is compared with a first threshold. When the saturation of the number of consultations of a certain doctor on that day is greater than the first threshold, but the platform still assigns the patient to the doctor, the consultation data is eliminated, wherein the first threshold is approximately 90%; due to limited human energy, if the total consultation time of a certain doctor exceeds 10 hours, then this consultation data also needs to be eliminated; since each doctor is good at treating different diseases and has different levels of expertise, in general, assuming that it takes about 10 minutes to complete a patient's consultation request, each doctor works about eight hours a day, and excluding rest time and other miscellaneous time, a doctor can handle about 36 patients a day. If the total number of consultations of the doctor on that day exceeds 36, the data also needs to be eliminated; the training data is pre-processed using methods including but not limited to the above methods;

[0089] Based on the preprocessed saturation of the number of patients, the first quantitative value corresponding to the total consultation time, the second quantitative value corresponding to the total number of patients, and the corresponding allocation identification information obtained by marking, a four-dimensional data matrix can be obtained, as shown in the following matrix (2):

[0090]

[0091] in,

[0092] P tti The first quantized value representing the total consultation time corresponding to the i-th doctor;

[0093] P tci The second quantitative value representing the total number of patients received by the ith doctor;

[0094] F i Indicates the assigned identification information corresponding to the i-th doctor;

[0095] The total number from 1 to i here represents all doctors registered on the online consultation platform;

[0096] a training data preparation module, connected to the preprocessing module, for constructing a deep neural network, taking the saturation of the number of patients received, the first quantized value, and the second quantized value as inputs of the deep neural network, and assigning identification information as outputs of the deep neural network;

[0097] Among them, there are i neural nodes in the input layer of the deep neural network, the hidden layer uses 3 layers, and the output layer also has i nodes. The saturation of the number of patients, the first quantization value, and the second quantization value are used as input parameters of the deep neural network, and the allocation identification information is used as the output parameter of the deep neural network to perform neural network training;

[0098] The training module is connected to the training data preparation module and is used to train the saturation of the number of patients, the first quantitative value, the second quantitative value, and the corresponding allocation identification information to obtain an allocation neural network model that can reasonably allocate doctors to diagnose and treat corresponding diseases and speed up the diagnosis and treatment time.

[0099] Furthermore, in the above-mentioned preferred embodiment, the allocation neural network model includes an input layer, multiple hidden layers and an output layer, wherein the input layer and each hidden layer use a logical function for neural network training, the logical function is a Sigmoid function, and an adaptive learning rate adjustment function is used to optimize the obtained allocation neural network model, and the adaptive learning rate adjustment function is an AdaDelta function.

[0100] Part 2: After the neural network model is trained, use the model to assign patients to doctors

[0101] The patient initiates a consultation in the platform to inquire about a certain disease. The receiving unit 5 receives the consultation request of the user to be consulted, and obtains the average processing time expected to be required for the disease according to the inquired disease. The average processing time here is calculated based on the time spent by all doctors who have treated the disease each time when treating the disease;

[0102] Furthermore, it also includes a database for storing the first average treatment time corresponding to each disease;

[0103] The database obtains all diseases handled by all doctors on the online consultation platform, as well as the time taken for each treatment of each disease, and uses the following formula (3) to obtain the first average treatment time corresponding to each disease:

[0104]

[0105] in,

[0106] T(DISk ) represents the processing time corresponding to the kth treatment of a specific disease;

[0107] P k represents the frequency corresponding to the treatment time of the kth treatment of the specific disease;

[0108] E(DIS) represents the first average processing time corresponding to each disease, which is specifically the average processing time of all doctors in the platform for the same disease;

[0109] Specifically, in an online medical consultation platform, for a specific disease, there may be multiple doctors who have treated the specific disease. Similarly, a doctor may have treated the specific disease multiple times. Therefore, the time taken by all doctors who have treated the disease in the platform each time to treat the disease is obtained. Assuming that the time taken to treat the disease once is 5 minutes, the number of times the time taken to treat the disease is 5 minutes is found from all the times taken to treat the disease obtained above. The number of times the time taken to treat the disease is 5 minutes is divided by the total number of times all doctors treat the specific disease (if a doctor has treated the disease multiple times, the total number of times includes the number of times the doctor treats the disease). The obtained ratio is the frequency corresponding to the treatment time of the specific disease once.

[0110] It should be noted that the estimated average processing time corresponding to each disease has been calculated and processed in advance and stored in the database. When a patient consultation is received, the corresponding first average processing time is directly retrieved from the database according to the disease to be consulted in the consultation request as the estimated average processing time corresponding to the disease.

