Post-sale processing method for analyte monitoring devices and related devices

Abstract

The present disclosure provides a post-sale processing method of an analyte monitoring device and related devices, the post-sale processing method comprising: obtaining post-sale request data corresponding to a post-sale request of a user and multi-dimensional data associated with the post-sale request data; performing preliminary judgment based on the multi-dimensional data to obtain a first judgment result; in response to the first judgment result indicating that the preliminary judgment is passed and the post-sale type is data deviation, obtaining monitoring data corresponding to the analyte monitoring device; identifying whether there is a preset abnormal pattern based on the monitoring data; if there is the preset abnormal pattern, determining that a second judgment result is passed, otherwise, obtaining first image data submitted by the user; extracting monitoring values and corresponding first time information, and control values and corresponding second time information from the first image data, and determining a deviation analysis result based thereon; determining the second judgment result based on the deviation analysis result; and processing a post-sale work order according to the second judgment result. Thus, the post-sale processing efficiency can be improved.

Description

Technical Field

[0001] This disclosure relates to the field of intelligent customer service technology, and in particular to an after-sales processing method and related equipment for an analytical substance monitoring device. Background Technology

[0002] During the use of analyte monitoring equipment, users may initiate after-sales requests based on anomalies (such as inaccurate data). These requests typically involve further verification of the anomalies and are quite frequent and complex in the actual use of analyte monitoring equipment.

[0003] In existing technologies, addressing such after-sales requests typically requires a comprehensive assessment combining various data points and user feedback. In some cases, users may also need to provide relevant supporting documentation to confirm the anomalies and improve the reliability of the assessment.

[0004] However, in practice, the processing of such after-sales requests relies heavily on the evidence provided by the user. When the information provided by the user is extensive or non-standard, it can increase the complexity of after-sales processing and prolong the overall after-sales decision-making process, thereby reducing the efficiency of after-sales processing to some extent. Summary of the Invention

[0005] This disclosure is made in view of the above-mentioned situation, and its purpose is to provide an after-sales processing method and related equipment for analytical monitoring equipment that can improve after-sales processing efficiency.

[0006] To this end, the first aspect of this disclosure provides an after-sales processing method for an analytical substance monitoring device, comprising: acquiring after-sales request data corresponding to a user's after-sales request and multi-dimensional data associated with the after-sales request data, wherein the multi-dimensional data includes user-dimensional data, device-dimensional data, and channel-dimensional data; performing a preliminary judgment based on the multi-dimensional data to obtain a first judgment result; in response to the first judgment result indicating that the preliminary judgment is passed and the after-sales type in the after-sales request data is data deviation, acquiring monitoring data corresponding to the analytical substance monitoring device; identifying whether a preset abnormal mode exists based on the monitoring data, wherein the preset abnormal mode includes a continuous low value mode, a no-fluctuation mode, and a jump point mode; in response to the existence of the preset abnormal mode, determining a second judgment result as passed; in response to the absence of the preset abnormal mode, acquiring first image data submitted by the user; extracting a monitoring value and corresponding first time information, and a control value and corresponding second time information from the first image data; determining a deviation analysis result based on the monitoring value, the first time information, the control value, and the second time information; determining the second judgment result based on the deviation analysis result; and processing an after-sales work order according to the second judgment result.

[0007] In the first aspect of this disclosure, preliminary judgment based on multi-dimensional data can improve the comprehensiveness of assessing the passability of after-sales requests, while reducing the computational load of subsequent complex analyses, thereby improving after-sales processing efficiency and the accuracy of after-sales access judgment. Furthermore, by automatically identifying preset anomaly patterns in the monitoring data, it is possible to quickly determine whether after-sales requests with data deviations are acceptable, thus reducing the need for users to submit evidence. This improves after-sales processing efficiency and user experience (for example, users may be able to pass after-sales requests without submitting any evidence of data deviation), and also reduces the pressure on the after-sales system to process user-submitted data (for example, image data processing typically requires significant hardware resources). Additionally, pre-identifying preset anomaly patterns, and then combining them with the first image data for data deviation analysis when the preset anomaly pattern is not identified, can reduce the interference of other data anomalies on the data deviation analysis, thereby improving the accuracy and reliability of the data deviation analysis.

[0008] Additionally, in the after-sales processing method according to the first aspect of this disclosure, optionally, identifying the persistent low-value pattern includes: setting a first sliding window; determining whether all monitoring values ​​of the monitoring data within the first sliding window are not higher than a first monitoring value threshold, and determining whether there are no data breakpoints exceeding a first preset time period within the first sliding window; in response to the existence of a target sliding window, acquiring second monitoring data within the most recent preset time period, and determining the highest monitoring value in the second monitoring data, wherein the target sliding window is the first sliding window where all monitoring values ​​are not higher than the first monitoring value threshold and there are no data breakpoints exceeding the first preset time period; and determining the existence of the persistent low-value pattern when the highest monitoring value is not higher than the second monitoring value threshold, wherein the second monitoring value threshold is greater than the first monitoring value threshold. In this case, it is possible to identify data points with excessively low monitoring values ​​while improving data continuity, thereby improving the accuracy of identifying the persistent low-value pattern. Furthermore, using monitoring values ​​over a longer time range to further confirm the existence of the persistent low-value pattern can further improve the accuracy of identifying the persistent low-value pattern.

[0009] Furthermore, in the after-sales processing method according to the first aspect of this disclosure, optionally, identifying the fluctuation-free mode includes: setting a second sliding window; acquiring multiple monitoring values ​​of the monitoring data within the second sliding window; determining the minimum monitoring value and the range of the multiple monitoring values ​​among the multiple monitoring values; and determining the existence of the fluctuation-free mode if the minimum monitoring value is not lower than a third monitoring value threshold and the range is not greater than a fluctuation threshold. This improves the accuracy of identifying the fluctuation-free mode.

[0010] Additionally, in the after-sales processing method according to the first aspect of this disclosure, optionally, at least two sets of data are acquired, each set including the monitored value, the first time information, the control value, and the second time information, wherein the monitored value, the first time information, the control value, and the second time information are extracted from the first image data based on optical character recognition; the activation duration of the analyte monitoring device is acquired and a no-bias condition is determined, the no-bias condition including: in response to the activation duration being in a first time period, the absolute difference between the monitored value and the control value does not exceed a first deviation threshold; in response to the activation duration being in a second time period, the monitored value is within a preset deviation range determined by the control value, the second time period being later than the first time period; the number of data sets satisfying the no-bias condition in the at least two sets of data is counted; if the number of data sets is greater than a preset number of sets, the deviation analysis result is determined to be without deviation. In this case, by distinguishing between early lenient standards and later precise standards, the reliability and accuracy of after-sales access judgment can be improved, while unnecessary manual intervention can be reduced.

[0011] Additionally, the after-sales processing method according to the first aspect of this disclosure may optionally include: in response to the first judgment result indicating that the preliminary judgment is passed and the after-sales type in the after-sales request data is sensor implantation failure, acquiring second image data submitted by the user; identifying the image type corresponding to the second image data; in response to the image type being a device appearance image, performing fault identification on the device appearance image based on a visual language model to obtain an image review result, the image review result including the fault type and the confidence level of the fault type; and determining the second judgment result based on the confidence level of the fault type. In this case, a more factual result can be output, and subsequent judgments can be made in conjunction with other data, reducing the risk of directly drawing after-sales conclusions based on a visual language model.

[0012] Furthermore, in the after-sales processing method according to the first aspect of this disclosure, optionally, it further includes: responding to the first judgment result indicating that the preliminary judgment is passed and the after-sales type in the after-sales request data is sensor abnormality, determining an abnormality identification result based on the identified abnormality state of the analyte monitoring device and the recovery time window; and determining the second judgment result based on the abnormality identification result. This avoids misjudging situations that can recover on their own as requiring after-sales service.

[0013] Furthermore, in the after-sales processing method according to the first aspect of this disclosure, optionally, in response to the analyte monitoring device being in use, the latest monitoring point in the monitoring data of the analyte monitoring device is determined; the device status and time information corresponding to the latest monitoring point are obtained; in response to the device status being an abnormal state related to the sensor, it is determined whether the time difference between the current time and the time information is greater than the recovery time window; if the time difference is greater than the recovery time window, the abnormal state identification result is determined to be an anomaly; if the time difference is not greater than the recovery time window, the abnormal state identification result is determined to be an undetermined anomaly. This improves the accuracy of determining whether a sensor anomaly has occurred during the use phase.

