Self-adaptive network switching insulin pump remote monitoring system

Abstract

The invention relates to the technical field of medical instruments, and discloses a self-adaptive network switching insulin pump remote monitoring system, which comprises an insulin pump equipment end, an insulin pump equipment end, an insulin pump equipment end, an insulin pump equipment end, an insulin pump equipment end, an insulin pump equipment end, an insulin pump equipment end, an insulin pump equipment end, an insulin pump equipment end, an insulin pump equipment end, an insulin pump equipment end, an insulin pump equipment end and an insulin pump equipment end, the relay equipment communicates with the insulin pump equipment end through the multimode communication module, is used for operating network switching management software and is connected with a cloud server; and the cloud server is used for receiving and storing the patient data transmitted by the insulin pump equipment end and providing a remote monitoring interface. By integrating the intelligent learning algorithm module, the network switching opportunity is predicted according to the signal strength change trend, the historical switching success rate and the user movement mode, the network performance reduction trend can be recognized in advance, actions can be taken, and unnecessary frequent switching is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of medical devices, and particularly to an insulin pump remote monitoring system with adaptive network switching. Background Art

[0002] An insulin pump is a medical device used for diabetes management, which helps patients control their blood glucose levels by continuously or on-demand delivering insulin. With the development of remote monitoring technology, existing insulin pump systems are usually equipped with wireless communication functions to transmit patient data (such as blood glucose values and insulin doses) to smartphones or cloud servers for real-time monitoring. Most existing systems adopt a single network communication method, such as connecting to a smartphone via Bluetooth or communicating with local area network devices through Wi-Fi. Some systems also integrate continuous glucose monitoring (CGM) functions. For example, Medtronic's MiniMed system and Tandem's t:slim X2 pump transmit data to a relay device via Bluetooth, and then the relay device uploads the data to the cloud. These systems focus on real-time data transmission and device portability in design to meet the daily use needs of patients.

[0003] However, the existing insulin pump remote monitoring systems have significant limitations in network switching. They mainly rely on fixed thresholds or manual operations to cope with network changes and cannot achieve intelligent prediction and optimization. For example, when a patient moves from an indoor environment covered by Wi-Fi to the outdoor, the system usually triggers a switch to another network (such as 4G) after the signal strength drops below a certain preset value (such as -80 dBm). This passive switching method is difficult to identify the signal attenuation trend in advance, resulting in a lag in the switching timing and prone to data transmission interruption or delay. In addition, existing systems lack the ability to analyze historical signal data and user movement patterns, and cannot optimize network selection according to the dynamic environment. Especially in scenarios with rapidly changing network conditions (such as urban mobility or signal interference areas), the continuity and stability of data transmission are affected. Summary of the Invention

[0004] In view of the deficiencies of the prior art, the present invention provides an insulin pump remote monitoring system with adaptive network switching, which solves the problem that the existing insulin pump remote monitoring system has significant limitations in network switching, mainly relying on fixed thresholds or manual operations to cope with network changes and unable to achieve intelligent prediction and optimization.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: An insulin pump remote monitoring system with adaptive network switching, comprising:

[0006] The insulin pump device side is configured with an embedded microcontroller, a multi-mode communication module, and a signal strength monitoring sensor, where the multi-mode communication module supports Bluetooth, 4G, and Wi-Fi communications;

[0007] A relay device communicates with the insulin pump device side through the multi-mode communication module, and is used to run network switching management software and connect to a cloud server;

[0008] A cloud server is used to receive and store patient data transmitted by the insulin pump device side, and provide a remote monitoring interface.

[0009] Preferably, the insulin pump device side further includes a data cache unit, which is used to temporarily store patient data during network switching, and its storage capacity supports at least 5 seconds of data volume.

[0010] Preferably, the network switching management software is configured with an adaptive network switching mechanism, and the adaptive network switching mechanism triggers network switching based on at least one of the following conditions:

[0011] The signal strength detected by the signal strength monitoring sensor is lower than a preset threshold;

[0012] The data transmission delay exceeds a preset time;

[0013] The current network connection is interrupted.

[0014] Preferably, the adaptive network switching mechanism includes the following steps: detecting the current network performance, selecting a target network, and synchronizing data.

[0015] Preferably, it further includes an intelligent learning algorithm module, which predicts the network switching timing according to the signal strength change trend. The input of the intelligent learning algorithm module includes historical signal strength data, network switching success rate, and user movement patterns, and the output is a network switching decision.