[0111] a second processing unit 6, connected to the receiving unit 5 and the training unit 4, respectively, for obtaining a list of doctors available for consultation at the current moment, wherein the list of doctors available for consultation is composed of all doctors online at the current moment (i.e. when the patient initiates a consultation request);

[0112] Specifically, the consultation data of all available doctors up to the time when the patient initiates a consultation request on the day are obtained, and the saturation of the number of patients is calculated according to the above formula (1), and the corresponding first quantized value and second quantized value are obtained according to the min-max normalization formula. The saturation of the number of patients, the first quantized value and the second quantized value corresponding to the current moment are input into the allocation neural network model to obtain an allocatable list and output it, wherein the allocatable list is composed of all doctors whose allocation identification information is obtained as the corresponding allocatable list after the allocation neural network model outputs.

[0113] According to the allocation identification information, all doctors on the platform can be divided into assignable doctors (i.e. those with the allocation identification information being assignable), and the rest are non-assignable doctors;

[0114] a third processing unit 7, connected to the second processing unit 6, for obtaining the second average processing time corresponding to each assignable doctor in the assignable list for treating the disease, selecting all doctors whose second average processing time is within a preset average processing time range, and obtaining an average processing time series output according to the second average processing time of the selected doctors;

[0115] In a preferred embodiment, the third processing unit 7 uses the following formula (4) to obtain the second average processing time corresponding to each doctor's treatment of the disease:

[0116]

[0117] in,

[0118] a represents the number of times each doctor treated a specific condition;

[0119] T x It represents the processing time corresponding to the xth time each doctor treats the disease;

[0120] The second average processing time for each doctor to handle a disease is represented, which specifically represents the average processing time for each doctor to handle the same disease multiple times.

[0121] The first average treatment time E(DIS) corresponding to each disease and the allowable error range d corresponding to each disease (the specific calculation method is shown in formula (8) below) are directly obtained from the database. The preset average treatment time range is [E(DIS)-d, E(DIS)+d]. The second average treatment time is selected from the list of assignable doctors. All doctors within [E(DIS)-d, E(DIS)+d].

[0122] Furthermore, the first average processing time is calculated based on the time taken by all doctors on the platform to treat a certain disease each time (see above for the specific calculation method). The first average processing time corresponds to a certain disease, that is, the estimated processing time corresponding to the disease, which is pre-calculated and stored in the database of the platform; and the second average processing time is calculated based on the time taken by a certain doctor on the platform to treat a certain disease each time (see above for the specific calculation method). The second average processing time corresponds to a certain doctor, that is, the average processing time corresponding to each doctor treating a certain disease. The second average processing time may be pre-calculated and stored in the platform, or may be calculated in real time when a user consults about a disease, after the assignable list is obtained through the above-mentioned assignment neural network model;

[0123] It should be noted that the advancement of medical diagnostic technology, the accumulation of experience of each doctor, or the complexity of the disease being treated will affect the time taken to treat a certain disease. Therefore, the above-mentioned first average processing time needs to be updated at regular intervals; if the second average processing time is calculated in advance, then the second average processing time also needs to be updated at regular intervals. The specific update interval can be determined based on the actual consultation and reception situation.

[0124] a fourth processing unit 8, connected to the third processing unit 7, for processing the order rate corresponding to each doctor according to the number of consultations and the number of orders for each doctor corresponding to a certain disease, and obtaining the order time ratio corresponding to each doctor according to the order rate corresponding to each selected doctor and the second average processing time, wherein the number of orders for each doctor refers to the total number of users who place orders and pay for the online consultation platform after the doctor completes the communication with the patient and prescribes medicine for the patient;

[0125] In a preferred embodiment, the fourth processing unit 8 calculates the order rate corresponding to each doctor according to the number of consultations and the number of orders received by the doctor corresponding to the disease, and uses the following formula (5):

[0126]

[0127] in,

[0128] C trt Indicates the number of patients each doctor receives for the corresponding disease;

[0129] C or Indicates the number of orders and consultations received by each doctor for the corresponding disease;

[0130] P or Indicates the order rate corresponding to each doctor.

[0131] In a preferred embodiment, the fourth processing unit 8 uses the following formula (6) to process the order placement time ratio corresponding to each doctor according to the order placement rate and the second average processing time corresponding to each selected doctor:

[0132]

[0133] in,

[0134] P or represents the order rate corresponding to the qth doctor;

[0135] T(D q ) z represents the second average treatment time of the qth doctor corresponding to the disease;

[0136] Dq Represents the order time ratio corresponding to the qth doctor.