[0014] Additionally, in the after-sales processing method involved in the first aspect of this disclosure, optionally, the user dimension data includes at least one of the user's historical after-sales records and device activation records. The preliminary judgment based on the user dimension data includes at least one of the following: determining whether the number of devices corresponding to consecutively passed after-sales records exceeds a first threshold and whether the usage time of the corresponding analytical substance monitoring devices is greater than a second preset duration based on the user's historical after-sales records; counting the number of consecutively passed after-sales records with the same fault type as the after-sales request data within a preset time range in the user's historical after-sales records, and determining whether the number exceeds a second threshold; determining whether the number of passed after-sales records exceeds a third threshold based on the user's historical after-sales records related to the analytical substance monitoring device; and obtaining the device activation records corresponding to the passed after-sales records in the user's historical after-sales records, and determining whether the corresponding analytical substance monitoring device is reactivated after the after-sales service is completed based on the device activation records and the passed after-sales records; and / or the channel dimension data includes at least one of channel type and channel configuration information. The preliminary judgment based on the channel dimension data includes: obtaining the channel type corresponding to the analytical substance monitoring device, obtaining the channel type corresponding to the channel type... The system uses channel configuration information to determine the after-sales restriction type corresponding to the channel type. In response to the after-sales restriction type indicating permission to enter, the system determines that the preliminary judgment based on the channel dimension data is passed. In response to the after-sales restriction type indicating a restriction on the number of after-sales visits, the system determines whether the number of after-sales visits corresponding to the analytical substance monitoring device has reached a preset number. If the preset number has not been reached, the system determines that the preliminary judgment based on the channel dimension data is passed. In response to the after-sales restriction type indicating prohibition of entry, the system determines that the first judgment result is failed. The system and / or the device dimension data includes at least one of the following: device identifier, activation status, time-related validity information, and historical after-sales records of the analytical substance monitoring device. The preliminary judgment based on the device dimension data includes at least one of the following: determining whether an analytical substance monitoring device corresponding to the device identifier exists based on the device identifier; determining whether the activation status matches the after-sales type; determining whether the analytical substance monitoring device is within the commitment period and the product validity period based on the validity information; and determining whether there is an ongoing after-sales work order associated with the analytical substance monitoring device based on the historical after-sales records of the analytical substance monitoring device.

[0015] Furthermore, in the after-sales processing method according to the first aspect of this disclosure, optionally, the second judgment result is determined based on at least one of the following target information: the risk level of the supply channel corresponding to the analyte monitoring equipment; the user risk score; and the fault cause classification corresponding to the after-sales type. In this case, at least one factor, such as the results of previous judgments (e.g., image review results or deviation analysis results), risk level, user risk score, and fault cause classification, can be integrated into the final judgment, thereby improving the reliability of the after-sales access judgment.

[0016] A second aspect of this disclosure provides an electronic device including a processor and a memory, the memory storing a computer program that, when executed by the processor, implements the after-sales processing method as described in the first aspect of this disclosure.

[0017] According to this disclosure, an after-sales processing method and related equipment for analytical substance monitoring equipment that can improve after-sales processing efficiency are provided. Attached Figure Description

[0018] This disclosure will now be explained in further detail by way of example only with reference to the accompanying drawings.

[0019] Figure 1 This is a schematic diagram illustrating the aftermarket environment covered by the examples in this disclosure.

[0020] Figure 2 This is a schematic diagram illustrating the monitoring equipment involved in the example of this disclosure.

[0021] Figure 3 This is an exemplary flowchart illustrating the after-sales processing method involved in the examples of this disclosure.

[0022] Figure 4 This is an exemplary flowchart illustrating an after-sales request for handling data discrepancies as described in this disclosure.

[0023] Figure 5 This is an exemplary flowchart illustrating the identification of persistent low value patterns as described in this disclosure.

[0024] Figure 6 This is an exemplary flowchart illustrating the identification of non-fluctuation patterns as described in this disclosure.

[0025] Figure 7 This is an exemplary flowchart illustrating the identification of jump point patterns as described in this disclosure.

[0026] Figure 8 This is an exemplary flowchart illustrating an after-sales request for handling a failed sensor implantation, as described in this disclosure.

[0027] Figure 9This is an exemplary block diagram illustrating an electronic device to which the present disclosure is based. Detailed Implementation

[0028] Hereinafter, preferred embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In the following description, the same reference numerals are used for the same components, and repeated descriptions are omitted. Furthermore, the drawings are merely schematic diagrams, and the proportions of the components or the shapes of the components may differ from actual figures. It should be noted that the terms "comprising" and "having," and any variations thereof, in this disclosure, do not necessarily limit the process, method, system, product, or apparatus to the explicitly listed steps or units, but may include or have other steps or units not explicitly listed or inherent to these processes, methods, products, or apparatuses.

[0029] First, let me introduce the relevant terminology used in this disclosure.

[0030] An "analyte monitoring device" can refer to a device for continuously monitoring analytes within a user's body. In some examples, an analyte monitoring device may include sensors and electronic components. The sensors may acquire analyte signals in relation to analyte concentrations. The electronic components may process and / or transmit the analyte signals. In some examples, the analyte monitoring device may also include a receiving device that can display the analyte concentration and / or provide a user interface related to the analyte monitoring device. In some examples, the analyte signal may be an electrical signal (e.g., a current signal), from which the corresponding analyte concentration can be obtained. In some examples, the sensor may be based on electrochemical measurements.

[0031] In some examples, the analyte may include at least one of acetylcholine, amylase, bilirubin, cholesterol, human chorionic gonadotropin, creatine kinase, creatine, creatine anhydride, DNA, fructosamine, glucose, glutamine, growth hormone, ketone bodies, lactate, peroxide, prostate-specific antigen, prothrombin, RNA, thyroid-stimulating hormone, and troponin.

[0032] A "Visual Language Model (VLM)" can refer to a multimodal model that can process both image and text information simultaneously. For example, a visual language model could be qwen3-vl-plus.

[0033] This disclosure also relates to an after-sales processing method for analyte monitoring equipment (hereinafter referred to as the after-sales processing method), which can improve after-sales processing efficiency. Furthermore, the after-sales processing method described in this disclosure may also be referred to as an after-sales service method, or an after-sales scheduling method, etc. In some examples, the after-sales processing method can centrally manage judgments related to whether or not entry into the after-sales process is permitted.

[0034] For ease of description, some examples below use glucose as the analyte and a continuous glucose meter as an example of the analyte monitoring device (hereinafter referred to as the monitoring device). It should be noted that this does not constitute a limitation of this disclosure; unless there is a contradiction, the relevant descriptions also apply to other types of analytes. Furthermore, unless there is a contradiction, some examples can also be applied to aftermarket medical devices or other arbitrary devices besides analyte monitoring devices. Additionally, the examples in this disclosure involve various thresholds, preset values, or similar values. It should be noted that, unless there is a contradiction, this disclosure does not limit the specific values, and adjustments can be made based on the performance (e.g., accuracy) of the corresponding processing.

[0035] Examples of this disclosure will now be described in detail with reference to the accompanying drawings.

[0036] Figure 1 This is a schematic diagram illustrating the after-sales environment involved in the examples disclosed herein. It should be noted that... Figure 1 This does not imply any limitation on this disclosure.

[0037] In some examples, reference Figure 1 In an after-sales environment, when a user needs to report a problem with the monitoring device 100 or has questions while using the monitoring device 100, they can connect to the after-sales system 1 through the session client 10 (i.e., the client that displays the interface between the user and customer service) executed by the receiving device 200. The after-sales system 1 can automatically handle the user's problem based on its nature, or it can transfer the user to the human customer service terminal 20. In some examples, the intelligent module 30 of the after-sales system 1 can automatically handle some of the user's problems. Furthermore, the intelligent module 30 can be any module capable of automatic execution. For example, the intelligent module 30 can perform intent recognition, knowledge base retrieval, answer generation, decision-making (e.g., whether to transfer to a human agent), and / or session routing with a human customer service representative. In some examples, the after-sales system 1 can also be configured to execute the after-sales processing methods involved in the examples of this disclosure to obtain a second judgment result and / or process after-sales work orders (described later).

[0038] Figure 2 This is a schematic diagram illustrating the monitoring device 100 involved in the example of this disclosure.

[0039] In some examples, reference Figure 2The monitoring device 100 may include a sensor 101 and an electronic component 102. The sensor 101 can acquire analyte signals related to analyte concentration. The electronic component 102 can receive the analyte signals generated by the sensor 101 and can process and / or transmit the analyte signals. Specifically, the electronic component 102 may be connected to the sensor 101 to receive analyte signals from the sensor 101. In some examples, the electronic component 102 may be electrically connected to the sensor 101. In some examples, the sensor 101 may be at least partially implanted under the skin of a user to acquire analyte signals related to analyte concentration.