[0016] Preferably, the intelligent learning algorithm module uses a long short-term memory network LSTM or a decision tree model to predict the signal strength change trend.

[0017] Preferably, the multi-mode communication module is configured with a default network priority, where the priority of Wi-Fi is higher than that of Bluetooth, and the priority of Bluetooth is higher than that of 4G, and the multi-mode communication module dynamically adjusts network selection according to real-time monitoring results.

[0018] Preferably, the multi-mode communication module includes:

[0019] A Bluetooth 5.0 chip for short-range low-power communication;

[0020] A 4G module for wide area network communication;

[0021] A Wi-Fi module for high-speed local area network communication.

[0022] Preferably, the system is configured to reduce data transmission latency and patient data loss during network switching.

[0023] Preferably, the system is configured to operate in the following scenarios:

[0024] Give priority to using Wi-Fi in a home environment and use Bluetooth as an auxiliary;

[0025] Give priority to using 4G in an outdoor mobile environment;

[0026] Give priority to using Wi-Fi for high-bandwidth data transmission in a hospital environment.

[0027] The present invention provides an insulin pump remote monitoring system with adaptive network switching. It has the following beneficial effects:

[0028] 1. By integrating an intelligent learning algorithm module, the present invention predicts the network switching timing based on the signal strength change trend, historical switching success rate, and user movement pattern. Compared with the traditional switching method that relies on fixed thresholds, it can identify the trend of network performance degradation in advance and take actions; this switching mechanism based on trend prediction improves the accuracy and stability of network selection, reduces unnecessary frequent switching, and thus significantly enhances the continuity of data transmission in a dynamic environment.

[0029] 2. The system of the present invention adopts a hybrid networking design of Bluetooth, 4G, and Wi-Fi, and combines an adaptive network switching mechanism. It can dynamically select the optimal network according to the real-time signal strength. Compared with the traditional insulin pump system with single-network communication, it avoids connection interruptions caused by insufficient network coverage or interference. With the cooperation of the data cache unit, patient data can be temporarily stored during the switching process and synchronized when the connection is restored, further ensuring the integrity of key data such as blood glucose values and insulin doses.

[0030] 3. The present invention optimizes the network switching process through pre-connection technology and cache mechanism, controlling the switching time within 50 milliseconds. Compared with the possible hundreds of milliseconds of delay in traditional insulin pump systems, it has significant improvements. Before the switching is triggered, the system establishes a standby link with the target network in advance and synchronizes the cached data immediately after the switching, avoiding monitoring interruptions caused by delays.

[0031] 4. The system of the present invention is configured with a default network priority (Wi-Fi > Bluetooth > 4G) and supports dynamic adjustment. It can automatically select an appropriate communication method according to different scenarios (such as at home, outdoors, in the hospital). Compared with the traditional insulin pump system with a fixed network, it has a wider range of applications. This multi-scenario adaptation ability improves the flexibility of the device in actual use and meets the monitoring needs of patients in different environments.

[0032] 5. By setting a data cache unit at the insulin pump device end, the system of the present invention can temporarily store patient data for at least 5 seconds during network switching and synchronize it to the cloud server through a sequence number verification and retransmission mechanism after the switching is completed. Compared with the traditional system without a cache design, the data loss rate is reduced to less than 1%. This data protection function enhances the reliability and medical safety of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a system architecture diagram of an insulin pump remote monitoring system with adaptive network switching according to the present invention;

[0034] Figure 2 It is a working principle diagram of the multi-mode communication module of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0035] Next, the technical solutions of the present invention will be clearly and completely described in conjunction with the drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0036] Please refer to the attached Figure 1 - attached Figure 2 , the embodiment of the present invention provides an insulin pump remote monitoring system with adaptive network switching, including:

[0037] The insulin pump device end is configured with an embedded microcontroller, a multi-mode communication module, and a signal strength monitoring sensor, wherein the multi-mode communication module supports Bluetooth, 4G, and Wi-Fi communications;

[0038] The relay device communicates with the insulin pump device end through the multi-mode communication module and is used to run network switching management software and connect to the cloud server;

[0039] The cloud server is used to receive and store the patient data transmitted by the insulin pump device end and provide a remote monitoring interface.

