Multi-modal sensing constraint monitoring system and monitoring method

Abstract

The invention discloses a multi-modal sensing constraint monitoring system and a monitoring method.The system comprises a restraint strap, a medical clamping device, a gyroscope, a flexible tension sensor, an edge calculation module and a wireless transmission module, by integrating an acceleration sensor, the gyroscope and the flexible tension sensor, physical displacement and biomechanical changes are synchronously captured, and the monitoring accuracy is improved. And comprehensive monitoring of the constraint state is realized. Through the multi-mode sensing and edge computing technology, accurate monitoring and timely alarming can be achieved, and the risk of self-release of the patient is remarkably reduced. Dynamic adjustment of constraint tightness is achieved, and constraint conditions are optimized in real time according to the actual state of a patient.

Description

Technical Field

[0001] The present invention relates to the technical field of medical care, and particularly to a multi-modal sensing restraint monitoring system and a monitoring method. Background Art

[0002] In the clinical care of psychiatry, some patients need to be subjected to protective restraint due to impulsive aggression, self-harm or suicide, or non-cooperation in treatment. Traditional restraint tools mainly include cloth restraint belts and magnetically controlled restraint belts, but these tools have the following significant defects:

[0003] 1. Insufficient safety:

[0004] Cloth restraint belts are prone to complications such as skin pressure ulcers and blood circulation disorders, and the self-liberation rate of patients is high, requiring frequent manual inspections; although the existing magnetically controlled restraint belts reduce the liberation rate, they rely on manual observation and cannot perceive the risk of limb movement or knot loosening in real time.

[0005] 2. Lack of intelligence:

[0006] Most of the existing restraint tools rely on manual operation and cannot achieve intelligent monitoring and early warning. Summary of the Invention

[0007] The present invention provides a multi-modal sensing restraint monitoring system and a monitoring method, aiming to solve the problems that the existing restraint tools cannot perceive the risk of patients' self-liberation in real time and the lack of intelligence.

[0008] The technical solution provided by the present invention is as follows:

[0009] A multi-modal sensing restraint monitoring system includes:

[0010] A restraint belt;

[0011] A medical clamping device equipped with an acceleration sensor for clamping the restraint belt;

[0012] A gyroscope integrated on the medical clamping device for capturing limb rotation characteristics;

[0013] A flexible tension sensor integrated on the medical clamping device and in contact with the restraint belt for real-time feedback of the restraint pressure;

[0014] An edge computing module capable of implanting a depthwise separable convolutional neural network model for intelligently distinguishing rope-untying actions from normal activities;

[0015] A wireless transmission module for realizing multi-terminal linkage alarm within a certain range.

[0016] Furthermore, the medical clamping device incorporates an anti-disassembly trigger mechanism and a temperature monitoring module.

[0017] Furthermore, the deep separable convolutional neural network model embedded in the edge computing module is the TinyML model, which performs real-time analysis on the collected multimodal data through the deep separable convolutional neural network.

[0018] Furthermore, the wireless transmission module adopts LoRaWAN low-power wireless transmission technology.

[0019] Furthermore, the acceleration sensor is a MWJS01 type acceleration sensor.

[0020] Furthermore, the medical clamping device is integrally molded using medical-grade TPU material.

[0021] Meanwhile, the present invention also provides a multimodal sensing constraint monitoring method, which employs the above-mentioned multimodal sensing constraint monitoring system and includes the following steps:

[0022] The restraint straps are clamped using a medical clamping device. The displacement of the knot is monitored by an accelerometer, the rotation characteristics of the limb are captured by a gyroscope, and the restraint pressure is fed back in real time by a flexible tension sensor to obtain sensor data.

[0023] The sensor data is processed to eliminate environmental noise using a dynamic baseline calibration algorithm to ensure the accuracy of the sensor data;

[0024] By combining wavelet transform to extract the frequency band features of rope-untying action from the sensor data, the features are input into the deep separable convolutional neural network model embedded in the edge computing module for behavior classification.

[0025] The deep separable convolutional neural network model is used to perform real-time analysis of the collected multimodal data to distinguish between the patient's normal activities and rope-untying behavior;

[0026] When the patient's rope-untying behavior is detected by the depthwise separable convolutional neural network model, a graded alarm is triggered via the wireless transmission module.

[0027] Furthermore, the depthwise separable convolutional neural network model is a TinyML model;

[0028] The wireless transmission module uses LoRaWAN low-power wireless transmission technology.

[0029] Furthermore, after triggering the graded alarm via the wireless transmission module, the following is also included:

[0030] Regularly collect system operation data to retrain and optimize the TinyML model, thereby improving its adaptability and accuracy.

[0031] Compared with the prior art, the beneficial effects of the present invention are:

[0032] (I) Significantly improves patient safety

[0033] Real-time monitoring and rapid response: The multimodal sensor restraint monitoring system can monitor the loosening status of restraints in real time, accurately identify the loosening action of 0.5N level knots, and trigger an alarm in a very short time, effectively reducing the safety risks after the patient loosens the restraints on their own;

[0034] Reduce the risk of complications: The medical clamping device, made of medical-grade TPU material in one piece, reduces irritation to the patient's skin and lowers the incidence of complications such as pressure sores and circulatory disorders.

