Queueing and calling methods, systems, and computer-readable storage media at payment counters

Abstract

This invention relates to the field of queuing technology at payment counters, specifically to a queuing and calling method, system, and computer-readable storage medium for payment counters. The invention calculates the window and time with the shortest waiting time based on the user's selected service type and business type. Simultaneously, it calculates the travel time from the user to the target window on each floor and recommends the floor with the shortest estimated waiting time to the user, thereby reducing queuing time. The queuing time calculation method of this invention is more accurate, effectively improving hospital queuing efficiency and enhancing the patient experience.

Description

Technical Field

[0001] This invention relates to the field of intelligent queuing technology, specifically to a queuing and calling method, system, and computer-readable storage medium at a payment counter. Background Technology

[0002] The existing queuing and calling system at payment counters often relies on historical statistical data for queue time calculations, which is not precise enough. The types of windows are generally fixed, as are the types of services and services that can be handled, making it impossible to dynamically adjust the types of services and services offered at each window based on actual circumstances. All services generally have the same priority, making it difficult to provide priority calling services for specific services. Users are often assigned a fixed floor for their queue, meaning they can generally only take a number for their current floor and are not recommended to other floors to save queuing time. This results in low queuing efficiency, long waiting times, and a poor patient experience in the current medical system. Summary of the Invention

[0003] In view of the above problems, embodiments of the present invention provide a queuing and calling method, system and computer-readable storage medium at the payment counter, which solves the problems of low queuing efficiency, long queuing time and poor medical experience in the existing medical system.

[0004] According to one aspect of the present invention, a queuing and calling method for a payment counter is provided, the method comprising at least:

[0005] Obtain the user's medical record number, and select the business type and service type based on the medical record number;

[0006] Based on the business type and service type, query the window type at the payment counter and obtain the matching window.

[0007] Based on the matching window, the queuing time of each matching window is obtained through the queuing algorithm model, and the matching window with the shortest queuing time for each floor is determined as the target window.

[0008] Each floor's target window is provided for the user to select;

[0009] The step of obtaining the queuing time for each matching window through the queuing algorithm model includes: calculating the queuing time for each matching window by summing the business processing time and the service pause time.

[0010] In one alternative approach, the queuing algorithm model specifically obtains the queuing time using the following formula:

[0011] T = T handle +T suspend ·P,

[0012] Among them, T handle For business processing time, Tsuspend Let P be the service downtime and P be the probability of service downtime.

[0013] In one alternative approach, calculating the business processing time further includes:

[0014] Based on the system logs, obtain the business priority of each matching window;

[0015] When calculating the service processing waiting time, the user is sorted according to the service priority order or placed last in the same service priority order, and the service processing time is calculated.

[0016] In an alternative approach, the processing time is calculated using the number of people in the queue of each service in the prior queue of the matching window, specifically using the following formula:

[0017]

[0018] in, This refers to the time it takes for a salesperson at the matching window to process the j-th user for the i-th type of service, where i = 1, 2, ..., N, and N represents the number of service types in the queue ahead of that user; j = 1, 2, ..., M. i M i It is the number of users of the i-th type of service, the It can be obtained through the following formula: in, It is the historical average time it takes for a salesperson to handle the i-th type of business. Obtained through system log statistics; It is the calibration value for the time when the salesperson processes the j-th user of the i-th category. Obtained through a BP neural network model.

[0019] The Obtained through a BP neural network model, specifically including:

[0020] Using the user's medical record number, the user's age, the province / city where the user's medical insurance is located, and the user's expense items are obtained, and the user's age, the province / city where the user's medical insurance is located, and the user's expense items are input into the input layer of the BP neural network model;

[0021] The hidden and output layers of the BP neural network model use the Leaky ReLU activation function to calculate and output...

[0022] The BP neural network model is obtained through the following steps:

[0023] Define the error function in Z is the target value for calibration.y The calculated value is the calibration value;

[0024] The system log records x1, x2, x3, As training samples, x1 represents the user's age, x2 represents the province / city where the user's medical insurance is located, and x3 represents the user's expense items; get, The actual time a salesperson spends processing any user's business in the i-th type of business. The historical average time for a salesperson to process the i-th type of business;

[0025] The weights from the hidden layer to the output layer and from the input layer to the hidden layer are updated using the gradient descent algorithm to obtain the BP neural network model.

[0026] In an alternative approach, the processing time is calculated using the number of people in the queue of each service in the prior queue of the matching window, specifically using the following formula:

[0027]

[0028] in, i = 1, 2, ..., N, j = 1, 2, ..., M i , is the time it takes for the window clerk to process the j-th user of the i-th type of service; N refers to the number of service types in the queue ahead of that user; M i It represents the number of users for the i-th type of service, and each previous item... All of this can be found in the system logs. For accurate prediction... It combines reference values ​​and calibration values:

[0029]

[0030] in, It is the baseline value of the time for the matching window salesperson to process the i-th type of business. This time is the historical average time for the matching window salesperson to process the i-th type of business, which can be obtained from the system log statistics. It is the calibration value for the processing time of the j-th user of the i-th type by the matching window salesperson. It refers to the deviation between the actual value and the benchmark value of the business processing time caused by various factors.

