Remote warning and monitoring system and method for insulin pump

Abstract

The invention discloses a remote warning and monitoring system and method for an insulin pump, and relates to the technical field of medical equipment monitoring, and the system comprises a data collection module which is used for obtaining multi-modal physiological and equipment data from user side equipment in real time; the risk assessment engine is used for performing fusion analysis based on the multi-modal physiological and equipment data to generate a comprehensive risk assessment result; the dynamic early warning module is used for dynamically triggering warning information of different risk levels according to the comprehensive risk assessment result; the remote communication module is used for simultaneously pushing the warning information to at least one remote monitoring terminal and user side equipment; and the remote intervention interface is integrated on the remote monitoring terminal and is used for receiving and displaying the warning information and enabling a remote guardian to initiate at least one intervention instruction. Alarm omission caused by the fact that a patient is alone or the response ability is limited is effectively avoided; the accuracy and reliability of risk assessment are improved, and false alarms are reduced; the risk response time is shortened, and the safety of patients is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of medical device monitoring technology, and more specifically, to a remote alert monitoring system and method for insulin pumps. Background Technology

[0002] Insulin pumps, as continuous subcutaneous infusion devices that mimic the function of the human pancreas, have become an important tool for diabetes management when used in conjunction with continuous glucose monitoring systems. However, in practical applications, patients are often asleep, alone, or in situations with limited responsiveness, such as children at school or elderly people alone at home. Critical alarms from existing devices, such as those for hypoglycemia, hyperglycemia, or catheter blockage, are difficult to detect in a timely manner. Current market solutions primarily push physiological data to the patient's personal smartphone, and their alarm mechanisms only function on the local device terminal, failing to overcome geographical limitations to notify remote caregivers. Some data sharing functions only display basic information, lacking in-depth correlation analysis of multimodal data such as real-time blood glucose levels, predicted blood glucose trends, remaining insulin infusion volume, and device blockage signals. Alarm triggering generally relies on a single threshold judgment, failing to integrate device operating status and physiological parameters for comprehensive risk quantification, leading to frequent false alarms and an inability to predict potential crises. More notably, even if a remote monitor receives an abnormal notification, the system does not provide a direct intervention channel. The monitor cannot send reminder messages to the patient's device, trigger the device to sound an alarm for location tracking, or establish a real-time voice connection. They can only contact the patient indirectly through external communication methods, which significantly prolongs the risk response time and exposes the patient to serious health threats.

[0003] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0004] In view of this, the present invention provides a remote warning monitoring system and method for insulin pumps, which enables remote real-time monitoring and immediate intervention during the use of insulin pumps, effectively avoiding missed alarms due to patients being alone or having limited response capabilities; through the fusion analysis of multimodal physiological and device data, the accuracy and reliability of risk assessment are improved, and false alarms are reduced; it provides a direct intervention channel for remote monitors, significantly shortens risk response time, and ensures patient safety.

[0005] In a first aspect, the present invention provides a remote alert monitoring system for insulin pumps, comprising: The data acquisition module is used to acquire multimodal physiological and device data from user-end devices in real time; the user-end devices include at least an insulin pump and a continuous glucose monitor. The risk assessment engine, which is connected in communication with the data acquisition module, is used to perform fusion analysis based on the multimodal physiological and device data to generate a comprehensive risk assessment result. The dynamic early warning module is connected in communication with the risk assessment engine and is used to dynamically trigger warning information of different risk levels based on the comprehensive risk assessment results. A remote communication module, which is communicatively connected to the dynamic early warning module, is used to simultaneously push the warning information to at least one remote monitoring terminal and the user terminal device; A remote intervention interface, integrated into the remote monitoring terminal, is used to receive and display the warning information and to allow the remote monitor to initiate at least one intervention command; the intervention command is fed back to the user terminal device through the remote communication module.

[0006] In one optional implementation, the multimodal physiological and device data includes real-time blood glucose levels, predicted blood glucose trend, remaining insulin infusion volume, current basal infusion rate, information on infused high doses, insulin pump blockage alarm signals, insulin pump power information, and CGM signal strength information.

[0007] In one alternative implementation, the risk assessment engine is configured to perform the following operations: Based on real-time blood glucose values ​​and predicted blood glucose trends, calculate the hypoglycemia risk index and hyperglycemia risk index for a specific future time period. Based on the remaining insulin infusion volume, current basal infusion rate, information on the large doses already infused, insulin pump blockage alarm signal, insulin pump power information, and CGM signal strength information, the equipment failure risk index is calculated. The hypoglycemia risk index, hyperglycemia risk index, and equipment failure risk index are weighted and fused using a fusion algorithm to generate the comprehensive risk assessment result; the comprehensive risk assessment result represents different risk levels.