[0040] Specifically, the embedded microcontroller built into the insulin pump device end is designed with low power consumption. For example, based on the ARM Cortex-M4 architecture, it has at least 32-bit processing capabilities and supports real-time data processing and communication control. The multi-mode communication module is integrated on the circuit board of the device end and is connected to the microcontroller through a hardware interface (such as SPI or I2C) to ensure the parallel operation of the three communication modes. The signal strength monitoring sensor specifically includes a radio frequency signal detection circuit that can collect the received signal strength indication (RSSI) of Bluetooth, 4G, and Wi-Fi in real time at a cycle of 1 second. Its measurement range is from -100 dBm to 0 dBm, and the accuracy reaches ±2 dBm.

[0041] The relay device can be a smartphone, a tablet computer, or a dedicated gateway device. It has a processor running the Android or iOS system built in, is equipped with at least 4GB of memory and a CPU that supports multi-threading to run the network switching management software. This software establishes two-way communication with the insulin pump device end through an application programming interface (API), supports at least 10 data interactions per second, and the transmitted content includes insulin dosage, blood glucose value, and device status.

[0042] The cloud server is deployed in a distributed architecture, such as a virtual private cloud (VPC) based on AWS or Alibaba Cloud. It has high availability and data redundancy functions, with a storage capacity of not less than 1TB and can save the data records of at least 1000 patients for one year. The server communicates with the relay device through the HTTPS protocol, uses 256-bit AES encryption to protect patient privacy, and provides a web interface and a mobile application as remote monitoring interfaces for doctors or family members to view the patient status in real time.

[0043] The specific implementation method of the hybrid networking is as follows: Bluetooth is used for short-distance (less than 10 meters) low-power connection, 4G is used for long-distance (coverage up to several kilometers) mobile communication, and Wi-Fi is used for high-speed (bandwidth up to 100 Mbps) local area network transmission. The automatic switching mechanism runs through a preset switching algorithm. For example, when the Bluetooth signal strength is lower than -80 dBm, it preferentially tries to connect to Wi-Fi. If Wi-Fi is unavailable, it switches to 4G to ensure that the data stream is not interrupted during the switching process.

[0044] The insulin pump device end also includes a data cache unit, which is used to temporarily store patient data during the network switching process, and its storage capacity supports at least 5 seconds of data volume.

[0045] Specifically, the data cache unit uses non-volatile memory (such as Flash or EEPROM), with a capacity designed to be 256KB, capable of storing at least 5 seconds of patient data, calculated based on the rate of approximately 50 bytes of data generated per second by the insulin pump (including timestamp, dose, and blood glucose value). The cache unit is connected to the microcontroller through DMA (Direct Memory Access) technology to ensure the efficiency of data writing and reading, with a writing speed not less than 1MB / s.

[0046] During the network switching process, for example, when switching from Wi-Fi to 4G, the cache unit activates the temporary storage mode, records the packet sequence numbers before and after the switch, and verifies the data integrity through a differential verification algorithm (such as CRC32) after the switch is completed. If data loss is detected, the retransmission mechanism is triggered, and the missing part is extracted from the cache and uploaded to the cloud server.

[0047] This design is particularly suitable for scenarios with unstable networks. For example, when the patient moves from indoors (Wi-Fi environment) to outdoors (4G environment), the cache unit can avoid missing blood glucose data due to signal interruption, ensuring that doctors can obtain continuous monitoring records.

[0048] The network switching management software is configured with an adaptive network switching mechanism, which triggers network switching based on at least one of the following conditions:

[0049] The signal strength detected by the signal strength monitoring sensor is lower than the preset threshold;

[0050] The data transmission delay exceeds the preset time;

[0051] The current network connection is interrupted.

[0052] Specifically, the adaptive network switching mechanism is built into the network switching management software and runs through an independent thread with a thread priority higher than the data transmission thread to ensure the real-time nature of the switching decision. The specific values for the signal strength lower than the preset threshold are: Bluetooth < -80dBm, Wi-Fi < -70dBm, 4G < -90dBm, and these thresholds can be adjusted by the user or technician through the software interface according to the actual environment.

[0053] The preset time for data transmission delay is 100 milliseconds, determined based on the real-time requirements of medical devices. If the delay exceeds this value, for example, due to congestion of the Wi-Fi access point, the switching evaluation is triggered. The detection of the interruption of the current network connection is achieved through heartbeat signals. A heartbeat packet is sent every 2 seconds, and if no response is received continuously for 3 times, it is determined as an interruption.