[0031]

[0032] In order to obtain This patent uses a BP neural network model. This BP neural network model consists of a total of 3 layers of neural networks.

[0033] The first layer is the input layer, consisting of 3 units: user age, the province / city where the user's medical insurance is located, and the user's expense items. The hidden layer has 7 units, and the output layer is the calibration value for the business processing time.

[0034] Hidden layer unit h j The calculation expression is:

[0035]

[0036] Where h j j = 1, 2, ..., 7, are hidden layer units; x i , i = 1, 2, 3, are input layer units; i = 1, 2, 3, j = 1, 2, ..., 7, where i is the weight from the input layer to the hidden layer; threshold b1 = 1, and the weight of b1 is... j = 1, 2, ..., 7.

[0037] Hidden layer unit h j Output Z j :

[0038]

[0039] Among them, Z j ,j=1,2,…,7, represents the output after activation of the hidden layer. The activation function of the hidden layer is the LeakyReLU function, where the coefficient a1=0.5.

[0040] Hidden layer output Z j The expression for calculating the value of the output layer unit y is:

[0041]

[0042] Among them W j 2 j = 1, 2, ..., 7, where j is the weight from the hidden layer to the output layer, and the threshold b2 = 1. The weight of b2 is...

[0043] Output Z of output layer unit y y :

[0044]

[0045] Among them, Z yThe output after activation of the output layer is the final calibration value. The activation function for the hidden layer is Leaky ReLU, with coefficient a² = 0.5. Both the hidden and output layers use Leaky ReLU activation. This is because Leaky ReLU is a linear function, much faster than sigmoid and tanh, making it suitable for scenarios with high real-time requirements in window queuing, as described in this patent. Secondly, the calibration values ​​in this patent can be positive or negative. Finally, compared to the ReLU function, it solves the Dead ReLU problem, meaning that Leaky ReLU remains effective even when the input is negative.

[0046] Define the error function

[0047] in Z is the target value for calibration. y The calculated value is the calibration value. Historical values ​​will be used. As training samples, x1, x2, and x3 refer to the user's age, the province / city where the user's medical insurance is located, and the user's expense items, respectively, which can be queried through the system logs. For calibration value, The actual time a salesperson spends processing the user's business for the i-th type of business. The historical average time for a salesperson to handle the i-th type of business can be obtained through system log queries and statistics.

[0048] The weights from the hidden layer to the output layer and from the input layer to the hidden layer are updated using the gradient descent method.

[0049]

[0050]

[0051] Where η is the learning rate, which is set to 0.5 here.

[0052] In one alternative approach, the average pause time T of the current salesperson suspend可 The expected probability of a salesperson pausing service before reaching the patient, obtained through system log statistics, is calculated using the following steps:

[0053] Based on the system logs, the frequency f(t) of the current salesperson pausing service in the matching window during various time periods is calculated, where t∈t. k ,k=1…K, where t k This represents the k-th time interval, where K represents the number of time intervals. The pause period for the salesperson, that is, the pause period for the salesperson during... The expected probability of pausing previously was 1.

[0054] If we obtain the service duration T0 of the current salesperson since the last time they started providing service, then the service duration of the salesperson when they are assigned to that user will be T0+T. handle The expected probability P of the salesperson pausing service before reaching the user is:

[0055]

[0056] in This indicates the floor function. j is the time interval number corresponding to the current user's current time. Where % is the remainder function, t j,begin and t j,end They are t j The start and end times of a time period.

[0057] In one alternative approach, the expected probability of the salesperson's service being suspended before the patient is reached is calculated through the following steps:

[0058] Based on the system logs, the frequency f(t) of the current salesperson pausing service in the matching window during various time periods is calculated, where t∈t. k k = 1…K, can be obtained through system log statistics, where t k This represents the k-th time interval, where K represents the number of time intervals. This can be understood as a pause period for the salesperson, that is, the salesperson during... The expected probability of pausing previously was 1.

[0059] If we obtain the service duration T0 of the current salesperson since the last time they started providing service, then the service duration of the salesperson when they are assigned to that user will be T0+T. handle The expected probability P of the salesperson pausing service before reaching the user is:

[0060]

[0061] in This represents the floor function. j refers to the time interval number corresponding to the user's current time. Where % is the remainder function, t j,begin and t j,end They are t j The start and end times of a time period.

[0062] In one alternative approach, the service downtime is obtained by using the average service downtime as statistically analyzed from system logs.

[0063] In one alternative approach, after determining the matching window with the shortest queuing time for each floor as the target window, the method further includes:

[0064] The user's floor can be confirmed by scanning information through the self-service machine at the payment counter or the mobile app;

[0065] Based on the user's floor, query the travel statistics table of each floor's payment counter to obtain the travel time required for the user to reach the target window on each floor, and provide the travel time to the user.