[0008] In one optional implementation, the dynamic early warning module is configured as follows: Multiple warning levels corresponding to different risk levels are pre-stored; When the comprehensive risk assessment result reaches the first risk level but does not reach the second risk level, a reminder-level alert is triggered, which is either silent or a mild prompt. When the comprehensive risk assessment result reaches the second risk level but does not reach the third risk level, a warning level alert is triggered, which includes an audio-visual prompt. When the comprehensive risk assessment result reaches the third risk level, a critical level warning is triggered, which includes continuous strong reminders and recommended measures.

[0009] In one optional implementation, the intervention instructions provided by the remote intervention interface include at least one of the following: sending a preset comfort or reminder message to the user terminal device, triggering the buzzer of the user terminal device to sound in order to locate the device, and initiating a real-time voice call connection with the user terminal device.

[0010] In one alternative implementation, the system further includes: The data anonymization module is located between the data acquisition module and the remote communication module. It is used to de-identify the data before sending it to the remote monitoring terminal to remove the user's personal identity information.

[0011] Secondly, the present invention also provides a remote alert monitoring method for an insulin pump, comprising the following steps: S1. Acquire multimodal physiological and device data in real time from user terminal devices; S2. Based on the multimodal physiological and device data, perform fusion analysis to generate a comprehensive risk assessment result; S3. Based on the comprehensive risk assessment results, dynamically trigger warning messages of different risk levels; S4. The warning information is simultaneously pushed to at least one remote monitoring terminal and the user terminal device; S5. Receive and display the warning information through the remote monitoring terminal, and in response to the operation of the remote monitor, send the intervention command to the user terminal device.

[0012] In one optional implementation, step S2 specifically includes the following steps: Calculate the hypoglycemia risk index and hyperglycemia risk index based on blood glucose data; Calculate the equipment risk index based on user-end device operation data; The hypoglycemia risk index, hyperglycemia risk index, and equipment failure risk index are weighted and fused using a fusion algorithm to generate the comprehensive risk assessment result; the comprehensive risk assessment result represents different risk levels.

[0013] In one optional implementation, step S3 specifically includes the following steps: The comprehensive risk assessment result is compared with multiple preset risk thresholds; each preset risk threshold corresponds to a different risk level. Based on the comparison results, select a corresponding level from multiple predefined risk levels to trigger an alert, which includes at least one non-critical warning level and one critical level.

[0014] In one alternative implementation, in S5, the intervention instruction includes initiating a real-time voice call with the user terminal device; The method also includes: temporarily pausing or reducing the intensity of the audio-visual alerts pushed to the remote monitoring terminal after the call is established.

[0015] As can be seen from the above, the remote warning monitoring system and method for insulin pumps provided by the present invention, through the integration of multimodal data acquisition, fusion analysis and dynamic risk assessment mechanisms, realizes real-time monitoring of the patient's physiological state and the equipment's operating status, and supports direct intervention by remote caregivers, effectively solving the problems of alarm omissions, frequent false alarms and delayed intervention in the prior art. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the structure of a remote alert monitoring system for an insulin pump according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating a remote alert monitoring method for an insulin pump according to an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] like Figure 1 As shown in the figure, this application proposes a remote alert monitoring system for insulin pumps, including: The data acquisition module is used to acquire multimodal physiological and device data from user-end devices in real time; the user-end devices include at least an insulin pump and a continuous glucose monitor. The risk assessment engine communicates with the data acquisition module to perform fusion analysis based on multimodal physiological and device data to generate comprehensive risk assessment results. The dynamic early warning module communicates with the risk assessment engine and is used to dynamically trigger warning messages of different risk levels based on the comprehensive risk assessment results. The remote communication module, which communicates with the dynamic early warning module, is used to simultaneously push warning information to at least one remote monitoring terminal and user terminal device; The remote intervention interface, integrated into the remote monitoring terminal, is used to receive and display warning information and to allow the remote monitor to initiate at least one intervention command; the intervention command is fed back to the user terminal device through the remote communication module.

[0020] In practical applications, a data acquisition module can be understood as a functional unit used to extract data from multiple devices. Its main function is to acquire real-time multimodal data from user-end devices. For example, data transmission can be achieved through wired or wireless communication protocols (such as Bluetooth, Wi-Fi, Zigbee), or some device status information can be supplemented by manual input. Its primary purpose is to achieve comprehensive monitoring of the user's physiological state and the device's operational status.

[0021] A risk assessment engine can be understood as a computational unit that performs data analysis and generates results based on algorithms. Specifically, it can process multimodal data through machine learning models, rule engines, or statistical analysis methods. For example, it can use decision tree algorithms to classify blood glucose trends or use time series analysis to predict the likelihood of equipment failure. Its main purpose is to generate quantitative results that reflect comprehensive risk.