[0054] The above conditions can be triggered individually or in combination. For example, when the signal strength is lower than the threshold and the delay exceeds 100 ms, the system preferentially switches to the sub-optimal network. This multi-condition trigger design improves the flexibility and reliability of the switch, adapting to the complex changes in different network environments.

[0055] The adaptive network switching mechanism includes the following steps: detecting the current network performance, selecting the target network, and synchronizing data.

[0056] Specifically, the specific method for detecting the current network performance includes calculating the sliding average value of the RSSI value in real time (window size is 5 seconds) and the packet loss rate (tested based on the UDP protocol). If the average RSSI drops by 10 dBm or the packet loss rate exceeds 5%, the switch evaluation is triggered. The target network selection is based on a weighted scoring algorithm, and the scoring factors include signal strength (weight 50%), bandwidth (weight 30%), and historical connection stability (weight 20%). The network with the highest score is selected as the target network.

[0057] The data synchronization process uses a lightweight protocol (such as MQTT). Before the switch, the current packet sequence number and timestamp are written into the cache. After the switch is completed, it is confirmed whether data needs to be retransmitted by comparing the sequence numbers. The synchronization time is controlled within 50 milliseconds, and fast switching is achieved through pre-connection technology (establishing a standby link with the target network in advance).

[0058] This process is particularly suitable for dynamic environments. For example, when a patient moves between the hospital corridor (Wi-Fi coverage) and the parking lot (4G coverage), it ensures the continuous transmission of blood glucose data and insulin dosage instructions.

[0059] It also includes an intelligent learning algorithm module. The intelligent learning algorithm module predicts the network switching timing according to the signal strength change trend. The input of the intelligent learning algorithm module includes historical signal strength data, network switching success rate, and user movement patterns, and the output is the network switching decision.

[0060] Specifically, the intelligent learning algorithm module runs in the microcontroller of the insulin pump device end, occupying no more than 128 KB of memory, and the computational complexity is controlled at the O(n) level to adapt to the resource limitations of embedded devices. The historical signal strength data is stored in units of 1 minute, and the records of the most recent 24 hours are retained. The data format includes timestamp and RSSI value. The network switching success rate is calculated by counting the number of successful switches in the most recent 100 times, and the failure cases (such as unstable connection after switching) are recorded as the basis for optimization.

[0061] The user's movement pattern is collected through an acceleration sensor or a GPS module on the relay device. For example, it detects the walking speed of a patient (about 1 m / s) or the moving speed of a vehicle (about 10 m / s), and correlates it with the trend of signal strength change to identify typical scenarios such as "indoor stationary" and "outdoor walking". The output decision includes the specific switching time (such as "switch to 4G after 3 seconds") and the target network type.

[0062] This module maintains its prediction ability through regular updates (such as pushing new model parameters through the cloud monthly), which is suitable for long-term used patient devices and improves the intelligence level of switching.

[0063] The intelligent learning algorithm module adopts a long short-term memory network (LSTM) or a decision tree model to predict the trend of signal strength change.

[0064] Specifically, the long short-term memory network (LSTM) model is designed as a single-layer network, containing 16 hidden units, with an input dimension of 3 (signal strength, handover success rate, moving speed) and an output dimension of 1 (signal drop probability). The training data comes from the simulation environment and real patient usage records, with a total of no less than 100,000 samples. The decision tree model uses the CART algorithm, with a depth of no more than 5 layers, splitting nodes based on the information gain ratio, and is suitable for fast inference scenarios.

[0065] The specific goal of predicting the trend of signal strength change is to estimate the probability that the RSSI drops by more than 10 dBm within the next 10 seconds. The prediction result is output as a continuous value from 0 to 1. If the probability exceeds 0.7, an early handover is triggered. When the model runs on the device, the prediction is updated every 5 seconds, and the power consumption is controlled within 1 mW.

[0066] The selection of the two models is dynamically adjusted according to the device hardware performance. For example, LSTM is preferred on high-performance devices, and the decision tree is used on low-power devices to ensure the universality and efficiency of the algorithm.

[0067] The multi-mode communication module is configured with a default network priority, where the priority of Wi-Fi is higher than that of Bluetooth, and the priority of Bluetooth is higher than that of 4G. Moreover, the multi-mode communication module dynamically adjusts network selection according to real-time monitoring results.