[0137] The matching unit 9 is connected to the fourth processing unit 8 and is used to assign the doctor corresponding to the largest order placement time ratio to the user to be consulted.

[0138] Due to the unreasonable allocation of online doctors for diagnosis and treatment of patients and the unreasonable use of doctor resources, patients have to wait in line after initiating a consultation request, resulting in a poor user experience;

[0139] By adopting this technical solution, firstly, in the process of preliminary investigation, a large amount of consultation data of each doctor for each disease during the actual operation of the online consultation platform is collected, and a stable network model is obtained through neural network training. The data of the online consultation doctors are substituted into the model to obtain a list of assignable doctors. By analyzing the processing time of each doctor for each disease, a list of doctors who can handle the disease (i.e., a list of assignable doctors) is screened out within the processing error range. According to the doctor's order time ratio per unit time, the most suitable doctor is screened out from the obtained list of assignable doctors and assigned to the consulting patient. This allows each patient to be reasonably and quickly assigned to the corresponding doctor for diagnosis and treatment, greatly reducing the waiting time for online patients and improving the patient experience. It solves the problem of different free time among doctors, rationally plans doctor resources, and at the same time, has doctors diagnose and prescribe medicines for patients to further increase the platform's total transaction volume.

[0140] Furthermore, the matching unit 9 may also sort the order time ratios in descending order, and then screen out a plurality of doctors with the highest rankings, and assign them to the currently consulting patient for the patient to select.

[0141] Preferably, in the above preferred embodiment, the matching unit 9 can also obtain the personal profile of each screened doctor and information such as typical cases of historical consultations, and push them to the patient for the patient's reference when making a choice.

[0142] In a preferred embodiment, a preset average treatment time range is obtained according to the first average treatment time and the allowable error range corresponding to each symptom, wherein the allowable error range corresponding to each symptom is obtained according to the following formula (7):

[0143]

[0144] in,

[0145] E(DIS) represents the first average treatment time corresponding to each disease;

[0146] T(DIS m ) represents the treatment time corresponding to the mth treatment of the disease;

[0147] d represents the allowable error range;

[0148] The preset average processing time range is [E(DIS)-d, E(DIS)+d].

[0149] The beneficial effects of the present invention are:

[0150] The present invention is based on the platform's early consultation data. By analyzing the processing time of each doctor for each condition, it can reasonably and quickly assign the corresponding doctor to each patient for diagnosis and treatment, greatly reducing the online patient waiting time and improving the patient experience; it solves the situation where doctors have different free time, reasonably plans doctor resources, and at the same time, through doctors diagnosing patients and placing orders for medicines, the total transaction volume of the platform can be further increased.

[0151] The above description is only a preferred embodiment of the present invention, and does not limit the implementation mode and protection scope of the present invention. For those skilled in the art, it should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the description and illustrations of the present invention should be included in the protection scope of the present invention.