[0022] A dynamic early warning module can be understood as a functional unit that dynamically adjusts its output behavior based on input data. Specifically, it can trigger different levels of warning information through preset grading rules or dynamically adjusted thresholds. For example, when a blood glucose level is detected to be approaching a dangerous range, the user can be notified via vibration; when a device malfunction is detected, a pop-up notification can alert the caregiver. Its main purpose is to achieve tiered alarms and reduce unnecessary interference.

[0023] A remote communication module can be understood as a functional unit used to achieve data transmission. Specifically, it can achieve bidirectional data transmission through Internet protocols (such as HTTP and MQTT) or dedicated communication protocols. For example, it can relay data through a cloud server or directly transmit information through point-to-point communication. Its main purpose is to ensure that warning information can be delivered to user-end devices and remote monitoring terminals simultaneously.

[0024] A remote intervention interface can be understood as a functional unit for receiving and responding to user operations. Specifically, it can input intervention commands through a graphical interface, voice interaction, or physical buttons. For example, a caregiver can send a preset message by clicking a button on the interface or trigger a specific operation through a voice command. Its main purpose is to enable remote caregivers to proactively intervene in the user's device.

[0025] The innovation of this invention lies in its integration of data acquisition, risk assessment, dynamic early warning, remote communication, and remote intervention functions to construct a closed-loop monitoring mechanism, solving the problems of passive alarms, limited analysis, and lack of intervention in existing technologies. The data acquisition module provides a comprehensive foundation of input data, the risk assessment engine enables multi-dimensional risk analysis, the dynamic early warning module reduces interference through a tiered alarm mechanism, the remote communication module overcomes the limitations of information transmission, and the remote intervention interface completes the closed loop from monitoring to intervention. These technical features collectively achieve proactive, intelligent, and remote monitoring of insulin pump therapy status.

[0026] The working principle of this invention is as follows: The system acquires multimodal physiological and device data in real time from user-end devices via a data acquisition module. These user-end devices include at least an insulin pump and a continuous glucose monitor, ensuring comprehensive monitoring of both the patient's physiological state and device operation. Furthermore, the risk assessment engine receives multimodal data from the data acquisition module and performs fusion analysis to generate a comprehensive risk assessment result. This avoids the limitations of single-indicator judgments and enables comprehensive prediction of hypoglycemia, hyperglycemia, and device failure risks. A dynamic early warning module dynamically triggers warning messages of different risk levels based on the comprehensive risk assessment results. Specifically, the tiered alarm mechanism ensures timely response to critical situations while reducing unnecessary alarm interference for users. A remote communication module simultaneously pushes warning information to at least one remote monitoring terminal and the user-end device, breaking the limitation of alarms being restricted to the patient's end and enabling remote monitors to obtain abnormal information and take simultaneous action. A remote intervention interface is integrated into the remote monitoring terminal to receive and display warning information and allow remote monitors to initiate at least one intervention command. These commands are fed back to the user-end device via the remote communication module, forming a closed-loop management system from monitoring to intervention. Through the synergistic operation of the above-mentioned technical features, the system achieves proactive, intelligent, and two-way remote security monitoring, effectively solving the problems of passive alarms, lack of intelligent risk assessment, and lack of remote intervention capabilities in existing technologies.

[0027] In one alternative implementation, the multimodal physiological and device data includes real-time blood glucose levels, predicted blood glucose trends, remaining insulin infusion volume, current basal infusion rate, information on infused bolus doses, insulin pump blockage alarm signals, insulin pump power information, and CGM signal strength information.

[0028] Specifically, real-time blood glucose levels refer to the patient's current blood glucose concentration data directly obtained through a continuous glucose monitor. This can be achieved using electrochemical or optical sensor technology, aiming to provide immediate feedback on blood glucose status. Blood glucose trend prediction can be understood as the future direction and magnitude of blood glucose changes calculated based on historical blood glucose data and dynamic models. This can be achieved through machine learning algorithms or time series analysis methods, aiming to identify potential risks in advance. Remaining insulin infusion volume refers to the remaining infusionable insulin dose in the insulin pump. This can be achieved through a built-in volume sensor or counter, aiming to prevent treatment interruption due to insulin depletion. The current basal infusion rate refers to the basal insulin dose continuously infused by the insulin pump per unit time. This can be adjusted through the infusion control module, aiming to maintain stable blood glucose levels. Recorded over-infusion doses refer to the recently manually added insulin doses, which can be stored through a dose recording module, aiming to track the impact of human intervention on blood glucose. The insulin pump blockage alarm signal is a status signal triggered when an abnormality in the insulin infusion tubing is detected. This can be achieved through a pressure sensor or flow monitoring device, aiming to quickly respond to physical faults. Insulin pump power information refers to the device's current remaining power status, which can be monitored by the battery management unit to ensure the device's continuous operation. CGM signal strength information refers to the communication signal quality between the continuous glucose monitor and the receiving device, which can be achieved through wireless signal strength indication technology to verify the reliability of the monitoring data.