[0068] Specifically, the default network priority is set based on power consumption and bandwidth requirements: Wi-Fi provides the highest bandwidth (up to 100 Mbps) and the lowest latency (<10 ms), which is suitable for hospital or home environments; Bluetooth has the lowest power consumption (standby <1 mA), which is suitable for short-range connections; 4G has the widest coverage (several kilometers), which is suitable for outdoor movement. The priority can be adjusted through a configuration file in the software. For example, the priority of Bluetooth is increased in power consumption-sensitive scenarios.

[0069] Real-time monitoring collects data once per second through a signal strength monitoring sensor and evaluates network performance by combining bandwidth tests (sending 100KB test packets per minute). If the Wi-Fi signal drops but is still available, the current connection is maintained; if the Bluetooth connection is interrupted, it immediately switches to 4G. This dynamic adjustment mechanism ensures the adaptability of the system in different usage scenarios.

[0070] The multi-mode communication module includes:

[0071] A Bluetooth 5.0 chip for short-range low-power communication;

[0072] A 4G module for wide-area network communication;

[0073] A Wi-Fi module for high-speed local area network communication.

[0074] Specifically, the Bluetooth 5.0 chip supports the BLE (Bluetooth Low Energy) mode, with a transmission distance of up to 10 meters, a data rate of up to 2Mbps, and a power consumption of less than 1mA in standby, suitable for near-field communication with relay devices. The 4G module supports the LTE Cat4 standard, with a downlink rate of up to 150Mbps, an uplink rate of up to 50Mbps, an internal SIM card slot, and supports global mainstream frequency bands (B1 / B3 / B5, etc.). The Wi-Fi module supports the 802.11n protocol, operates in the 2.4GHz frequency band, has a maximum bandwidth of 150Mbps, and a coverage range of approximately 50 meters.

[0075] The three modules are connected to the microcontroller through a multiplexer (MUX), sharing antenna resources, avoiding signal interference, and the switching time is controlled within 20 milliseconds. The integrated design of the module takes into account the size limitations of medical devices, and the total weight does not exceed 50 grams.

[0076] The system is configured to reduce data transmission latency and patient data loss during network switching.

[0077] Specifically, the reduction of data transmission latency is achieved through pre-connection technology and caching mechanisms. For example, when detecting a decline in the current network performance, a standby connection is established with the target network in advance, and data is transmitted immediately after switching. The control of patient data loss depends on the cache unit and retransmission mechanism. If it is detected that the packet sequence numbers are not continuous, the missing data is extracted from the cache and retransmitted.

[0078] The system simulated 1000 network switches in laboratory tests, with the average latency controlled within 40 milliseconds and the data loss rate less than 0.5%, meeting the requirements of medical devices for real-time performance and reliability. This design is particularly suitable for emergency situations, such as data transmission when a patient's blood sugar drops suddenly.

[0079] The system is configured to operate in the following scenarios:

[0080] Prefer to use Wi-Fi in the home environment and use Bluetooth as a supplement;

[0081] Prefer to use 4G in the outdoor mobile environment;

[0082] Prefer to use Wi-Fi for high-bandwidth data transmission in the hospital environment.

[0083] Specifically, the Wi-Fi priority mode in the home environment utilizes the stable connection of the home router, with Bluetooth as a backup to maintain communication at the edge of the Wi-Fi signal coverage (such as balconies or basements). The 4G priority mode in the outdoor mobile environment adapts to the scenarios of patients walking or driving. The system automatically detects the GPS speed and switches to 4G to ensure that data transmission is not interrupted due to the unavailability of Wi-Fi. The Wi-Fi high-bandwidth mode in the hospital environment supports simultaneous connection of multiple devices, such as data synchronization between insulin pumps and blood glucose meters and nurse station terminals, with a transmission rate of over 50 Mbps.

[0084] The network selection for each scenario is loaded through a preset configuration file and dynamically optimized according to the real-time signal strength during operation. For example, it switches to a backup Wi-Fi channel when the hospital Wi-Fi signal is congested. This multi-scenario adaptation design improves the practicality of the system and the convenience of use for patients.

[0085] Algorithm Description

[0086] Long Short-Term Memory Network (LSTM) Prediction Algorithm

[0087] This system uses the Long Short-Term Memory Network (LSTM) to predict the changing trend of signal strength for early judgment of network switching timing. The algorithm inputs include the historical signal strength sequence Switching success rate R s and the user's moving speed V t . The model is a single-layer LSTM with 16 hidden units, and outputs the probability of signal decline . If P d > 0.7, then an early switch is triggered. The model is trained with 100,000 samples in the cloud, the loss function is , and the parameters are updated through the Adam optimizer with a learning rate of 0.001, and it runs once every 5 seconds when deployed on the device side.