[0040] In some examples, sensor 101 may include a working electrode, and the analyte signal can be obtained by measuring the electrical signal at the working electrode. The working electrode may be an enzyme-containing electrode. That is, an analyte enzyme (e.g., glucose oxidase, glucose dehydrogenase, or hydroxybutyrate dehydrogenase) may be disposed on the working electrode. The analyte enzyme can be used to promote an electrochemical reaction. Specifically, the analyte enzyme on the working electrode can contact the analyte in body fluids to promote an electrochemical reaction (e.g., a redox reaction), thereby generating an electrical signal. In some examples, sensor 101 may also include a counter electrode and a reference electrode. Thus, a three-electrode system can be formed.

[0041] Figure 3 This is an exemplary flowchart illustrating the after-sales processing method involved in the examples of this disclosure.

[0042] In some examples, reference Figure 3 The after-sales processing method may include obtaining the after-sales request data corresponding to the user's after-sales request (step S101). The after-sales request data may be the data returned by the user when submitting the after-sales request. In some examples, the after-sales request data may at least include the after-sales type. In some examples, the after-sales request data may also include the user's user identifier and the device identifier of the monitoring device 100. This facilitates the acquisition of user information and device information.

[0043] In some examples, reference Figure 3 The after-sales processing method may include obtaining multi-dimensional data associated with the after-sales request data (step S102). That is, obtaining multi-dimensional data related to the after-sales request. Furthermore, the multi-dimensional data may include data from different dimensions (e.g., user dimension, device dimension, or channel dimension data). In this case, making a preliminary judgment based on multi-dimensional data can improve the comprehensiveness of assessing the passability of the after-sales request, while reducing the computational load of subsequent complex analyses, thereby improving after-sales processing efficiency and the accuracy of after-sales access judgment.

[0044] In some examples, multi-dimensional data may include at least one of user-dimensional data, device-dimensional data, and channel-dimensional data. In other examples, multi-dimensional data may include user-dimensional data, device-dimensional data, and channel-dimensional data. In this case, including data from all three dimensions simultaneously can further improve the comprehensiveness of the after-sales request availability assessment.

[0045] Additionally, user-dimensional data can refer to data related to the user who submitted the after-sales request. In some examples, user-dimensional data may include at least one of the user's historical after-sales records and device activation records.

[0046] Additionally, device-level data can refer to data related to the monitoring device 100 involved in the after-sales request. In some examples, device-level data may include at least one of the monitoring device 100's device identifier, activation status, time-related validity information, and historical after-sales records. In some examples, time-related validity information may include at least one of the commitment period and the product validity period. Furthermore, the commitment period may be the promised after-sales service time window for the monitoring device 100, calculated from the activation date.

[0047] Additionally, channel-level data can refer to data about the supply channels to which monitoring device 100 belongs. In some examples, channel-level data may include at least one of channel type and channel configuration information.

[0048] In some examples, reference Figure 3 The after-sales processing method may include making a preliminary judgment based on multi-dimensional data to obtain a first judgment result (step S103). The first judgment result can be used to indicate whether the preliminary judgment is passed. For example, the first judgment result may include pass and fail. In some examples, steps S102 and S103 may not be necessary.

[0049] In some examples, a first machine learning model can be applied to multi-dimensional data to obtain an initial judgment. For example, the first machine learning model may include support vector machines, random forests, and / or neural networks. In some examples, a first machine learning model may not be used.

[0050] In some examples, a preliminary judgment can be made based on user-dimensional data, device-dimensional data, and channel-dimensional data in sequence to determine the first judgment result. For example, a first machine learning model can be trained using data from different dimensions, and the trained first machine learning model can be applied to each dimension of data to obtain the corresponding result, thereby obtaining the first judgment result.

[0051] In some examples, a preliminary judgment based on user-dimensional data can determine whether a user meets after-sales eligibility. Furthermore, user-dimensional data can be historical data. For example, user-dimensional data can be data from a recent preset time range (e.g., the last 90 days).

[0052] In some examples, the initial judgment based on user-dimensional data may include at least one sub-judgment. In some examples, at least one sub-judgment may include at least one of a first initial judgment, a second initial judgment, a third initial judgment, and a fourth initial judgment, which will be described later.

[0053] In some examples, the first preliminary judgment can be based on whether the number of devices corresponding to consecutively passed after-sales records exceeds a first threshold, and whether the usage time of the corresponding monitored devices is greater than a second preset time. In some examples, if the number of devices exceeds the first threshold and the usage time of the corresponding monitored devices is greater than the second preset time, the result of the first preliminary judgment can be "fail"; otherwise, the result of the first preliminary judgment can be "pass". That is, if multiple sets of devices have consecutively passed after-sales services (e.g., replacement or exchange), and the actual usage days of each set are not less than the specified number of days, then the first preliminary judgment fails.

[0054] In some examples, the second preliminary judgment can be to count the number of consecutive passed after-sales service records with the same fault type as the after-sales request data within a preset time range in the user's historical after-sales records, and determine whether the number exceeds a second threshold. In some examples, if the number exceeds the second threshold, the result of the second preliminary judgment can be "failed"; otherwise, the result of the second preliminary judgment can be "passed".

[0055] In some examples, different second thresholds can be set based on the fault cause category to which the fault type belongs. This can improve the accuracy of the second preliminary judgment. For example, a lower threshold (e.g., 2 consecutive faults) can be used for fault causes belonging to user operation problems, while a higher threshold (e.g., 3 consecutive faults) can be used for fault causes belonging to product damage itself.

[0056] In some examples, the third preliminary judgment may be based on the historical after-sales records of users associated with monitoring device 100 to determine whether the number of approved after-sales records exceeds a third threshold (hereinafter referred to as the third preliminary judgment). In some examples, if the number of approved after-sales records exceeds the third threshold, the result of the third preliminary judgment may be "not approved"; otherwise, the result of the third preliminary judgment may be "approved".

[0057] For example, if the first device applied for a second device through after-sales service, and the second device applied for a third device through after-sales service, then the second and third devices can be considered as two consecutive after-sales services related to the first device.

[0058] In some examples, the fourth preliminary judgment can be to obtain the device activation record corresponding to the approved after-sales record in the user's historical after-sales records, and determine whether the corresponding monitoring device has been reactivated after the after-sales service is completed based on the device activation record and the approved after-sales record. In some examples, if the corresponding monitoring device has been reactivated after the after-sales service is completed, the result of the fourth preliminary judgment can be "failed"; otherwise, the result of the fourth preliminary judgment can be "passed".

[0059] In some examples, if the result of any sub-judgment (e.g., the first preliminary judgment, the second preliminary judgment, the third preliminary judgment, and the fourth preliminary judgment) in the preliminary judgment based on user dimension data indicates that the corresponding sub-judgment has failed, then the result of the first judgment can be determined as failed. In some examples, if the results of all sub-judgments in the preliminary judgment based on user dimension data indicate that the corresponding sub-judgment has passed, then the result of the preliminary judgment based on user dimension data can be determined as passed.

[0060] In some examples, a first preliminary judgment, a second preliminary judgment, a third preliminary judgment, and a fourth preliminary judgment can be performed in sequence. In response to the result of the current sub-judgment indicating that the current sub-judgment fails, the subsequent sub-judgments are stopped.

[0061] In some examples, a preliminary judgment based on user-dimensional data can be made in response to the user's profile assessment switch being turned on; if the user's profile assessment switch is not turned on, the preliminary judgment based on user-dimensional data is deemed passed. In some examples, a preliminary judgment based on user-dimensional data can be made in response to the user's profile assessment switch being turned on and the user not being a new user. In some examples, a preliminary judgment based on user-dimensional data can be deemed passed in response to the user's profile assessment switch being turned on and the user being a new user. Additionally, a new user can refer to a user with zero historical after-sales records.

[0062] In some examples, for an initial assessment based on user-level data, the response to the first assessment result being "failed," the second assessment result could be "pending review." That is, after-sales requests that fail at the user level can undergo further review. In this case, it facilitates routing to human customer service for processing the after-sales request, reducing the likelihood of special cases being handled automatically.

[0063] In some examples, a preliminary assessment based on device-level data can determine whether monitoring device 100 meets after-sales eligibility criteria. That is, it determines whether monitoring device 100 itself meets the basic conditions for after-sales service.

[0064] In some examples, the initial judgment based on device-level data may include at least one sub-judgment. In some examples, at least one sub-judgment may include at least one of the fifth, sixth, seventh, and eighth initial judgments, which will be described later.

[0065] In some examples, the fifth preliminary judgment can be based on the device identifier to determine whether a monitoring device 100 corresponding to the device identifier exists. In some examples, if no monitoring device 100 corresponding to the device identifier exists, the result of the fifth preliminary judgment can be "fail"; otherwise, the result of the fifth preliminary judgment can be "pass".