[0029] Specifically, the above solution constructs a comprehensive information system that simultaneously reflects the patient's physiological state and the equipment's operational status by comprehensively collecting multi-dimensional data. Real-time blood glucose levels serve as the core input, providing direct feedback on the patient's current blood glucose level; predicted blood glucose trends are dynamically modeled to predict future blood glucose trends, enabling the system to make forward-looking judgments. Based on this, the remaining insulin infusion volume and current basal infusion rate supplement key parameters of equipment operation from the perspectives of resource reserves and dose stability, respectively, avoiding risks caused by insufficient resources or improper settings. Information on infused high doses records human intervention behaviors, helping the system more accurately analyze the causes of postprandial blood glucose fluctuations. Insulin pump blockage alarm signals and insulin pump battery information provide safety assurance from the perspectives of equipment failure and battery life, ensuring the equipment is always available. CGM signal strength information further enhances the reliability of data acquisition, avoiding misjudgments due to signal problems. The synergistic effect of these data items allows the risk assessment engine to comprehensively analyze the patient's physiological indicators and equipment status. For example, when a rapid decline in blood glucose occurs simultaneously with an insulin pump blockage signal, the system can identify a compound risk rather than an isolated event, thereby significantly improving the accuracy of risk prediction and reducing unnecessary alarm interference. Furthermore, this solution is closely integrated with components such as the data acquisition module and risk assessment engine to form an intelligent risk assessment system, effectively solving the technical problem that a single data dimension is insufficient to comprehensively capture the patient's condition.

[0030] In one alternative implementation, the risk assessment engine is configured to perform the following operations: Based on real-time blood glucose values ​​and predicted blood glucose trends, calculate the hypoglycemia risk index and hyperglycemia risk index for a specific future time period. Based on the remaining insulin infusion volume, current basal infusion rate, information on the large doses already infused, insulin pump blockage alarm signal, insulin pump power information, and CGM signal strength information, the equipment failure risk index is calculated. The hypoglycemia risk index, hyperglycemia risk index, and equipment failure risk index are weighted and fused using an algorithm to generate a comprehensive risk assessment result; the comprehensive risk assessment result represents different risk levels.

[0031] Specifically, the hypoglycemia risk index quantifies the probability of a hypoglycemic event occurring within a specific future timeframe based on real-time blood glucose levels and their predicted trends, combined with historical patient data and clinical experience models. It can be implemented by training multidimensional features using machine learning algorithms (such as support vector machines or random forests) to identify potential hypoglycemia risks and provide early warnings. The hyperglycemia risk index is similar to the hypoglycemia risk index, but it focuses on the risks associated with excessively high blood glucose levels. Specifically, it predicts blood glucose trends through time series analysis and dynamically adjusts these trends in conjunction with the patient's personalized treatment goals. The device failure risk index assesses the probability of potential malfunctions or abnormalities in insulin pumps through multidimensional monitoring of their operational status. This can be achieved using a rule engine combined with threshold judgments, such as triggering a warning when the remaining insulin infusion volume falls below a preset safety value. The weighted fusion algorithm is a method that comprehensively calculates multiple risk indices according to preset weights. It can be implemented using linear weighting or nonlinear mapping functions, aiming to assign differentiated weights based on the clinical importance of different risk types, thereby generating more accurate comprehensive risk assessment results.

[0032] In detail, the above technical solution constructs a complete intelligent risk assessment system through multi-source data fusion and prediction mechanisms. First, based on real-time blood glucose levels and predicted blood glucose trends, the system can comprehensively capture blood glucose fluctuation risks from both the immediate state and future evolution paths, avoiding the lag problem of traditional single-threshold judgments. Second, by integrating equipment operating parameters such as remaining insulin infusion volume, current basal infusion rate, and information on infused high doses, the system can effectively identify potential equipment-level hazards, such as catheter blockage or insufficient power, thus overcoming the one-sidedness of existing technologies that rely solely on physiological data for risk assessment. Furthermore, the introduction of a weighted fusion algorithm allows different risk types to be uniformly represented in a quantitative form, considering both the priority of hypoglycemia risk and the impact of hyperglycemia and equipment failure. The resulting comprehensive risk assessment accurately reflects the overall risk level. In addition, this solution works closely with the system's data acquisition module and dynamic early warning module. The former provides rich multimodal data sources, while the latter implements tiered and precise alerts based on the comprehensive risk assessment results, significantly improving the system's early warning capabilities and intervention efficiency.