[0066] The travel statistics table for each floor's payment counter includes the time it takes to travel from any payment counter on one floor to payment counters on other floors.

[0067] In one optional approach, obtaining the user's medical record number and selecting the business type and service type based on the medical record number includes:

[0068] The system receives the user's medical record number, service type, and business type through the self-service machine at the payment counter or via a mobile app.

[0069] Query user information and fee information for medical record numbers, and determine whether the user has selected the correct service type;

[0070] If so, then match the window type according to the business type and service type;

[0071] Otherwise, if the business type or service type is selected incorrectly, a prompt will appear indicating that you should select it again.

[0072] The business types include outpatient and inpatient types. The outpatient type specifically includes payment, refund, receipt printing, and special preferential groups. The inpatient type specifically includes admission, discharge, and special preferential groups. The window type includes business type priority and service type. The business type priority is determined by each window staff member or through the system. The service type includes one or more of Mandarin, local language, and English.

[0073] In one alternative approach, the window type of the payment counter refers to the types of business that can be handled at the window, the business priority, and the types of services that can be provided, all of which can be selected by the salesperson; and / or, the various window types of the payment counter are adjusted in real time by the administrator based on the actual queuing situation at each window.

[0074] According to another aspect of the present invention, a queuing and calling system for payment counters is provided, the system comprising:

[0075] The information acquisition module is used to acquire the user's medical record number and select the business type and service type based on the medical record number;

[0076] The information matching module is used to query the window type at the payment counter based on the business type and service type, and obtain the matching window.

[0077] The information processing module is used to obtain the queuing time of each matching window according to the matching window through a queuing algorithm model, and determine the matching window with the shortest queuing time for each floor as the target window; and provide the target window for each floor to the user for selection; wherein, obtaining the queuing time of each matching window through the queuing algorithm model includes: calculating the queuing time of each matching window by the sum of the business processing time and the service pause time.

[0078] According to another aspect of the present invention, a computer-readable storage medium is provided, the storage medium storing at least one executable instruction, which, when executed on a toll booth queuing system as described above, causes the toll booth queuing system to perform the operation of a toll booth queuing method as described above.

[0079] This invention provides a queuing and calling method, system, and computer-readable storage medium for payment counters. Its advantages include: Firstly, it combines historical data-based baseline values ​​and neural network-based calibration values ​​to calculate processing time, taking into account staff downtime, thus accurately calculating the user's estimated queuing time. Secondly, it matches service types and service types to windows and prioritizes services to calculate the window with the shortest waiting time. Thirdly, it calculates the travel time from the user to windows on each floor and recommends the target window and estimated waiting time to the user, thereby reducing queuing time. This invention provides more accurate queuing time calculations, effectively improving hospital queuing efficiency and enhancing the patient experience.

[0080] The above description is merely an overview of the technical solutions of the embodiments of the present invention. In order to better understand the technical means of the embodiments of the present invention and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0081] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0082] Figure 1 A flowchart illustrating the queuing and calling method at the payment counter provided by the present invention is shown.

[0083] Figure 2 A schematic diagram of the queuing and calling system at the payment counter provided by the present invention is shown;

[0084] Figure 3 This invention provides a schematic diagram of the hardware system for queuing and calling numbers at payment counters.

[0085] Figure 4 A schematic diagram of the BP neural network model provided by this invention is shown. Figure 1 ;

[0086] Figure 5 A schematic diagram of the BP neural network model provided by this invention is shown. Figure 2 ;

[0087] Figure 6 This diagram illustrates the expected probability distribution of when a salesperson's service is suspended before the patient is reached, as provided by the present invention. Detailed Implementation

[0088] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0089] Example 1,

[0090] See Figure 1 According to one aspect of the present invention, a queuing and calling method for a payment counter is provided, the method comprising at least:

[0091] Step 101 involves obtaining the user's medical record number and selecting the service type and business type based on that number. This includes: receiving the user's medical record number, service type, and business type via a self-service payment machine or mobile app; querying the user's information and fees related to the medical record number; and determining if the selected service type is correct. If correct, matching the window type according to the selected service type; otherwise, indicating an incorrect service type selection and prompting the user to reselect. Service types include outpatient and inpatient types. Outpatient types specifically include payment, refund, receipt printing, and special preferential treatment groups. Inpatient types specifically include admission, discharge, and special preferential treatment groups. User information includes the user's name, address, phone number, age, past medical records, and preferential treatment type. User fee information refers to the details of items to be paid or already paid. Special preferential treatment groups include retired cadres, military personnel, seniors over 65 years old, disabled persons, and critically ill patients. Window types include business type priority and service type. Business type priority is determined by each window clerk or through the system. Service type includes one or more of Mandarin, local language, and English.