[0066] In some examples, the sixth preliminary judgment can be to determine whether the activation status matches the after-sales type. In some examples, if the activation status does not match the after-sales type, the result of the sixth preliminary judgment can be "fail"; otherwise, the result of the sixth preliminary judgment can be "pass". For example, different after-sales types require different activation statuses. When monitoring device 100 is not activated, an after-sales request for sensor implantation failure can be submitted; when monitoring device 100 is activated, an after-sales request for sensor malfunction can be submitted.

[0067] In some examples, the seventh preliminary judgment may be based on validity information to determine whether the monitoring device 100 is within the warranty period and whether it is within the product validity period. In some examples, if the monitoring device 100 is not within the warranty period or not within the product validity period, the result of the seventh preliminary judgment may be "fail"; otherwise, the result of the seventh preliminary judgment may be "pass".

[0068] In some examples, the eighth preliminary judgment can be based on the historical after-sales records of monitoring device 100 to determine whether there is an ongoing after-sales work order associated with monitoring device 100. In some examples, if there is an ongoing after-sales work order associated with monitoring device 100, the result of the eighth preliminary judgment can be "failed." This avoids duplicate after-sales service. In some examples, if there is no ongoing after-sales work order and no draft after-sales work order, the result of the eighth preliminary judgment can be "passed." In some examples, if there is a draft after-sales work order, the result of the eighth preliminary judgment can be determined as "passed" after receiving the user's instruction to cancel the after-sales work order.

[0069] In some examples, if the result of any sub-judgment (e.g., the fifth, sixth, seventh, and eighth preliminary judgments) in the initial judgment based on device dimension data indicates that the corresponding sub-judgment failed, the first judgment result can be determined as failed. In some examples, if the results of all sub-judgments in the initial judgment based on device dimension data indicate that the corresponding sub-judgment passed, the initial judgment based on device dimension data can be determined as passed.

[0070] In some examples, the fifth, sixth, seventh, and eighth preliminary judgments can be performed sequentially. In response to the result of the current sub-judgment indicating that the current sub-judgment fails, the first judgment result is determined to be unsuccessful, and subsequent sub-judgments are stopped.

[0071] In some examples, for initial judgments based on device-level data, in response to a first judgment result of "failed," a second judgment result can also be "failed." That is, after-sales requests that fail at the device level can be automatically rejected. This improves after-sales processing efficiency.

[0072] In some examples, a preliminary judgment based on channel-level data can determine whether the supply channel to which monitoring device 100 belongs is suitable for the current after-sales service entry point. In some examples, the suitability of the supply channel for the current after-sales service entry point can be determined based on the channel type to which the supply channel belongs. In some examples, differentiated judgments can be made based on channel configuration information of the channel type.

[0073] In some examples, channel types may include at least one of institutional channels (such as medical institutions), membership channels, retail channels, event channels, special channels, and agency distribution channels. An event channel is a type of supply channel through which a merchant distributes the monitoring device 100 to users based on promotional activities.

[0074] In some examples, the initial judgment based on channel-level data may include obtaining the channel type corresponding to monitoring device 100; obtaining the channel configuration information corresponding to the channel type; determining the after-sales restriction type corresponding to the channel type based on the channel configuration information; and determining the first judgment result based on the after-sales restriction type. This can improve after-sales processing efficiency.

[0075] In some examples, after-sales restriction types can include at least one of allowed access, limited after-sales service attempts, and prohibited access. For example, the after-sales restriction type for institutional and membership channels can be prohibited access. This facilitates guiding users of these channel types to submit after-sales requests through a more suitable after-sales portal. As another example, the after-sales restriction type for retail channels can be allowed access. And as yet another example, the after-sales restriction type for event channels can include allowed access, limited after-sales service attempts, and prohibited access.

[0076] In some examples, responding to the after-sales restriction type indication allowing entry, the preliminary judgment based on channel dimension data can be determined as passed. In some examples, responding to the after-sales restriction type indication limiting the number of after-sales services, it can be determined whether the number of after-sales services corresponding to monitoring device 100 has reached a preset number, and if the preset number has not been reached, the preliminary judgment based on channel dimension data is determined as passed. In some examples, responding to the after-sales restriction type indication prohibiting entry, the first judgment result is determined as failed.

[0077] In some examples, the initial judgment based on channel dimension data may include obtaining the channel blacklist in response to the channel type belonging to the type that needs to be checked for blacklisting; and determining the first judgment result as failing in response to the supply channel of monitoring device 100 being in the channel blacklist.

[0078] In some examples, in response to the fact that the supply channel of monitoring device 100 is not in the channel blacklist, the preliminary judgment based on channel-dimensional data is deemed successful. In other examples, in response to the fact that the supply channel of monitoring device 100 is not in the channel blacklist, the above-mentioned judgment related to the after-sales restriction type can be performed. For example, the after-sales restriction type for a special channel can be "allowed entry, but it needs to be checked whether it is on the blacklist."

[0079] In some examples, the initial judgment based on channel-level data may include, in response to the channel type being an agency / distribution channel, retrieving after-sales processing records from an external system; in response to the presence of after-sales processing records on monitoring device 100, determining the first judgment result as "failed"; and in response to the absence of after-sales processing records on monitoring device 100, determining the result of the initial judgment based on channel-level data as "passed." In some examples, in response to the absence of after-sales processing records on monitoring device 100, the above-mentioned judgment related to the after-sales restriction type can be performed. For example, the after-sales restriction type for an agency / distribution channel may be "allowed entry," but it is necessary to first confirm whether after-sales processing records exist on monitoring device 100 in the external system. Furthermore, the external system can be any system that allows agents / distributors to submit after-sales processing records.

[0080] In some examples, for initial judgments based on channel-level data, if the first judgment result is "failed," the second judgment result can also be "failed." That is, after-sales requests that fail at the channel level can be automatically rejected. This improves after-sales processing efficiency.

[0081] In some examples, if the initial judgment for any dimension of data is "fail," the first judgment result can be determined as "fail." In other examples, if the initial judgment for each dimension of data is "pass," the first judgment result can be determined as "pass."

[0082] In some examples, reference Figure 3 The after-sales processing method may include, in response to a first judgment result indication, obtaining a second judgment result based on the after-sales type in the after-sales request data through preliminary judgment (step S104). In this case, by first filtering out after-sales requests that clearly do not meet the conditions or are not suitable for entering the subsequent process through preliminary judgment, and then obtaining the second judgment result based on the after-sales type, the efficiency of after-sales processing can be improved and unnecessary data analysis can be reduced. In addition, this hierarchical judgment method can select appropriate analysis processes for different after-sales types, thereby improving the accuracy of after-sales access judgment and the scalability of the system.

[0083] Additionally, the after-sales service type can refer to the type of problem reported by the user related to the monitoring device 100. In some examples, the after-sales service type may include at least one of data deviation, sensor implantation failure, and sensor malfunction. Data deviation can refer to inaccurate monitoring data from the monitoring device 100. Sensor implantation failure can refer to the failure of sensor 101 of the monitoring device 100 to be implanted successfully. Sensor malfunction can refer to a malfunction of sensor 101 of the monitoring device 100 (e.g., sensor failure or sensor detachment).

[0084] In some examples, the after-sales service type can include other types. In some examples, in response to the after-sales service type being another type, the second judgment result can be pending review. That is, when the after-sales service type selected by the user is a non-standard type, review can be mandatory.

[0085] Additionally, the second judgment result can indicate whether the after-sales request is approved. In some examples, the second judgment result can include "approved". In some examples, the second judgment result can include "approved" and "not approved". In some examples, the second judgment result can include "approved", "not approved", and "pending review". Furthermore, "pending review" in the second judgment result can indicate that further confirmation is needed regarding the approval of the after-sales request.

[0086] Figure 4 This is an exemplary flowchart illustrating an after-sales request for handling data discrepancies as described in this disclosure.

[0087] In some examples, after-sales requests for handling data discrepancies can determine whether data anomalies in monitoring device 100 qualify for after-sales service based on monitoring data collected by monitoring device 100. Specifically, time-series analysis can be performed on the monitoring data to identify data anomaly patterns and determine whether they constitute eligible equipment malfunctions. Additionally, the monitoring data may include analyte concentration data, such as continuous glucose levels.

[0088] In some examples, reference Figure 4The after-sales request for handling data deviation may include: in response to the after-sales type being data deviation in the after-sales request data, obtaining the monitoring data corresponding to the monitoring device 100 (step S201); identifying whether a preset abnormal mode exists based on the monitoring data (step S202); in response to the existence of a preset abnormal mode, determining the second judgment result as passed (step S203); in response to the absence of a preset abnormal mode, obtaining the first image data submitted by the user (step S204); extracting the monitoring value and corresponding first time information, as well as the reference value and corresponding second time information (hereinafter sometimes referred to as deviation proof information) from the first image data (step S205); determining the deviation analysis result based on the monitoring value, first time information, reference value, and second time information (step S206); and determining the second judgment result based on the deviation analysis result (step S207).