[0033] In one optional implementation, the dynamic early warning module is configured as follows: Multiple warning levels corresponding to different risk levels are pre-stored; When the comprehensive risk assessment result reaches the first risk level but does not reach the second risk level, a reminder-level alert is triggered. This alert is either silent or a mild prompt. When the comprehensive risk assessment result reaches the second risk level but does not reach the third risk level, a warning level alert is triggered, which includes an audio-visual prompt. When the comprehensive risk assessment results reach the third risk level, a critical level warning is triggered, which includes continuous strong reminders and recommended measures.

[0034] Specifically, the dynamic early warning module is a functional unit that can dynamically adjust the warning level based on the comprehensive risk assessment results. It can be implemented using a hierarchical storage structure, pre-setting multiple warning levels and their corresponding risk thresholds to ensure that the appropriate warning method can be quickly matched upon receiving the risk assessment results. The purpose of introducing this module is to solve the inefficiency caused by the lack of hierarchical alarms, thereby improving the system's response speed and intervention efficiency.

[0035] In practical applications, pre-stored multiple alert levels corresponding to different risk levels refer to the system maintaining a complete risk-alert mapping table, which can be implemented through databases, configuration files, or embedded algorithms. This design, based on accurate judgment of risk assessment results, avoids delays in real-time decision-making and ensures timely warnings. When the comprehensive risk assessment result reaches the first risk level but not the second, a reminder-level alert is triggered, using silent or mild prompts. This design effectively prevents alert fatigue. When the comprehensive risk assessment result reaches the second risk level but not the third, a warning-level alert is triggered, including audible and visual prompts. This design attracts attention, prompting users or guardians to promptly view and handle the anomaly. When the comprehensive risk assessment result reaches the third risk level, a critical-level alert is triggered, including continuous strong reminders and suggested measures. This design not only ensures immediate response but also provides specific action suggestions, significantly shortening the time from anomaly detection to intervention.

[0036] In detail, the dynamic early warning module achieves precise response to risk assessment results through a tiered early warning mechanism. In low-risk situations, the system issues only silent or mild alerts to avoid excessive interference with users and caregivers; in medium-risk situations, it attracts attention with audio-visual alerts to prevent the problem from escalating; in high-risk situations, continuous strong reminders combined with specific suggested measures ensure rapid response in emergencies. Furthermore, the dynamic early warning module works in conjunction with the risk assessment engine, performing tiered matching based on the comprehensive risk assessment results generated by the latter, thus forming a complete closed-loop monitoring system. This design not only solves the inefficiency problem caused by non-tiered alerts but also improves the system's practicality and reliability through precise tiering, making it particularly suitable for ensuring the safety of diabetic patients using insulin pump therapy.

[0037] In one alternative implementation, the intervention instructions provided by the remote intervention interface include at least one of the following: sending a preset comfort or reminder message to the user device, triggering the buzzer of the user device to locate the device, and initiating a real-time voice call connection with the user device.

[0038] Specifically, the remote intervention interface refers to the application interface integrated on the caregiver's mobile phone, which can provide intervention instructions through graphical buttons, shortcut menus, and other methods. The preset comfort or reminder messages are standardized text content pre-configured in the system, designed to avoid delays and errors caused by manual input in emergencies, ensuring that critical guidance information is delivered promptly. Triggering the buzzer on the user device refers to activating the acoustic signal function of the user device via remote command, aiming to quickly locate the physical device and address the problem of patients being unable to retrieve the device independently due to environmental interference or mobility limitations. Initiating a real-time voice call connection with the user device refers to establishing a direct voice channel using an end-to-end communication protocol, aiming to significantly shorten the time interval from alarm recognition to intervention implementation.

[0039] In detail, when a caregiver receives an alert through the remote intervention interface, they can choose appropriate intervention measures based on the specific situation. For example, in cases of abnormal but non-critical blood sugar levels, the caregiver can choose to send a preset comforting or reminder message, and the patient's phone will immediately display a prompt such as "Mom: Quickly check your blood sugar and drink juice!" When the alert indicates an abnormal device status, the caregiver can click the "Sound to Locate Pump" button, and the patient's phone and insulin pump will sound at maximum volume to help quickly locate the device. In critical situations, the caregiver can directly establish an internet voice call through the "One-Click Call" function. This integrated design eliminates the traditional dialing steps, allowing the caregiver to immediately provide voice guidance based on the risk level of the alert. These intervention methods work closely with other modules of the system to form a complete closed-loop management mechanism, effectively solving the response delay problem and improving intervention efficiency.

[0040] Through the above technical solutions, guardians can quickly select targeted actions based on the real-time content of the warning information, realizing an efficient transition from alarm recognition to intervention implementation, thereby providing timely and effective remote assistance in emergency scenarios.