Claims

1. A doctor allocation system based on consultation data, applied to online consultation platform; It is characterized in that include: an acquisition unit, configured to acquire historical allocation information from the online consultation platform, and acquire corresponding doctor's consultation data according to the historical allocation information, wherein the consultation data includes the first number of consultations being processed by each doctor, the second number of consultations to be processed by each doctor, the total number of consultations within a preset time period, and the corresponding total consultation duration; a first processing unit, connected to the acquisition unit, for obtaining the saturation of the number of patients received by each doctor according to the first number of patients received and the second number of patients received, and for normalizing the total number of patients received and the total consultation time corresponding to each doctor in the consultation data to obtain a first quantitative value of the total consultation time corresponding to each doctor and a second quantitative value of the total number of patients received; a marking unit, connected to the acquisition unit and the first processing unit, respectively, for marking the corresponding doctor with allocation identification information according to the historical allocation information, and associating the allocation identification information with the patient number saturation corresponding to the doctor, the first quantitative value, and the second quantitative value; a training unit connected to the marking unit, used to construct a deep neural network, and train the deep neural network according to the saturation of the number of patients received, the first quantized value, the second quantized value, and the corresponding allocation identification information to obtain an allocation neural network model and output it; a receiving unit, configured to receive a consultation request from a user to be consulted, wherein the consultation request includes a symptom corresponding to the user to be consulted, and an estimated average processing time corresponding to the symptom; a second processing unit, connected to the receiving unit and the training unit respectively, for obtaining a list of available doctors at the current moment, and the patient saturation, the first quantized value and the second quantized value corresponding to each doctor in the list of available doctors, and inputting the obtained values ​​into the allocation neural network model to obtain an allocatable list and output it; a third processing unit, connected to the second processing unit, for obtaining the second average processing time corresponding to each assignable doctor in the assignable list for treating the disease, selecting all the doctors whose second average processing time is within a preset average processing time range, and obtaining an average processing time series output according to the second average processing time of the selected doctors; a fourth processing unit, connected to the third processing unit, for processing and obtaining the order rate corresponding to each doctor according to the number of consultations and the number of consultations with orders for each doctor corresponding to the disease, and obtaining the order time ratio corresponding to each doctor according to the order rate corresponding to each selected doctor and the second average processing time in the average processing time series, wherein the number of consultations with orders refers to the total number of users who place orders on the online consultation platform after consultation; a matching unit, connected to the fourth processing unit, configured to assign the doctor corresponding to the largest order placement time ratio to the user to be consulted; The first processing unit obtains the patient reception saturation according to the following formula: ; in, Indicates the number of first admissions that the doctor is currently handling; Indicates the number of second patients to be treated by the doctor; Indicates the saturation of the number of patients received; Also included is a database, the database is used to store the first average treatment time corresponding to each of the symptoms; The database obtains all the diseases handled by all the doctors on the online consultation platform, as well as the time taken for each treatment corresponding to each disease, and obtains the first average treatment time corresponding to each disease by the following process: ; in, n represents the number of times each of said doctors treated each specific said condition; Indicates The treatment time corresponding to each specific condition is treated; Indicates The frequency of treatment times corresponding to each of the specific symptoms; Indicates the first average treatment time corresponding to each of the symptoms; The third processing unit uses the following formula to obtain the second average processing time corresponding to each doctor processing the disease: ; in, Indicates the number of times each of the doctors treated the condition; Indicates that each of the doctors The treatment time corresponding to the treatment of the disease for the first time; The second average processing time of each of the doctors for processing the disease is represented.

2. The doctor allocation system based on consultation data according to claim 1, It is characterized in that The first processing unit performs deviation standardization processing on the total number of patients and the total consultation time, respectively, to obtain the first quantitative value and the second quantitative value corresponding to each doctor.

3. The doctor allocation system based on consultation data according to claim 1, It is characterized in that The preset average treatment time range is obtained according to the first average treatment time and the allowable error range corresponding to each of the symptoms, wherein the allowable error range corresponding to each of the symptoms is obtained according to the following formula: ; in, Indicates the first average treatment time corresponding to each of the symptoms; Indicates The treatment time corresponding to the treatment of the disease; Indicates the allowable error range; Then the preset average processing time range is [E(DIS)-d, E(DIS)+d].

4. The doctor allocation system based on consultation data according to claim 1, It is characterized in that The training unit comprises: a preprocessing module, used for preprocessing the saturation of the number of patients received, the first quantized value and the second quantized value, and the associated allocation identification information; a training data preparation module, connected to the preprocessing module, and used for taking the saturation of the number of patients, the first quantized value, and the second quantized value as inputs of the deep neural network, and taking the allocation identification information as output of the deep neural network; A training module, connected to the training data preparation module, is used to train the saturation of the number of patients received, the first quantized value, the second quantized value, and the corresponding allocation identification information to obtain the allocation neural network model.

5. The doctor allocation system based on consultation data according to claim 1, It is characterized in that After the allocation neural network model is trained, an adaptive learning rate adjustment function is used to optimize the obtained allocation neural network model.

6. The doctor allocation system based on consultation data according to claim 1, It is characterized in that The fourth processing unit calculates the order rate corresponding to each doctor according to the number of consultations and the number of orders for the doctor corresponding to the disease by using the following formula: ; in, Indicates the number of patients each doctor has received for the disease; Indicates the number of consultations ordered by each doctor corresponding to the disease; Indicates the order rate corresponding to each doctor.

7. The doctor allocation system based on consultation data according to claim 1, It is characterized in that The fourth processing unit obtains the order placement time ratio corresponding to each of the selected doctors according to the order placement rate and the second average processing time respectively, by using the following formula: ; in, represents the order rate corresponding to the qth doctor; represents the second average processing time of the qth doctor corresponding to the disease; Represents the order time ratio corresponding to the qth doctor.