[0089] In this context, by automatically identifying preset anomaly patterns in the monitoring data, it is possible to quickly determine whether an after-sales request for data deviation should be approved. This reduces the need for users to submit supporting evidence, improving both after-sales processing efficiency and user experience (for example, users may be able to approve their after-sales requests without submitting any evidence of data deviation). Furthermore, it reduces the burden on the after-sales system 1 when processing user-submitted data (for example, image data processing typically consumes significant hardware resources). Additionally, pre-identifying preset anomaly patterns, and then combining them with the first image data for data deviation analysis when those patterns are not identified, reduces interference from other data anomalies, thereby improving the accuracy and reliability of the data deviation analysis.

[0090] In some examples, reference Figure 4 In step S201, the monitoring data may include multiple consecutive monitoring values. Furthermore, the monitoring values ​​are also the monitoring values ​​of the analyte, such as glucose concentration or ketone body concentration.

[0091] In some examples, the raw monitoring data can be preprocessed to obtain the monitoring data. For example, preprocessing may include removing data points from the monitoring device 100 during the initial activation phase and retaining data points from the valid time period. This reduces the interference of unstable data during the initial activation phase on abnormal pattern recognition and improves the effectiveness of the monitoring data in identifying abnormal patterns. In some examples, the initial activation phase may refer to a certain period after wearing the device (e.g., the first hour). In some examples, the valid time period may refer to the period during which the monitoring data is in a non-steady state (i.e., the fluctuation amplitude is greater than a preset amplitude). In some examples, the valid time period may refer to the time period excluding fasting periods, such as 06:00-23:59 daily.

[0092] In some examples, in response to the number of data points in the monitoring data being less than a preset number (e.g., 12), the after-sales request to process data deviations can be cancelled. This reduces the likelihood of inaccurate judgments of data deviations due to insufficient data volume.

[0093] In some examples, reference Figure 4 In step S202, the preset anomaly mode can refer to a pre-set anomaly mode. In some examples, the preset anomaly mode can be an anomaly mode unrelated to equipment failure. In this case, distinguishing preset anomaly modes unrelated to equipment failure and automatically requesting after-sales service when such anomalies are identified can reduce invalid analysis and unnecessary processing of the actual equipment status, reduce user confusion, and thus improve after-sales processing efficiency and user experience.

[0094] In some examples, a second machine learning model can be applied to the monitoring data to identify preset anomaly patterns. In other examples, the analyte curves corresponding to the monitoring data can be input into a deep learning-based second machine learning model to identify preset anomaly patterns. Thus, the second machine learning model can automatically learn the corresponding features to identify preset anomaly patterns. In some examples, features can be extracted from the monitoring data, and the second machine learning model can be applied to the extracted features to identify preset anomaly patterns. A description of the second machine learning model can be found in the first machine learning model. In some examples, the second machine learning model may not be used.

[0095] In some examples, the preset abnormal modes may include at least one of a persistent low value mode, a no-fluctuation mode, and a jump point mode. Additionally, the persistent low value mode, the no-fluctuation mode, and the jump point mode may be abnormal modes unrelated to device malfunction. In some examples, the persistent low value mode, the no-fluctuation mode, and the jump point mode may be identified sequentially; if one preset abnormal mode is identified, the process exits.

[0096] In addition, a sustained low-value pattern can refer to monitoring data that remains consistently low over a certain period. A fluctuation-free pattern can refer to monitoring data with low fluctuation range over a certain period. That is, a fluctuation-free pattern can refer to an ultra-stable pattern that does not conform to normal physiological laws. Normal physiological indicators have natural fluctuation characteristics; excessive stability may actually be a false reading caused by external factors. A jump-point pattern can refer to a large difference between adjacent monitoring values. A jump-point pattern can include short-term data jumps caused by physical interference (such as squeezing, scraping, or loosening).

[0097] In some examples, when a pre-defined abnormal pattern is identified, corresponding explanatory information can be obtained. This explanatory information can be used to explain to the user the reasons for the occurrence of the pre-defined abnormal pattern that are unrelated to equipment malfunction. In this case, it can reduce the likelihood of similar problems occurring when users use new monitoring equipment in the future, thereby improving the user experience and reducing the number of users requesting after-sales service for similar issues. For example, the explanatory information for a persistently low value pattern could be "Persistently low values ​​are more likely to be caused by physiological factors or medication effects, rather than equipment malfunction." For example, the explanatory information for a fluctuation-free pattern could be "Extremely stable readings are more likely to be related to diet, rest, activity levels, or medication factors, rather than equipment malfunction." For example, the explanatory information for a jump-point pattern could be "Short-term continuous jumps are usually caused by physical interference to the sensor, which is an external factor rather than a malfunction of the equipment itself."

[0098] Figure 5 This is an exemplary flowchart illustrating the identification of persistent low value patterns as described in this disclosure.

[0099] In some examples, reference Figure 5 Identifying persistent low value patterns may include the following steps.

[0100] Step S301: Set the first sliding window (e.g., 4 hours or 48 data points). The first sliding window can slide across the monitored data. In some examples, the step size of the first sliding window can be 1 data point.

[0101] Step S302: Determine whether all monitored values ​​within the first sliding window are not higher than the first monitoring value threshold, and determine whether there are any data breakpoints within the first sliding window that exceed the first preset time period (e.g., 30 minutes). In this case, it is possible to identify data points with excessively low monitored values ​​while improving data continuity, thereby improving the accuracy of identifying continuous low value patterns.

[0102] Step S303: In response to the existence of a target sliding window, acquire the second monitoring data within the most recent preset time period relative to the current time, and determine the highest monitoring value in the second monitoring data. The target sliding window can be a first sliding window where all monitoring values ​​are not higher than a first monitoring value threshold and there are no data breakpoints exceeding a first preset time duration. Furthermore, the time span of the preset time period in step S303 can be greater than that of the first sliding window. For example, the preset time period can be 24 hours.

[0103] Step S304: If the highest monitored value is not higher than the second monitored value threshold, a persistent low value pattern is determined. In this case, monitoring values ​​over a longer time range are used to further confirm the existence of the persistent low value pattern, thereby improving the accuracy of identifying the persistent low value pattern. Furthermore, the second monitored value threshold can be greater than the first monitored value threshold. For example, the first monitored value threshold can be 2.8 mmol / L, and the second monitored value threshold can be 7.8 mmol / L.

[0104] Figure 6 This is an exemplary flowchart illustrating the identification of non-fluctuation patterns as described in this disclosure.

[0105] In some examples, reference Figure 6 Identifying a non-fluctuation pattern may include the following steps.

[0106] Step S401: Set a second sliding window (e.g., 8 hours). The second sliding window can slide on the monitoring data.

[0107] Step S402: Obtain multiple monitoring values ​​within the second sliding window.

[0108] Step S403: Determine the minimum monitoring value among the multiple monitoring values ​​and the range of the multiple monitoring values. The range can be the absolute value of the difference between the maximum and minimum values.

[0109] Step S404: If the minimum monitored value is not lower than the third monitored value threshold (e.g., 4.5 mmol / L) and the range is not greater than the fluctuation threshold (e.g., 1.0 mmol / L), a non-fluctuation pattern is determined to exist. This improves the accuracy of identifying non-fluctuation patterns.

[0110] Figure 7 This is an exemplary flowchart illustrating the identification of jump point patterns as described in this disclosure.

[0111] In some examples, reference Figure 7 Identifying jump point patterns can include the following steps.

[0112] Step S501: Determine the absolute difference between adjacent monitoring values ​​in the monitoring data.

[0113] Step S502: In response to an absolute difference greater than a jump threshold (e.g., 3.0 mmol / L), the corresponding monitored value is determined as a jump point. In some examples, a counter can be maintained; if the current point is a jump point, the counter is incremented by 1; if the current point is not a jump point, the counter is reset to zero.

[0114] Step S503: If the number of consecutive jump points is not less than a preset number (e.g., 3), it is determined that a jump point pattern exists.

[0115] In some examples, a reference is returned. Figure 4 In step S204, a user interface can be provided to receive the first image data. For example, if a preset abnormal pattern is not detected, a user interface can be output for the user to submit the first image data. After the user submits the first image data through the user interface, the first image data can be received. Alternatively, the first image data can be supporting data related to data deviation.

[0116] In some examples, the first image data may include images of monitoring data and control data. This allows for the identification and comparison of monitoring and control values. For example, the first image data may include images of the data interface of a continuous glucose meter and images of the data interface of a finger-prick glucose meter. In some examples, the first image data may consist of multiple sets of data.