[0041] In one optional embodiment, the system of the present invention further includes: The data anonymization module is located between the data acquisition module and the remote communication module. It is used to de-identify the data before sending it to the remote monitoring terminal to remove the user's personal identity information.

[0042] Specifically, a data anonymization module is a functional unit specifically designed to process sensitive information. It can be implemented using methods such as data masking, data replacement, or data encryption. In practical applications, the data anonymization module is deployed in the transmission link between the data acquisition module and the remote communication module. The purpose is to ensure that the raw data stream is intercepted and processed before entering the remote communication stage, thereby preventing sensitive information from being exposed in the transmission channel.

[0043] In detail, the data anonymization module proactively removes users' personally identifiable information by performing de-identification processing. For example, identifiable identifiers such as names and ID numbers are replaced with randomly generated anonymous IDs, while only essential medical data such as blood glucose levels and device status are retained. The key to this design is to prioritize privacy protection during data transmission, rather than remedial measures afterward, thereby preventing the possibility of identity information leakage at the source. Furthermore, the module's design fully considers regulatory requirements, meeting the anonymization requirements of medical data privacy regulations such as HIPAA and GDPR, while ensuring the normal operation of remote monitoring functions. Even if data is intercepted during transmission or cloud storage, it cannot be traced back to a specific individual, achieving a seamless integration of privacy and functional requirements.

[0044] Building upon this foundation, the data anonymization module works closely with other modules. The data acquisition module is responsible for obtaining multimodal physiological and device data from user devices. This data undergoes processing by the data anonymization module before being transmitted to the remote communication module. This process not only ensures that the data received by the remote monitoring terminal contains only anonymized medical indicators but also enhances the overall system's privacy protection mechanism. In this way, the system addresses the issue of potentially carrying personally identifiable information during the transmission of user physiological and device data while effectively reducing privacy and security risks caused by data leaks, thereby significantly improving the overall security and compliance of the system.

[0045] In addition, such as Figure 2 As shown in the embodiments of this application, a remote alert monitoring method for an insulin pump is also proposed, including the following steps: S1. Acquire multimodal physiological and device data in real time from user terminal devices; S2. Based on the fusion analysis of multimodal physiological and equipment data, a comprehensive risk assessment result is generated; S3. Based on the comprehensive risk assessment results, dynamically trigger warning messages of different risk levels; S4. Simultaneously push the warning information to at least one remote monitoring terminal and one user terminal device; S5. Receive and display warning information through the remote monitoring terminal, and in response to the operation of the remote monitor, send intervention instructions to the user terminal device.

[0046] The core innovation of this embodiment lies in combining the data acquisition module with the risk assessment engine through multimodal data analysis, and introducing a dynamic early warning module and a remote intervention interface. This solves the problems in insulin pump monitoring, such as alarms being limited to the patient's device, lack of risk assessment capabilities involving multi-dimensional data fusion, and the inability of remote monitors to directly intervene. It achieves the effect of proactive, intelligent, and two-way closed-loop monitoring.

[0047] Specifically, step S1 acquires multimodal physiological and equipment data in real time, providing the system with input sources covering multiple dimensions such as blood glucose levels and equipment status, avoiding assessment blind spots caused by reliance on single data and ensuring the comprehensiveness of risk analysis; step S2 performs fusion analysis based on multimodal data to generate comprehensive risk assessment results, realizing the quantitative calculation of hypoglycemia, hyperglycemia, and equipment failure risks by integrating physiological trends and equipment operating status, overcoming the false alarm defects of traditional single threshold triggering; step S3 dynamically triggers warning information of different risk levels based on the comprehensive risk assessment results, automatically adjusting the warning intensity according to the degree of risk, ensuring timely response to critical situations while avoiding excessive interference from non-critical alarms; step S4 pushes warning information to both the remote monitoring terminal and the user terminal device simultaneously, ensuring that information is delivered synchronously to both the patient and the caregiver, breaking the limitation of alarms being limited to the patient end and creating conditions for remote intervention; step S5 receives warnings and sends intervention instructions through the remote monitoring terminal, enabling caregivers to directly initiate operations such as message reminders or voice calls, establishing a rapid channel from risk discovery to proactive intervention. These features work together closely: S1 provides the data foundation for S2, the evaluation results of S2 drive the accurate warnings of S3, S4 ensures two-way coverage of warning information, and S5 realizes real-time intervention based on the push of S4. The whole process forms a closed-loop mechanism of data-driven, dynamic response and two-way interaction, which together solves the technical problems of passive alarms, unintelligent risk assessment and lack of intervention capabilities.