[0092] 102. Based on the business type and service type, query the window type at the payment counter and obtain a matching window; in step 102, after obtaining the business type and service type, query the system logs of each floor's payment counter queuing system; select the matching window that matches the business type and service type of the user's current transaction as the matching window. A matching window refers to a window that satisfies the business type and service type selected by the patient.

[0093] 103. Based on the matching windows, the queuing time of each matching window is obtained through a queuing algorithm model, and the matching window with the shortest queuing time for each floor is determined as the target window. In step 103, obtaining the queuing time of each matching window through the queuing algorithm model includes: calculating the queuing time of each matching window by summing the business processing waiting time and the service pause time. This application calculates the shortest window for each floor after obtaining the matching windows and provides it to the user for selection. The target window for each floor mentioned here can be a target window for different floors in different buildings within the hospital; and there is only one target window for each floor.

[0094] 104. Provide the target windows for each floor for the user to select. In step 104, after obtaining the target windows for each floor using this method, send them to the user via the self-service machine at the payment counter or a mobile app, and the user confirms by clicking. Providing multiple target windows for different floors can avoid increasing travel time for users unfamiliar with the hospital and prevent them from missing their appointment.

[0095] This invention calculates the window and time with the shortest waiting time based on service type and priority; and recommends the target window and estimated waiting time to the user, thereby reducing queuing time. The queuing time calculation method of this invention is more accurate, effectively improving hospital queuing efficiency and enhancing the patient experience.

[0096] In one alternative approach, a large number of training samples are first obtained from system logs. x1, x2, and x3 refer to the user's age, province, and cost item, respectively, which can be queried through the system logs. For calibration value, The actual time a salesperson spends processing the user's business for the i-th type of business. The historical average time for a salesperson to process the i-th type of business. and All of these can be obtained through system log queries and statistics. The BP neural network model is trained using training samples, and the weight parameters are updated to obtain the trained BP neural network model.

[0097] See Figure 4-5 The queuing algorithm model specifically uses the following formula to obtain the queuing time: T = T handle +Tsuspend ·P, where T handle T represents the waiting time for business processing. suspend Let P be the service pause time, and P be the expected probability that the service provider will pause service before the patient is reached. In this embodiment, after the user takes a number and selects the service type, for a given floor, the system checks each window to see if the selected service is available. If so, the user is added to the queue for that window. For the inserted window, the queue is sorted according to the priority of the service from highest to lowest. The user is placed at the end of the queue with the same priority. The waiting time T for the user at that window is calculated using a queuing time algorithm, as follows: T = T handle +T suspend ·P.

[0098] In one alternative approach, the service processing wait time is calculated using the number of people in the queue of each service that is already in the matching window, specifically using the following formula: in, i = 1, 2, ..., N, j = 1, 2, ..., M i , is the time it takes for the window clerk to process the j-th user of the i-th type of service; N refers to the number of service types in the queue ahead of that user; M i It represents the number of users for the i-th type of service, and each previous item... All of this can be found in the system logs. For accurate prediction... It combines reference values ​​and calibration values: in, It is the baseline value of the time for the matching window salesperson to process the i-th type of business. This time is the historical average time for the matching window salesperson to process the i-th type of business, which can be obtained from the system log statistics. This is a calibration value for the processing time of the j-th user in the i-th category by the matching window salesperson. A calibration model based on a backpropagation neural network is used to obtain the baseline value for the processing time, which is determined by the salesperson and the business type. Based on this, the calibration model is used to correct the processing time, improving the accuracy of time prediction. Furthermore, calculating the processing time may also include: querying system logs based on the business type to obtain the business priority of each window; when calculating the processing time, sorting the user according to the business priority order or placing them last in the same business priority order, and calculating the processing waiting time; inserting the user's business priority before other users with lower priorities or after other users with higher priorities, and then calculating the user's processing time.

[0099] In one alternative approach, the processing time is calculated by matching the number of people in the queue for each service that is already in the queue, specifically using the following formula:

[0100]

[0101] in, This refers to the time it takes for a salesperson at the matching window to process the j-th user for the i-th type of service, where i = 1, 2, ..., N, and N represents the number of service types in the queue ahead of that user; j = 1, 2, ..., M. i M i It represents the number of users of the i-th type of service. It can be obtained through the following formula: in, It is the historical average time it takes for a salesperson to handle the i-th type of business. Obtained through system log statistics; It is the calibration value for the time when the salesperson processes the j-th user of the i-th category. Obtained through a BP neural network model.

[0102] Obtained through a BP neural network model, specifically including:

[0103] The system obtains the user's age, the province / city where the user's medical insurance is located, and the user's expense items through the user's medical record number, and then inputs the user's age, the province / city where the user's medical insurance is located, and the user's expense items into the input layer of the BP neural network model.

[0104] The hidden and output layers of the BP neural network model use the Leaky ReLU activation function to compute and output.