[0117] In some examples, image data submitted in after-sales requests (such as first or second image data) can be identified by image type, and processed accordingly. In this case, different image types can be processed according to different logics, thereby improving the effectiveness of image analysis. Furthermore, it helps to provide analysis results with different granularities for different after-sales scenarios (e.g., providing deviation analysis results related to deviation analysis for physiological data deviations, and providing image review results related to the confidence level of the fault type for equipment appearance).

[0118] In some examples, image type can be used to check if the image type matches the after-sales service type, prompting the user to replace the image if they don't match. For example, if the after-sales service type is data deviation, but the image type is not a physiological data image, it can be considered a mismatch. If the after-sales service type is sensor implantation failure, but the image type is not a device appearance image, it can also be considered a mismatch. In such cases, matching based on image type can improve the stability of the after-sales processing flow.

[0119] In some examples, the image type may include at least one of physiological data images (also known as analyte data images), device appearance images, device status images, and other images. For example, device appearance images may include photos related to sensor implantation failures (e.g., photos of device damage, patch malfunctions, or packaging defects). Device status images may include photos of a page displaying the current status of the device.

[0120] In some examples, reference Figure 4In step S205, deviation verification information can be extracted from the first image data based on optical character recognition (OCR). Specifically, the image type corresponding to the first image data can be identified; in response to the image type being a physiological data image, character information in the first image data is obtained through optical character recognition; and deviation verification information is extracted from the character information. In this case, the image type can be automatically identified and information for deviation analysis can be extracted, thereby improving data processing efficiency. In addition, key information can be obtained without manual intervention, which can improve the automation and reliability of after-sales access judgment.

[0121] In some examples, reference Figure 4 In step S206, the deviation analysis results can be used to indicate whether a data deviation exists. In some examples, the deviation analysis results may include whether a deviation exists or not.

[0122] In some examples, determining the outcome of a deviation analysis may include acquiring at least two sets of data; determining the no-bias condition; counting the number of data sets that meet the no-bias condition from the at least two sets of data; and determining the deviation analysis outcome based on the number of data sets. Each set of data may include a monitored value, first-time information, a control value, and second-time information.

[0123] In some examples, determining the deviation analysis results may also include obtaining the activation duration of monitoring device 100. In some examples, the no-bias condition may be related to the activation duration of monitoring device 100. Specifically, the no-bias condition may include: in response to an activation duration of a first time period, the absolute difference between the monitored value and the control value does not exceed a first deviation threshold (e.g., 7 mmol / L); in response to an activation duration of a second time period, the monitored value is within a preset deviation range determined by the control value. Furthermore, the deviation range corresponding to the preset deviation range is smaller than the deviation range corresponding to the first deviation threshold, and the second time period may be later than the first time period. For example, the first time period may be within 48 hours, and the second time period may be after 48 hours. In this case, by distinguishing between early lenient standards and later precise standards, the reliability and accuracy of after-sales access judgment can be improved, while unnecessary human intervention can be reduced.

[0124] In some examples, the preset deviation range may include a preset percentage interval and a preset fixed interval. In some examples, in response to a control value greater than a first control threshold, the monitored value may fall within the preset percentage interval corresponding to the control value (e.g., 80%-120%). In some examples, in response to a control value not greater than the first control threshold, the monitored value may fall within the preset fixed interval corresponding to the control value (e.g., ±1.1 mmol / L). Thus, segmented judgment can improve the accuracy of deviation analysis.

[0125] In some examples, determining the deviation analysis result based on the number of data sets may include determining that there is no deviation when the number of data sets is greater than a preset number (e.g., 1 set).

[0126] In some examples, before determining whether each set of data meets the unbiased condition, it can be determined whether the first and second time information of each set of data corresponds (i.e., whether they can be considered the same time). If the correspondence between the first and second time information of each set of data is determined, then it can be determined whether each set of data meets the unbiased condition. In this case, being able to compare the monitored values ​​and the control values ​​when they correspond at different times can improve the effectiveness of the deviation analysis results.

[0127] In some examples, reference Figure 4 In step S207, in response to the deviation analysis result indicating no deviation, the second judgment result can be determined as failing. In some examples, in response to the deviation analysis result indicating a deviation, the second judgment result can be determined as passing. In some examples, in response to the deviation analysis result indicating a deviation, the second judgment result can also be determined in conjunction with target information (described later). This further improves the accuracy of the second judgment result.

[0128] Figure 8 This is an exemplary flowchart illustrating an after-sales request for handling a failed sensor implantation, as described in this disclosure.

[0129] In some examples, reference Figure 8 The process for handling after-sales requests related to sensor implantation failure can include: responding to the after-sales request data indicating sensor implantation failure as the after-sales type, obtaining second image data submitted by the user (step S601); identifying the image type corresponding to the second image data (step S602); responding to the image type being a device appearance image, performing fault identification on the device appearance image based on a visual language model to obtain an image review result (step S603); and determining a second judgment result based on the image review result (step S604). Therefore, using a visual language model for fault identification can improve the judgment efficiency of after-sales requests related to sensor implantation failure.

[0130] In some examples, a user interface may be provided in step S601 to receive the second image data. Refer to the description of the first image data for details. Alternatively, the second image data may be evidentiary data related to sensor implantation failure.

[0131] In some examples, the image review result may be related to factual information corresponding to the second image data, rather than directly determining whether the after-sales request is approved. That is, it does not directly decide whether the after-sales request is approved. In this case, a more factual result can be output, which, when combined with other data for subsequent judgment, can reduce the risk of making after-sales conclusions directly from the visual language model. In some examples, the image review result may include the fault type and the confidence level of the fault type. In some examples, the second judgment result can be determined based on the confidence level of the fault type. In some examples, the confidence level of the fault type can be any value between 0 and 10.

[0132] In some examples, a second judgment result can be determined as pending review if the confidence level for the fault type is less than a first preset confidence level but greater than a second preset confidence level. In some examples, a second judgment result can be determined as failing if the confidence level for the fault type is not greater than the second preset confidence level.

[0133] In some examples, a second judgment result can be determined as pass if the confidence level of the fault type is not less than a first preset confidence level. In some examples, a second judgment result can also be determined by combining target information (described later) if the confidence level of the fault type is not less than the first preset confidence level.

[0134] In addition, the first pre-set reliability can be greater than the second pre-set reliability. For example, a low score may indicate insufficient evidence and failure of image review, with the second judgment result being "failure"; a medium score may indicate that some evidence exists but requires manual review, with the second judgment result being "pending review"; and a high score may indicate sufficient evidence and approval of image review.

[0135] In some examples, handling after-sales requests for sensor malfunctions may include determining whether an abnormal state of the monitoring device 100 (such as sensor detachment or sensor failure) qualifies for after-sales service.

[0136] In some examples, handling after-sales requests for sensor malfunctions may include, in response to an after-sales request of sensor malfunction, determining a second judgment result by confirming whether the identified malfunction of the monitoring device 100 is temporary or a genuine malfunction. In some examples, the malfunction identification result may be determined based on the identified malfunction of the monitoring device 100 and a recovery time window, and the second judgment result may be determined based on the malfunction identification result. This avoids misjudging situations that can recover on their own as requiring after-sales service.

[0137] In some examples, the abnormal status of monitoring device 100 may not be input by the user. In other examples, the abnormal status of monitoring device 100 can be automatically identified by the after-sales system 1. In this case, the accuracy of device abnormality judgment can be improved, user input can be reduced, and after-sales processing efficiency can be increased.

[0138] Additionally, anomaly identification results can be used to indicate whether an anomaly is temporary or a genuine anomaly. In some examples, anomaly identification results may include both definitive and undetermined anomalies.

[0139] Additionally, the recovery time window refers to the time window during which the monitored device 100 can recover on its own after malfunctioning. For example, the recovery time window could be 1 hour, 2 hours, 3 hours, 4 hours, or 6 hours. Within the recovery time window, it can be assumed that the device malfunction still has the potential to recover on its own, thus temporarily disqualifying it from after-sales service. In some examples, different malfunction states can correspond to different recovery time windows. This reduces user waiting time and minimizes misjudgments.

[0140] In some examples, the abnormal state may include an initialization abnormality. An initialization abnormality may be a sensor abnormality that occurs during the initialization phase. In some examples, a post-sales request to handle an initialization abnormality may include, in response to the monitoring device 100 being in the initialization phase, determining whether the duration of the initialization abnormality is greater than a recovery time window; if the duration is greater than the recovery time window, determining the abnormal state identification result as a confirmed abnormality; if the duration is not greater than the recovery time window, determining the abnormal state identification result as an unconfirmed abnormality.