[0048] Through the above technical solutions, this application constructs a complete process from data collection to remote intervention, realizing proactive and intelligent monitoring and closed-loop management of insulin pump usage, effectively solving the core problems of passive alarms, one-sided risk assessment and lack of intervention in existing technologies.

[0049] In one optional implementation, step S2 specifically includes the following steps: Calculate the hypoglycemia risk index and hyperglycemia risk index based on blood glucose data; Calculate the equipment risk index based on user-end device operation data; The hypoglycemia risk index, hyperglycemia risk index, and equipment failure risk index are weighted and fused using an algorithm to generate a comprehensive risk assessment result; the comprehensive risk assessment result represents different risk levels.

[0050] In one optional implementation, step S3 specifically includes the following steps: The comprehensive risk assessment results are compared with multiple preset risk thresholds; each preset risk threshold corresponds to a different risk level. Based on the comparison results, select a corresponding level from multiple predefined risk levels to trigger an alert, which includes at least one non-critical warning level and one critical level.

[0051] Specifically, the comprehensive risk assessment result refers to a quantitative indicator generated by integrating multi-dimensional physiological and equipment data, which can be implemented using numerical ranges, percentages, or scoring systems. Multiple preset risk thresholds are critical points set according to different risk scenarios. These can be fixed values ​​derived from historical data statistics or dynamically adjusted adaptive thresholds, aiming to ensure the system can accurately distinguish risk scenarios of varying severity. The design of non-critical and critical warning levels is to cover the completeness of basic risk scenarios, which can be achieved through a tiered response mechanism, aiming to balance warning sensitivity and user experience.

[0052] In detail, this solution achieves precision and intelligence in its dynamic early warning mechanism by constructing a multi-level risk threshold comparison framework. First, it compares the comprehensive risk assessment results with multiple preset risk thresholds. This process utilizes risk assessment results that integrate multi-dimensional physiological and device data, comprehensively reflecting the complexity and dynamic changes of risk. This avoids false alarms caused by relying on a single indicator, thus ensuring the objectivity and reliability of risk identification. Multiple preset risk thresholds correspond to different risk levels, establishing a continuous risk spectrum from low to high, enabling the system to distinguish subtle risk differences. For example, when device blockage is analyzed in conjunction with blood glucose trends, a moderate risk can be identified rather than a direct assessment of a critical situation, providing a structured basis for tiered responses. The corresponding risk level is selected to trigger an alert based on the comparison results. This ensures that the alert intensity strictly matches the actual risk level, triggering only a mild alert to reduce user interference when the risk is within a controllable range, and automatically escalating the alert level when the risk rises sharply, achieving dynamic adaptive adjustment of the early warning strategy. In addition, the design of non-critical warning and critical levels ensures the system's complete coverage of basic risk scenarios. At the warning level, moderate intervention such as audible and visual prompts is used to avoid excessively disturbing users. At the critical level, continuous strong reminders are activated along with suggested measures to ensure that emergencies are responded to in a timely manner. This effectively solves the problem of alarm overload or insufficient response caused by the lack of risk classification.

[0053] Through the above technical solutions, the dynamic early warning mechanism can operate with precision, significantly improving the reliability and practicality of the insulin pump remote warning monitoring system, and providing more efficient technical support for the safety monitoring of diabetic patients.

[0054] In one alternative implementation, in step S5, the intervention instruction includes initiating a real-time voice call with the user terminal device; The method of the present invention further includes: temporarily pausing or reducing the intensity of the audio-visual warnings pushed to the remote monitoring terminal after the call is established.

[0055] Specifically, real-time voice calls refer to a two-way voice communication channel established through network communication protocols. This can be implemented using VoIP technology, the WebRTC framework, or traditional telephone networks. The purpose is to provide caregivers and patients with an immediate means of communication to quickly deliver crucial guidance information. Pausing or reducing the intensity of audio-visual warnings can be achieved by adjusting the alarm volume, turning off the flashlight, or switching the notification method from a strong alert to a weak alert. The aim is to reduce interference during the call and ensure clear voice communication between both parties.

[0056] In detail, during the operation of the aforementioned system, when a guardian initiates a real-time voice call with the user's device via the remote intervention interface, the system automatically detects the call status and dynamically adjusts the audible and visual alert behavior on the remote monitoring terminal based on the call establishment trigger. For example, after a successful call connection, the system will switch the originally high-frequency flashing alert light to a low-frequency flashing, or reduce the sharp alarm sound to a soft background prompt, or even completely pause the output of audible and visual alerts. This dynamic adjustment mechanism not only maintains the continuity of risk monitoring but also optimizes the human-computer interaction experience, allowing guardians to focus on the call content without interference, thereby significantly improving the efficiency and accuracy of emergency intervention. Furthermore, since the audible and visual alerts can automatically resume after the call ends, the overall solution ensures timely response capabilities in critical situations while avoiding the obstruction of voice communication by continuous high-intensity alarms, demonstrating the ingenuity and practicality of the design.