[0105] The BP neural network model is obtained through the following steps:

[0106] Define the error function in Z is the target value for calibration. y The calculated value is the calibration value;

[0107] The system log records x1, x2, x3, As training samples, x1 represents the user's age, x2 represents the province / city where the user's medical insurance is located, and x3 represents the user's expense items; get, The actual time a salesperson spends processing any user's business in the i-th type of business. The historical average time for a salesperson to process the i-th type of business;

[0108] The weights from the hidden layer to the output layer and from the input layer to the hidden layer are updated using the gradient descent algorithm to obtain the BP neural network model.

[0109] In one alternative approach, the processing time is calculated by matching the number of people in the queue for each service that is already in the queue, specifically using the following formula:

[0110] in, i = 1, 2, ..., N, j = 1, 2, ..., M i , is the time it takes for the window clerk to process the j-th user of the i-th type of service; N refers to the number of service types in the queue ahead of that user; M i It represents the number of users for the i-th type of service, and each previous item... All can be

[0111] Found in the system logs. For accurate predictions... It combines reference values ​​and calibration values:

[0112] in, It is the baseline value of the time for the matching window salesperson to process the i-th type of business. This time is the historical average time for the matching window salesperson to process the i-th type of business, which can be obtained from the system log statistics. It is the calibration value for the processing time of the j-th user of the i-th type by the matching window salesperson. It refers to the deviation between the actual value and the benchmark value of the business processing time caused by various factors.

[0113]

[0114] In order to obtain This patent employs a BP neural network model. See also Figure 4-5 This BP neural network model consists of a total of 3 layers of neural networks.

[0115] The first layer is the input layer, consisting of 3 units: user age, the province / city where the user's medical insurance is located, and the user's expense items. The hidden layer has 7 units, and the output layer is the calibration value for the business processing time.

[0116] Hidden layer unit h j The calculation expression is:

[0117]

[0118] Where h j j = 1, 2, ..., 7, are hidden layer units; x i , i = 1, 2, 3, are input layer units; i = 1, 2, 3, j = 1, 2, ..., 7, where i is the weight from the input layer to the hidden layer; threshold b1 = 1, and the weight of b1 is... j = 1, 2, ..., 7.

[0119] Hidden layer unit h j Output Z j :

[0120]

[0121] Among them, Z j ,j=1,2,…,7, represents the output after activation of the hidden layer. The activation function of the hidden layer is the LeakyReLU function, where the coefficient a1=0.5.

[0122] Hidden layer output Z j The expression for calculating the value of the output layer unit y is:

[0123]

[0124] Among them W j 2 j = 1, 2, ..., 7, where j is the weight from the hidden layer to the output layer, and the threshold b2 = 1. The weight of b2 is...

[0125] Output Z of output layer unit y y :

[0126]

[0127] Among them, Z y The output after activation of the output layer is the final calibration value. The activation function for the hidden layer is Leaky ReLU, with coefficient a² = 0.5. Both the hidden and output layers use Leaky ReLU activation. This is because Leaky ReLU is a linear function, much faster than sigmoid and tanh, making it suitable for scenarios with high real-time requirements in window queuing, as described in this patent. Secondly, the calibration values ​​in this patent can be positive or negative. Finally, compared to the ReLU function, it solves the Dead ReLU problem, meaning that Leaky ReLU remains effective even when the input is negative.

[0128] Define the error function

[0129] in Z is the target value for calibration. y The calculated value is the calibration value. Historical values ​​will be used. As training samples, x1, x2, and x3 refer to the user's age, the province / city where the user's medical insurance is located, and the user's expense items, respectively, which can be queried through the system logs. For calibration value, The actual time a salesperson spends processing the user's business for the i-th type of business. The historical average time for a salesperson to handle the i-th type of business can be obtained through system log queries and statistics.

[0130] The weights from the hidden layer to the output layer and from the input layer to the hidden layer are updated using the gradient descent method.

[0131]

[0132]

[0133] Where η is the learning rate, which is set to 0.5 here.

[0134] In one specific embodiment, T is the time point at which processing of this user begins (i.e., the total time spent processing previous users). handle_N To determine the processing time for this user's service, we can now calculate the user's waiting time T at this window.

[0135] Similarly, calculate the waiting time for the user at other service windows where the user's business can be handled, and take the minimum waiting time at all windows to obtain the estimated waiting time for that floor.

[0136] Similarly, the estimated waiting time for each floor is calculated, and the floor with the shortest waiting time is recommended to the user. For example, a hospital has three floors with payment windows. The payment area on the first floor has nine windows. The cashier at window number 1 is configured to handle all outpatient services, including outpatient payments, refunds, receipt printing, special priority groups, and other services. The priority for special priority groups is set to the highest (1), the priority for refunds is set to 2, and the priority for payments, receipt printing, and other services is set to the lowest (3). If a user takes a refund number for an outpatient service, and there are still two people in the special priority group, two people in the refund queue, and four people in the payment queue at window number 1, then the user's queue number will be inserted at the end of the queue for the two people in the refund queue (the end of the queue with the same priority) and in front of the four people in the payment queue, i.e., fifth place. This is obtained through system log statistics. The user's age, the province / city where their medical insurance is located, and the expense items are input into the trained BP neural network to calculate the calibration value. Adding the two together gives and through The business processing time is calculated.