[0141] In some examples, handling after-sales requests for sensor anomalies during the usage phase may include, in response to the monitoring device 100 being in use, determining the latest monitoring point in the monitoring data; obtaining the device status and time information corresponding to the latest monitoring point; in response to the device status being an abnormal state related to sensor 101, determining whether the time difference between the current time and the time information is greater than a recovery time window; if the time difference is greater than the recovery time window, determining the abnormal state identification result as a confirmed anomaly; if the time difference is not greater than the recovery time window, determining the abnormal state identification result as an unconfirmed anomaly. This improves the accuracy of determining whether a sensor anomaly has occurred during the usage phase. Furthermore, the monitoring data used to obtain the latest monitoring point can be the latest monitoring data from the monitoring device 100.

[0142] In some examples, after-sales requests for sensor malfunctions can be handled differently based on the device model. In some examples, during the initialization phase, in response to the device model indicating that monitoring device 100 is a modular device, the system automatically identifies whether physical assembly is complete. If complete, the aforementioned initialization malfunction judgment is performed. If not complete, the number of assembly attempts is counted. If the number of attempts is less than a first preset number, the malfunction is determined as an undetermined malfunction, and self-help guidance is provided. If the number of attempts is not less than the first preset number but less than a second preset number, the malfunction is determined as an undetermined malfunction, and an entry point for transferring to human assistance is provided. If the number of attempts is greater than the second preset number, the malfunction is determined as a confirmed malfunction, and an assembly failure message is displayed. The second preset number can be greater than the first preset number. That is, the first and second preset numbers can identify whether the user's attempts are in the early stages, intermediate stages, or multiple failed attempts to determine the malfunction and provide corresponding assistance to the user.

[0143] In some examples, during the usage phase, in response to the device model indicating that monitoring device 100 is an integrated device, the abnormal state identification result can be determined as "undetermined abnormality" even if no monitoring data is acquired. In some examples, in response to the device model indicating that monitoring device 100 is a separate device, in the case where no monitoring data is acquired and the after-sales system 1 has already indicated a sensor abnormality, the abnormal state identification result can be determined as "determined abnormality." This enables differentiated processing for different device types in the event of data loss.

[0144] Furthermore, an integrated device can refer to a monitoring device 100 that is assembled before being worn. A split device can refer to a monitoring device 100 in which one part (e.g., electronic component 102) is assembled with another part (e.g., sensor base) after it is worn. Thus, split devices can reuse one part, reducing monitoring costs for users.

[0145] In some examples, a second judgment result can also be determined based on at least one of the following target information: the risk level of the supply channel corresponding to the monitoring device 100, the user risk score (i.e., user profile score), and the fault cause classification corresponding to the after-sales type. In this case, at least one factor such as the results of the previous judgment (e.g., image review results or deviation analysis results), risk level, user risk score, and fault cause classification can be integrated into the final judgment, thereby improving the reliability of the after-sales access judgment.

[0146] As mentioned above, in some examples, the second judgment result can also be determined by combining the target information. In some examples, the second judgment result can be determined based on the intermediate judgment result corresponding to each after-sales type and the target information. In some examples, in response to the intermediate judgment result indicating that the after-sales request has been approved, the second judgment result can be determined based on the intermediate judgment result corresponding to each after-sales type and the target information. That is, if all the preceding judgments have passed, the second judgment result is then determined by combining the target information.

[0147] Furthermore, intermediate judgment results can be the final judgment result for each after-sales type before determining the second judgment result. For example, for after-sales type data deviation, the final judgment result can be the deviation analysis result. For after-sales type sensor implantation failure, the final judgment result can be the image review result. For after-sales type sensor anomaly, the final judgment result can be the anomaly status identification result.

[0148] In some examples, a third machine learning model can be applied to the intermediate judgment results and target information to determine the second judgment result. A description of the third machine learning model can be found in the documentation for the first machine learning model. In some examples, a third machine learning model may not be used. For instance, rule matching can be performed based on the intermediate judgment results and target information to determine the second judgment result.

[0149] In some examples, the second judgment result is determined to be pending review because the risk level is higher than the preset risk level. In some examples, the second judgment result is determined to be pending review because the user's risk score is greater than the preset risk score. In some examples, the second judgment result is determined to be pending review because the fault cause classification belongs to the preset manual intervention category (i.e., a situation requiring manual intervention).

[0150] In some examples, for various judgments involved in the after-sales processing method, if the current judgment fails, the process can terminate and output the result of the current judgment, without continuing with subsequent judgments; if the current judgment succeeds, the process can continue to the next judgment until the result of the second judgment is obtained. This improves after-sales processing efficiency and avoids unnecessary data queries.

[0151] In some examples, when information is missing for various judgments involved in after-sales processing, the second judgment result can be determined as pending review. This reduces misjudgments caused by automated handling of special cases.

[0152] In some examples, a reference is returned. Figure 3 The after-sales processing method may include processing the after-sales work order based on the second judgment result (step S105). In some examples, processing the after-sales work order may include not creating the after-sales work order, creating the after-sales work order, and / or updating the work order status of the after-sales work order.

[0153] In some examples, when the second judgment result indicates that an after-sales work order needs to be created, if the after-sales work order corresponding to the after-sales request already exists, the work order status can be updated; otherwise, the after-sales work order can be created. In some examples, in response to the need to receive user input data (such as first image data or second image data) in the after-sales request, an after-sales work order can be created, and the work order status can be set to draft.

[0154] In some examples, in response to a "pass" result in the second judgment, the after-sales work order can be processed to initiate the fulfillment process. For example, the work order status can be set to "passed" to initiate the fulfillment process. In some examples, in response to a "pending review" result in the second judgment, the after-sales work order can be processed to mark its status as "pending review" (e.g., pending manual review). In some examples, in response to a "fail" result in the second judgment, the after-sales work order may not be created or the status of an already created after-sales work order may be marked as "invalid."

[0155] In some examples, the status of an after-sales work order can include at least one of the following: draft, under after-sales access assessment, rejected, pending review (e.g., pending manual review), under review (e.g., under manual review), review failed, approved, awaiting shipment, in transit, completed, and voided. In some examples, the initial status of an after-sales work order can be draft. In some examples, when a new after-sales work order for monitoring device 100 is generated, the status of an existing after-sales work order can be changed from draft to void. Additionally, draft can refer to the work order status before it has been formally submitted.

[0156] In some examples, an ongoing after-sales work order can be an incomplete after-sales work order. For example, an ongoing after-sales work order can include an unfinished after-sales work order that is pending review, under review, approved, awaiting shipment, in transit, or otherwise not in a draft state.

[0157] In some examples, after a work order is approved, it can enter the fulfillment process. Specifically, depending on the logistics situation, after-sales work orders can sequentially enter the pending shipment, in transit, and completed work order statuses.

[0158] In some examples, in response to a second judgment result of "fail" and / or "pending review," a prompt message corresponding to the second judgment result can be output. In some examples, the prompt message may include at least one of a first prompt message, a second prompt message, and a third prompt message. The first prompt message may include a result identifier, which can uniquely identify a second judgment result and be used for internal circulation and statistics within the after-sales system 1. The second prompt message may include a user-oriented natural language description, which explains the current second judgment result and suggests the next step. The language of the second prompt message should be easy to understand and avoid technical jargon. The third prompt message may include professional scripts for human customer service, such as explanations of technical terms and suggestions. This helps human customer service representatives better explain complex reasons for rejection to users.

[0159] In some examples, the prompt message can simultaneously include a first prompt message, a second prompt message, and a third prompt message. In this case, the session client 10 can display the second prompt message, the human customer service client 20 can display the third prompt message, and the backend system can use the first prompt message for statistics and processing. That is, the three-tiered prompt message can provide different granularities of information content to the user, the human customer service client, and the backend system, respectively. This avoids information compression or loss during processing (e.g., not reducing the three-tiered prompt message to a single Boolean value or a simple reason text). In addition, different receiving ends can perform corresponding processing based on the information at the corresponding level, thereby improving the information transmission efficiency and after-sales processing efficiency of the after-sales system 1.

[0160] Figure 9 This is an exemplary block diagram illustrating an electronic device 8 as described in this disclosure.

[0161] The examples of this disclosure also relate to an electronic device 8, see reference 8. Figure 9 The electronic device 8 may include a processor 801 (e.g., a central processing unit or a graphics processing unit) and a memory 802. The memory 802 may store a computer program that, when executed by the processor 801, implements one or more steps of the after-sales processing method described above.