[0057] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A remote alert monitoring system for insulin pumps, characterized in that, include: The data acquisition module is used to acquire multimodal physiological and device data from user-end devices in real time; The user-end device includes at least an insulin pump and a continuous glucose monitor; The risk assessment engine, which is connected in communication with the data acquisition module, is used to perform fusion analysis based on the multimodal physiological and device data to generate a comprehensive risk assessment result. The dynamic early warning module is connected in communication with the risk assessment engine and is used to dynamically trigger warning information of different risk levels based on the comprehensive risk assessment results. A remote communication module, which is communicatively connected to the dynamic early warning module, is used to simultaneously push the warning information to at least one remote monitoring terminal and the user terminal device; A remote intervention interface, integrated into the remote monitoring terminal, is used to receive and display the warning information and to allow the remote monitor to initiate at least one intervention command; the intervention command is fed back to the user terminal device through the remote communication module.

2. The system according to claim 1, characterized in that, The multimodal physiological and equipment data includes real-time blood glucose levels, predicted blood glucose trends, remaining insulin infusion volume, current basal infusion rate, information on infused high doses, insulin pump blockage alarm signals, insulin pump power information, and CGM signal strength information.

3. The system according to claim 1, characterized in that, The risk assessment engine is configured to perform the following operations: Based on real-time blood glucose values ​​and predicted blood glucose trends, calculate the hypoglycemia risk index and hyperglycemia risk index for a specific future time period. Based on the remaining insulin infusion volume, current basal infusion rate, information on the large doses already infused, insulin pump blockage alarm signal, insulin pump power information, and CGM signal strength information, the equipment failure risk index is calculated. The hypoglycemia risk index, hyperglycemia risk index, and equipment failure risk index are weighted and fused using a fusion algorithm to generate the comprehensive risk assessment result. The comprehensive risk assessment results represent different risk levels.

4. The system according to claim 3, characterized in that, The dynamic early warning module is configured as follows: Multiple warning levels corresponding to different risk levels are pre-stored; When the comprehensive risk assessment result reaches the first risk level but does not reach the second risk level, a reminder-level alert is triggered, which is either silent or a mild prompt. When the comprehensive risk assessment result reaches the second risk level but does not reach the third risk level, a warning level alert is triggered, which includes an audio-visual prompt. When the comprehensive risk assessment result reaches the third risk level, a critical level warning is triggered, which includes continuous strong reminders and recommended measures.

5. The system according to claim 1, characterized in that, The intervention commands provided by the remote intervention interface include at least one of the following: sending a preset comfort or reminder message to the user terminal device, triggering the buzzer of the user terminal device to sound in order to locate the device, and initiating a real-time voice call connection with the user terminal device.

6. The system according to any one of claims 1, characterized in that, The system also includes: The data anonymization module is located between the data acquisition module and the remote communication module. It is used to de-identify the data before sending it to the remote monitoring terminal to remove the user's personal identity information.

7. A remote alert monitoring method for an insulin pump based on the system described in any one of claims 1-6, characterized in that, Includes the following steps: S1. Acquire multimodal physiological and device data in real time from user terminal devices; S2. Based on the multimodal physiological and device data, perform fusion analysis to generate a comprehensive risk assessment result; S3. Based on the comprehensive risk assessment results, dynamically trigger warning messages of different risk levels; S4. The warning information is simultaneously pushed to at least one remote monitoring terminal and the user terminal device; S5. Receive and display the warning information through the remote monitoring terminal, and in response to the operation of the remote monitor, send the intervention command to the user terminal device.

8. The method according to claim 7, characterized in that, S2 specifically includes the following steps: Calculate the hypoglycemia risk index and hyperglycemia risk index based on blood glucose data; Calculate the equipment risk index based on user-end device operation data; The hypoglycemia risk index, hyperglycemia risk index, and equipment failure risk index are weighted and fused using a fusion algorithm to generate the comprehensive risk assessment result; the comprehensive risk assessment result represents different risk levels.

9. The method according to claim 7, characterized in that, S3 specifically includes the following steps: The comprehensive risk assessment result is compared with multiple preset risk thresholds; each preset risk threshold corresponds to a different risk level. Based on the comparison results, select a corresponding level from multiple predefined risk levels to trigger an alert, which includes at least one non-critical warning level and one critical level.

10. The method according to claim 7, characterized in that, In S5, the intervention instruction includes initiating a real-time voice call with the user terminal device; The method also includes: temporarily pausing or reducing the intensity of the audio-visual alerts pushed to the remote monitoring terminal after the call is established.

Images