[0088] Decision Tree Switching Decision Algorithm

[0089] The system uses the CART decision tree model to quickly generate switching decisions. The inputs are the current signal strength RSSI current , delay time D t and the list of available networks N. The tree depth does not exceed 5 layers, and it splits based on the information gain ratio. An example rule is: if RSSI current < -80 dBm and Dt > 100 ms, select the network with max(RSSI Ni ). The model is built offline and stored as a lookup table, occupying < 10 KB of memory and running once per second.

[0090] Weighted Scoring Network Selection Algorithm

[0091] This algorithm evaluates available networks and selects the target network. The inputs are signal strength S i , bandwidth B i and historical stability H i , and the score is calculated where and are normalization values. The output is the network N with the highest score target , taking < 10 ms and applied to the network selection phase of the handover process.

[0092] Data Serial Number Verification and Re - transmission Algorithm

[0093] The system ensures data integrity through serial number verification. The sender generates sequence Seq sent , and the receiver returns Seq recv . After comparison, missing data packets are extracted from the cache. CRC32 is used for verification, and the time is synchronized within < 50 ms after re - transmission. The cache capacity is 256 KB, supporting 5 - second data storage.

[0094] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An insulin pump remote monitoring system with adaptive network switching, characterized in that Comprising: An insulin pump device end, configured with an embedded microcontroller, a multi-mode communication module, and a signal strength monitoring sensor, wherein the multi-mode communication module supports Bluetooth, 4G, and Wi-Fi communications; A relay device, communicating with the insulin pump device end through the multi-mode communication module, for running network switching management software and connecting to a cloud server; A cloud server, for receiving and storing patient data transmitted by the insulin pump device end, and providing a remote monitoring interface.

2. The insulin pump remote monitoring system with adaptive network switching according to claim 1, wherein The insulin pump device end further includes a data cache unit, which is used to temporarily store patient data during network switching, and its storage capacity supports at least 5 seconds of data volume.

3. The insulin pump remote monitoring system with adaptive network switching according to claim 1, characterized in that, The network switching management software is configured with an adaptive network switching mechanism, and the adaptive network switching mechanism triggers network switching based on at least one of the following conditions: The signal strength detected by the signal strength monitoring sensor is lower than a preset threshold; The data transmission delay exceeds a preset time; The current network connection is interrupted.

4. The insulin pump remote monitoring system with adaptive network switching according to claim 1, characterized in that The adaptive network switching mechanism includes the following steps: detecting the current network performance, selecting a target network, and synchronizing data.

5. The insulin pump remote monitoring system with adaptive network switching according to claim 1, characterized in that, It further includes an intelligent learning algorithm module, which predicts the network switching timing according to the signal strength change trend. The input of the intelligent learning algorithm module includes historical signal strength data, network switching success rate, and user movement pattern, and the output is a network switching decision.

6. The insulin pump remote monitoring system for adaptive network switching according to claim 5, wherein The intelligent learning algorithm module adopts a long short-term memory network LSTM or a decision tree model to predict the signal strength change trend.

7. The insulin pump remote monitoring system with adaptive network switching according to claim 1, characterized in that, The multi-mode communication module is configured with a default network priority, where the priority of Wi-Fi is higher than that of Bluetooth, and the priority of Bluetooth is higher than that of 4G, and the multi-mode communication module dynamically adjusts network selection according to real-time monitoring results.

8. The insulin pump remote monitoring system with adaptive network switching according to claim 1, characterized in that, The multi-mode communication module includes: A Bluetooth 5.0 chip for short-distance low-power communication; A 4G module for wide area network communication; A Wi-Fi module for high-speed local area network communication.

9. The insulin pump remote monitoring system with adaptive network switching according to claim 1, characterized in that, The system is configured to reduce data transmission delay and patient data loss during network switching.

10. The insulin pump remote monitoring system with adaptive network switching according to claim 1, characterized in that, The system is configured to operate in the following scenarios: In a home environment, Wi-Fi is preferentially used and Bluetooth is used as an auxiliary; In an outdoor mobile environment, 4G is preferentially used; In a hospital environment, Wi-Fi is preferentially used for high-bandwidth data transmission.

Images