[0137] The expected probability of a salesperson pausing service before reaching the patient is calculated using the following steps: Based on system logs, the frequency f(t) of the current salesperson pausing service in each time period within the matching window is calculated, where t∈t. i ,i=1…K, where t k Let K represent the k-th time period, and K represent the number of time periods. If we obtain the service duration T0 of the current salesperson since the last time they started providing service, then the service duration of the salesperson when they are assigned to that user is T0+T. handle The expected probability P of the salesperson pausing service before reaching the user is: in This represents the floor function. j refers to the time interval number corresponding to the user's current time. Where % is the remainder function, t j,begin and t j,end They are t j The start and end times of the time period. Service pause time is obtained from the average service pause time calculated from system logs. In this embodiment, the service pause time is calculated using the formula T. suspend ·P, where T suspend Let be the average downtime of the salesperson, which can be obtained through system log statistics; P is the probability of the salesperson pausing service. Here, frequency is used instead of probability, and the frequency f(t), t∈t, of the salesperson pausing service in each fixed time period is statistically calculated from the system logs. i i = 1…K, t i Let K represent the i-th time period, and K represent the number of time periods. Figure 6 As shown, f(t) is a piecewise function that approximates a normal distribution function. According to probability theory, events such as a salesperson pausing service generally follow a normal distribution. The service duration T0+T when the user is scheduled is... handle Substitution This gives us the probability P of service suspension. Then, based on T... suspend The service interruption time is calculated using P. Assume the average interval T between service interruptions for this salesperson is obtained from system log statistics. suspend The service interruption period is divided into 12 segments, each lasting 10 minutes, which constitutes the interruption cycle. Suppose the salesperson suspends service during t1…t 12 The frequencies are 1 / 80, 1 / 20, 1 / 16, 1 / 12, 1 / 8, 1 / 6, 1 / 6, 1 / 8, 1 / 12, 1 / 16, 1 / 20, 1 / 80. If, when the customer arrives, the salesperson's service time has already reached 55 minutes, i.e., T0+T... handle =55, then t j =t6,

[0138] In one optional approach, after selecting the window with the shortest queuing time on each floor as the target window, the process further includes: obtaining the user's current floor from the self-service machine via a mobile app; querying the travel statistics table for each floor's payment counter to obtain the travel time required for the user to reach the target window on each floor, and providing this travel time to the user; wherein the travel statistics table for each floor's payment counter includes the travel time from any floor's payment counter to other floors' payment counters. The user's waiting time at other windows can be calculated separately, and finally, for each window (excluding those not handling the specific service), the minimum waiting time is taken to obtain the estimated waiting time for that floor. Similarly, the estimated waiting time for each floor can be calculated and displayed on the self-service machine and mini-program ticket retrieval interfaces, recommending the floor with the shortest waiting time to the user, thereby reducing the user's waiting time. After selecting a floor, the user will enter the waiting queue of the window with the shortest waiting time on that floor. When the target window calls the user, the user will be removed from the target window's waiting queue and enter the completion queue. If a user doesn't arrive at the window in time, the cashier can let them pass through the queue, and the user will be called from there when they return. If a user takes the wrong number, the cashier can cancel it, and the user will be placed in the cancellation queue and cannot be called again. If the cashier needs to leave the window for any reason, they can change the window status to "Service Paused" and then change it back to "Service Resumed" upon returning. The window status will be displayed on the call screen. There are three calling methods. The most common method is sequential calling, which calls the first user in the waiting queue. In special cases, selective calling can be used, where a user in the queue is selected for a call. Search calling can also be used, where a user is searched for by medical record number and then called.

[0139] In one optional method, the user's medical record number is obtained, and the service type is selected based on the medical record number. This includes: receiving the user's medical record number and service type through a self-service machine at the payment counter or a mobile app; querying the user's information and fee information for the medical record number, and determining whether the user's selected service type is correct; if so, matching the window type according to the selected service type; otherwise, if the service type is incorrectly selected, prompting the user to select again. Service types include outpatient and inpatient types. Outpatient types specifically include payment, refund, receipt printing, and specific preferential groups; inpatient types specifically include admission, discharge, and specific preferential groups. User information includes the user's name, address, phone number, age, completed medical record, and preferential type; user fee information refers to the details of items to be paid or already paid. Specific preferential groups include retired cadres, military personnel, seniors over 65 years old, disabled persons, and critically ill patients. Querying the user's information and fee information for the medical record number involves querying the user's information through the hospital database. If the user is registering for the first time, basic user information must be filled in and stored in the hospital database.