[0162] In some examples, memory 802 may include a computer-readable storage medium. The computer-readable storage medium may, for example, include volatile memory and / or non-volatile memory. Volatile memory may, for example, include random access memory (RAM) and / or cache memory, etc. For example, computer-executable instructions may be loaded from memory unit 807 (described later) into random access memory (RAM) to execute computer-executable instructions. Non-volatile memory may, for example, include read-only memory (ROM), hard disk, USB storage, flash memory, etc.

[0163] In some examples, reference Figure 9 The processor 801 and memory 802 can be connected to each other via bus 803. In some examples, the electronic device 8 may also include an input / output interface 804, which can be connected to bus 803. In some examples, the electronic device 8 may also include an input unit 805 (e.g., touch screen, keyboard, mouse, camera, microphone, etc.), an output unit 806 (e.g., display, speaker, etc.), a storage unit 807 (e.g., magnetic tape, hard disk, flash memory, etc.), and / or a communication unit 808, which can be connected to the input / output interface 804. Additionally, the communication unit 808 can be used to enable the electronic device 8 to communicate wirelessly or wiredly with other devices.

[0164] Examples of this disclosure also disclose a computer-readable storage medium that can store at least one instruction, which, when executed by a processor, implements one or more steps of the after-sales processing method described above. The computer-readable storage medium can be, but is not limited to, any type of disk, including floppy disks, optical disks, DVDs, CD-ROMs, microdrives, and magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic cards or optical cards, nanosystems (including molecular memory ICs), or any type of medium or device suitable for storing instructions and / or data.

[0165] While the present disclosure has been specifically described above in conjunction with the accompanying drawings and examples, it is to be understood that the foregoing description does not limit the present disclosure in any way. Those skilled in the art can make modifications and variations to the present disclosure as needed without departing from its essential spirit and scope, and all such modifications and variations shall fall within the scope of the present disclosure.

Claims

1. A method for after-sales service of an analyte monitoring device, characterized in that, include: Obtain after-sales request data corresponding to the user's after-sales request and multi-dimensional data associated with the after-sales request data, wherein the multi-dimensional data includes user dimension data, device dimension data and channel dimension data; A preliminary judgment is made based on the multi-dimensional data to obtain a first judgment result; in response to the first judgment result indicating that the preliminary judgment is passed and the after-sales type in the after-sales request data is a data deviation, the monitoring data corresponding to the analytical monitoring device is obtained; Based on the monitoring data, it is possible to identify whether there is a preset abnormal pattern. The preset abnormal patterns include a continuous low value pattern, a no-fluctuation pattern, and a jump point pattern. In response to the existence of the preset abnormal mode, the second judgment result is determined to be passed; In response to the absence of the preset abnormal mode, the first image data submitted by the user is obtained; Extract the monitoring value and corresponding first time information, as well as the control value and corresponding second time information from the first image data; The deviation analysis result is determined based on the monitored value, the first time information, the control value, and the second time information; the second judgment result is determined based on the deviation analysis result; and the after-sales work order is processed according to the second judgment result.

2. The after-sales service method according to claim 1, characterized in that, Identifying the persistent low value pattern includes: Set the first sliding window; Determine whether the monitored values ​​of the monitored data within the first sliding window are all not higher than the first monitoring value threshold, and determine whether there are any data breakpoints within the first sliding window that exceed the first preset time. In response to the existence of a target sliding window, the second monitoring data within the most recent preset time period is obtained, and the highest monitoring value in the second monitoring data is determined. The target sliding window is the first sliding window in which all monitoring values ​​are not higher than the first monitoring value threshold and there are no data breakpoints exceeding the first preset time period. If the highest monitored value is not higher than the second monitored value threshold, the persistent low value pattern is determined to exist, wherein the second monitored value threshold is greater than the first monitored value threshold.

3. The after-sales service method according to claim 1, characterized in that, Identifying the fluctuation-free pattern includes: Set a second sliding window; Obtain multiple monitoring values ​​of the monitoring data within the second sliding window; Determine the minimum monitoring value among the plurality of monitoring values ​​and the range of the plurality of monitoring values; If the minimum monitored value is not lower than the third monitored value threshold and the range is not greater than the fluctuation threshold, then the absence of fluctuation mode is determined.

4. The after-sales service method according to claim 1, characterized in that, At least two sets of data are acquired, each set of data including the monitoring value, the first time information, the reference value, and the second time information, wherein the monitoring value, the first time information, the reference value, and the second time information are extracted from the first image data based on optical character recognition; The activation duration of the analyte monitoring device is obtained and a no-bias condition is determined. The no-bias condition includes: in response to the activation duration being in a first time period, the absolute difference between the monitored value and the control value does not exceed a first deviation threshold; in response to the activation duration being in a second time period, the monitored value is within a preset deviation range determined by the control value, and the second time period is later than the first time period. Count the number of data sets that satisfy the unbiased condition among the at least two sets of data; If the number of data sets is greater than the preset number of sets, the deviation analysis result is determined to be that there is no deviation.

5. The after-sales service method according to claim 1, characterized in that, Also includes: In response to the first judgment result indicating that the preliminary judgment is passed and the after-sales type in the after-sales request data is sensor implantation failure, the second image data submitted by the user is obtained; Identify the image type corresponding to the second image data; In response to the image type being a device appearance image, fault identification is performed on the device appearance image based on a visual language model to obtain an image review result, the image review result including the fault type and the confidence level of the fault type; The second judgment result is determined based on the confidence level of the fault type.

6. The after-sales service method according to claim 1, characterized in that, Also includes: In response to the first judgment result indicating that the preliminary judgment is passed and the after-sales type in the after-sales request data is sensor abnormality, the abnormality identification result is determined based on the identified abnormality status of the analytical monitoring device and the recovery time window; The second judgment result is determined based on the abnormal state identification result.

7. The after-sales service method according to claim 6, characterized in that, In response to the analyte monitoring device being in use, the latest monitoring point in the monitoring data of the analyte monitoring device is determined; Obtain the device status and time information corresponding to the latest monitoring point; In response to the device state being an abnormal state related to the sensor, determine whether the time difference between the current time and the time information is greater than the recovery time window; If the time difference is greater than the recovery time window, the abnormal state identification result is determined to be an abnormality. If the time difference is not greater than the recovery time window, the abnormal state identification result is determined to be an undetermined abnormality.

8. The after-sales service method according to claim 1, characterized in that, The user dimension data includes at least one of the user's historical after-sales records and device activation records. The preliminary judgment based on the user dimension data includes at least one of the following: determining whether the number of devices corresponding to consecutively passed after-sales records exceeds a first threshold and whether the usage time of the corresponding analyte monitoring devices is greater than a second preset duration based on the user's historical after-sales records; counting the number of consecutively passed after-sales records with the same fault type as the after-sales request data within a preset time range in the user's historical after-sales records, and determining whether the number exceeds a second threshold; determining whether the number of passed after-sales records exceeds a third threshold based on the historical after-sales records of users related to the analyte monitoring devices; and obtaining the device activation records corresponding to the passed after-sales records in the user's historical after-sales records, and determining whether the corresponding analyte monitoring devices are reactivated after the after-sales service is completed based on the device activation records and the passed after-sales records; and / or The channel dimension data includes at least one of channel type and channel configuration information. The preliminary judgment based on the channel dimension data includes: obtaining the channel type corresponding to the analyte monitoring device; obtaining the channel configuration information corresponding to the channel type; determining the after-sales restriction type corresponding to the channel type based on the channel configuration information; in response to the after-sales restriction type indicating permission, determining the result of the preliminary judgment based on the channel dimension data as passed; in response to the after-sales restriction type indicating a restriction on the number of after-sales services, determining whether the number of after-sales services corresponding to the analyte monitoring device has reached a preset number; and if the preset number has not been reached, determining the result of the preliminary judgment based on the channel dimension data as passed; and in response to the after-sales restriction type indicating prohibition, determining the first judgment result as failed; and / or The device dimension data includes at least one of the following: device identifier, activation status, time-related validity information, and historical after-sales records of the analyte monitoring device. The preliminary judgment based on the device dimension data includes at least one of the following: determining whether there is an analyte monitoring device corresponding to the device identifier based on the device identifier; determining whether the activation status matches the after-sales type based on the activation status; determining whether the analyte monitoring device is within the warranty period and the product validity period based on the validity information; and determining whether there is an ongoing after-sales work order associated with the analyte monitoring device based on the historical after-sales records of the analyte monitoring device.

9. The after-sales processing method according to any one of claims 1 to 8, characterized in that, The second judgment result is also determined based on at least one of the following target information: The risk level of the supply channel corresponding to the analyte monitoring equipment; User risk assessment; The fault cause classification corresponding to the after-sales type.

10. An electronic device, characterized in that, It includes a processor and a memory, the memory storing a computer program that, when executed by the processor, implements the after-sales processing method as described in any one of claims 1 to 9.

Images