[0140] In one optional approach, the service types and priorities of each window at the payment counter are selected and confirmed by the clerk; and / or, the service types and priorities of each window at the payment counter are adjusted in real time by the administrator based on the queuing situation for each service type. Specifically, the window clerk can set the parameters of the current window through the window queuing software, including the services that can be processed and the priority of each service. The administrator can set the parameters of each window through the back-end management system. The parameters are the service type and priority of the window. In this embodiment, each window can handle multiple services, and the service type is selected and confirmed by the clerk. When a particular service is busy, the administrator can coordinate the addition of a target service type to the window to reduce queuing. In this embodiment, each service type has a different priority order. Services can be queued according to their priority, either at the end of the queue for the same priority or ahead of services with a lower priority than the current user's priority.

[0141] Example 2,

[0142] See Figure 2-3 According to another aspect of the present invention, a queuing and calling system for payment counters is provided. This system includes:

[0143] The information acquisition module 201 is used to acquire the user's medical record number and select the business type and service type based on the medical record number; the information matching module 202 is used to query the window type of the payment counter based on the business type and service type and obtain the matching window; the information processing module 203 is used to obtain the queuing time of each matching window through a queuing algorithm model based on the matching window, and determine the matching window with the shortest queuing time on each floor as the target window; and provide the target window on each floor for the user to choose from; wherein, obtaining the queuing time of each matching window through the queuing algorithm model includes calculating the queuing time of each matching window by summing the business processing time and the service pause time. In this embodiment, the queuing and calling hardware system of the payment counter includes a ticket dispensing terminal, a calling terminal, a server, a database, a large display screen in the waiting area, a small display screen in the window, a sound generator box, a power amplifier, and a speaker. First, when the user takes a number at the ticket dispensing terminal, the server is triggered to compare information in the database. If the conditions are met, the ticket dispensing is successful; otherwise, the ticket dispensing fails. The ticket dispensing terminal includes a self-service machine, a mobile APP, and a WeChat mini program. After obtaining a number, users enter the waiting queue for their number to be called. When the cashier calls a user's number at the calling terminal, the server triggers a push notification to the large screen in the waiting area, the small screen at the service window, and the speaker. The large screen and small screen update their information, and the speaker controls the power amplifier to drive the speaker to play the call announcement. The large screen displays each window number, the current queue number and user's name, and the next queue number and user's name. The small screen displays the current window number, the queue number and user's name, and the voice announcement says, "Please go to window number * to process your service." Simultaneously, a message is pushed to the user's mobile app and WeChat official account, reminding them to go to the corresponding window. A message is also pushed to the fourth person in the queue, reminding them to return to the waiting area and wait for their number to be called. After completing their service, users can rate and provide suggestions on the service via the mobile app or mini-program. At the hospital's cashier, users can log in by swiping their medical card at the self-service machine or manually entering their medical record number. If the database does not find the user's medical record number, the self-service machine will display "Medical record number does not exist." After successful login, select the service you need, which mainly includes two categories: outpatient and inpatient services. Outpatient services include payment, refund, receipt printing, services for specific preferential groups, and other services. Inpatient services include admission, discharge, special needs, services for specific preferential groups, and other services. If the database does not find information for this service for the user, the self-service machine will display "No information found for this service." After successfully selecting a service, the system will generate the user's queue number on each floor and the estimated waiting time, recommending the floor with the shortest waiting time to the user. The user then selects the floor to obtain their queue number and successfully obtains it.In addition to checking in at the self-service machine, users can also obtain a number online by logging into the hospital's APP or WeChat mini-program on their mobile phones. The number-obtaining process is the same as that of the self-service machine, reducing the queuing time for users to obtain a number at the self-service machine.

[0144] Example 3,

[0145] According to another aspect of the present invention, a computer-readable storage medium is provided, storing at least one executable instruction. When executed on a queuing system for a payment counter as described above, the executable instruction causes the queuing system to perform the operation of a queuing method for a payment counter as described above. The present invention calculates the window and time with the shortest waiting time based on service type and service priority; simultaneously, the present invention also calculates the travel time from the user to the target window on each floor and recommends the target window and estimated waiting time to the user, thereby reducing queuing time. The queuing time calculation method of the present invention is more accurate, which can effectively improve hospital queuing efficiency and enhance the medical experience.

[0146] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. Similarly, for the sake of brevity and to aid in understanding one or more aspects of the invention, in the description of exemplary embodiments of the invention above, various features of the embodiments are sometimes grouped together in a single embodiment, figure, or description thereof. The claims, which follow the detailed description, are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of the invention.

[0147] Those skilled in the art will understand that the modules in the device of the embodiment can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiment can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components, except that at least some of such features and / or processes or units are mutually exclusive.

[0148] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be construed as limiting the order of execution.

Claims

1. A method of queuing for a toll booth by number, characterized by, The method includes at least: Obtain the user's medical record number, and select the business type and service type based on the medical record number; Based on the business type and service type, query the window type at the payment counter and obtain the matching window; Based on the matching window, the queuing time of each matching window is obtained through the queuing algorithm model, and the matching window with the shortest queuing time for each floor is determined as the target window. Each floor's target window is provided for the user to select; The step of obtaining the queuing time for each matching window through the queuing algorithm model includes: calculating the queuing time for each matching window by summing the business processing time and service pause time, using the following formula: , in, For business processing time, Let P be the service pause time, and P be the expected probability that the service will be paused before the patient is scheduled. The expected probability of the salesperson pausing service before reaching the patient is calculated through the following steps: Based on system logs, the frequency f(t) of service pauses for the current salesperson in the matching window across various time periods is calculated, where t∈ , k=1...K, where, This represents the k-th time interval, where K represents the number of time intervals. Get the service duration of the current salesperson since the last time the service was activated. Then, when the user is reached, the service time for the salesperson is... + The expected probability P of the salesperson pausing service before reaching the user is: ,in, This represents the floor function, where j is the time interval number corresponding to the current user's current time. This is a pause period.

2. The queuing and calling method at a payment counter according to claim 1, characterized in that, The processing time is calculated based on the number of people in the queue for each service in the matching window, specifically using the following formula: in, , is the matching window for the business staff to process the first The first type of business Time per user This refers to the number of different types of services in the queue ahead of the user. , It is the first The number of users of the type of service, the It can be obtained through the following formula: ,in, The salesperson handles the first Historical average time for similar business transactions Obtained through system log statistics; The salesperson handles the first The class of Calibration value for each user's time. Obtained through a BP neural network model.

3. A queuing and calling method at a payment counter according to claim 2, characterized in that, The Obtained through a BP neural network model, specifically including: Using the user's medical record number, the user's age, the province / city where the user's medical insurance is located, and the user's expense items are obtained, and the user's age, the province / city where the user's medical insurance is located, and the user's expense items are input into the input layer of the BP neural network model; The hidden and output layers of the BP neural network model use the Leaky ReLU activation function to calculate and output... ; The BP neural network model is obtained through the following steps: Define the error function ,in The target value for calibration. The calculated value is the calibration value; Recorded in the system log As training samples, among which For user age, For the province and city where the user's medical insurance is located, For user fee items; get, For the salesperson to handle the first The actual time of any user service in the same category. For the salesperson to handle the first Historical average time for this type of business; The weights from the hidden layer to the output layer and from the input layer to the hidden layer are updated using the gradient descent algorithm to obtain the BP neural network model.

4. A queuing and calling method at a payment counter according to claim 3, characterized in that, Calculating the business processing time also includes: Query the system logs to obtain the business priority of each matching window; When calculating the processing time of the service, the user is sorted according to the service priority order or placed last in the same service priority order, and the processing time is calculated accordingly.

5. A queuing and calling method at a payment counter according to claim 1, characterized in that, Average pause time for the business type employees This can be obtained by querying system log statistics.

6. A queuing and calling method at a payment counter according to claim 1, characterized in that, After determining the matching window with the shortest queuing time for each floor as the target window, the process also includes: The user's floor can be confirmed by scanning information through the self-service machine at the payment counter or the mobile app; Based on the user's floor, query the travel statistics table of each floor's payment counter to obtain the travel time required for the user to reach the target window on each floor, and provide the travel time to the user. The travel statistics table for each floor's payment counter includes the time it takes to travel from any payment counter on one floor to payment counters on other floors.

7. A queuing and calling method at a payment counter according to claim 1, characterized in that, The step of obtaining the user's medical record number and selecting the business type and service type based on the medical record number includes: The system receives the user's medical record number, service type, and business type through the self-service machine at the payment counter or via a mobile app. Query user information and fee information for medical record numbers, and determine whether the user has selected the correct service type; If so, then match the window type according to the business type and service type; Otherwise, if the business type or service type is selected incorrectly, a prompt will appear indicating that you should select it again. The business types include outpatient and inpatient types. The outpatient type specifically includes payment, refund, receipt printing, and special preferential groups. The inpatient type specifically includes admission, discharge, and special preferential groups. The window type includes business type priority and service type. The business type priority is determined by each window staff member or through the system. The service type includes one or more of Mandarin, local language, and English.

8. A queuing and calling system for a payment counter, characterized in that, The system is used to execute a queuing and calling method for a payment counter as described in any one of claims 1-7, the system comprising: The information acquisition module is used to acquire the user's medical record number and select the business type and service type based on the medical record number; The information matching module is used to query the window type at the payment counter based on the business type and service type, and obtain the matching window. The information processing module is used to obtain the queuing time of each matching window according to the matching window through a queuing algorithm model, and determine the matching window with the shortest queuing time for each floor as the target window; and provide the target window for each floor to the user for selection; wherein, obtaining the queuing time of each matching window through the queuing algorithm model includes: calculating the queuing time of each matching window by the sum of the business processing time and the service pause time.

9. A computer-readable storage medium, characterized in that, The storage medium stores at least one executable instruction, which, when executed on a queuing system for a payment counter as described in claim 8, causes the queuing system for a payment counter to perform the operation of a queuing method for a payment counter as described in any one of claims 1-7.

Images