[0035] (II) Improving nursing efficiency and quality

[0036] Intelligent Behavior Recognition: The multimodal sensing constraint monitoring system uses TinyML edge computing technology and embeds a deep separable convolutional neural network model to intelligently distinguish between the patient's normal activities and rope-untying behavior, significantly reducing the false alarm rate and improving the pertinence and efficiency of nursing work;

[0037] Dynamic adjustment of restraint tightness: The multimodal sensor restraint monitoring system can dynamically adjust the tightness of the restraint based on real-time monitoring data, optimize the patient's restraint experience, and reduce patient discomfort and safety hazards caused by improper restraint;

[0038] Reduce the frequency of nurses' rounds: Through digital constraint logs and risk warning functions, the multimodal sensor constraint monitoring system can automatically record the constraint status and issue real-time warnings, reducing the frequency of rounds by medical staff and saving labor costs;

[0039] (III) Promoting the intelligent transformation of nursing models

[0040] Multi-terminal linkage alarm: Using LoRaWAN low-power wireless transmission technology, multi-terminal linkage alarm can be realized within a certain range; alarm signals can be sent simultaneously to the mobile terminals of medical staff, ward monitoring systems and central nursing stations, ensuring that medical staff can receive alarm information in a timely manner and improve nursing efficiency;

[0041] Data sharing and remote monitoring: The multimodal sensing constraint monitoring system supports data sharing and remote monitoring, and can transmit monitoring data to the hospital's central nursing system in real time, which facilitates remote management and data analysis by medical staff and promotes the application of medical Internet of Things;

[0042] (iv) Improve the industry's technological level

[0043] Filling a technological gap: This invention fills a technological gap in intelligent restraint devices, providing a precise and humane solution for psychiatric nursing and promoting the intelligent transformation of the medical and nursing field.

[0044] Market potential and industry impetus: The domestic market is expected to be large, and this invention is expected to become an important driving force for the upgrading of the medical Internet of Things industry. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the framework of the multimodal sensing constraint monitoring system in an embodiment of the present invention;

[0046] Figure 2 This is a flowchart of the multimodal sensing constraint monitoring method in an embodiment of the present invention.

[0047] The attached figures are labeled as follows:

[0048] 1- Restraint strap, 2- Medical clamping device, 3- Accelerometer, 4- Gyroscope, 5- Flexible tension sensor, 6- Edge computing module, 7- Wireless transmission module. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this application, and not all embodiments.

[0050] See Figure 1 This invention provides a multimodal sensing constraint monitoring system, comprising:

[0051] 1. Restraint strap; 2. Medical clamping device; 4. Gyroscope; 5. Flexible tension sensor; 6. Edge computing module; and 7. Wireless transmission module.

[0052] The medical clamping device 2 is equipped with an acceleration sensor 3, which is used to clamp the restraint strap 1.

[0053] The gyroscope 4 is integrated into the medical clamping device 2 to capture limb rotation characteristics.

[0054] The flexible tension sensor 5 is integrated on the medical clamping device 2 and contacts the restraint strap 1 to provide real-time feedback of the restraint pressure.

[0055] The edge computing module 6 can embed a deep separable convolutional neural network model to intelligently distinguish between rope movements and normal activities.

[0056] The wireless transmission module 7 is used to realize multi-terminal linkage alarm within a certain range, ensuring that medical staff can receive alarm information in a timely manner and improve nursing efficiency.

[0057] Optionally, the medical clamping device 2 is integrally molded from medical-grade TPU material, which reduces irritation to the patient's skin and lowers the incidence of complications such as pressure sores and circulatory disorders.

[0058] The medical clamping device 2 has a built-in anti-disassembly trigger mechanism and temperature monitoring module, which enhances the safety and reliability of the system.

[0059] Optionally, the deep separable convolutional neural network model implanted in the edge computing module 6 is the TinyML model. The TinyML model uses a deep separable convolutional neural network to perform real-time analysis on the collected multimodal data, distinguish between the patient's normal activities and rope-untying behavior, reduce the false alarm rate, and improve the intelligence level of the system.

[0060] Optionally, the wireless transmission module 7 adopts LoRaWAN low-power wireless transmission technology, which can realize multi-terminal linkage alarm within a certain range. The alarm signal can be sent to the mobile terminals of medical staff, the ward monitoring system and the central nursing station at the same time, ensuring that medical staff can receive alarm information in a timely manner and improve nursing efficiency.

[0061] Optionally, the accelerometer 3 is a MWJS01 type accelerometer 3, and the MWJS01 type accelerometer 3 is a high-precision accelerometer 3.

[0062] This invention provides a multimodal sensing constraint monitoring system that integrates an accelerometer 3, a gyroscope 4, and a flexible tension sensor 5 to simultaneously capture physical displacement and biomechanical changes, achieving comprehensive monitoring of the constraint state. Through multimodal sensing and edge computing technology, it can accurately monitor and promptly issue alarms, significantly reducing the risk of patients escaping on their own. It also enables dynamic adjustment of constraint tightness, optimizing constraint conditions in real time based on the patient's actual condition.

[0063] like Figure 2 As shown, the present invention also provides a multimodal sensing constraint monitoring method, which employs the above-mentioned multimodal sensing constraint monitoring system and includes the following steps:

[0064] Step S1: Use medical clamping device 2 to clamp the restraint belt 1, monitor the knot displacement through acceleration sensor 3, capture limb rotation characteristics through gyroscope 4, and provide real-time feedback of restraint pressure through flexible tension sensor 5 to obtain sensor data.

[0065] Step S2: Use a dynamic baseline calibration algorithm to process the sensor data to eliminate environmental noise and ensure the accuracy of the sensor data;

[0066] Step S3: Combine wavelet transform to extract the frequency band features of rope-untying action from the sensor data, and input them into the deep separable convolutional neural network model implanted in the edge computing module 6 for behavior classification.

[0067] Step S4: Real-time analysis of the collected multimodal data is performed using a deep separable convolutional neural network model to distinguish between the patient's normal activities and rope-untying behavior;

[0068] Step S5: When the patient's rope-untying behavior is distinguished by the deep separable convolutional neural network model, a graded alarm is triggered through the wireless transmission module 7.

[0069] Optionally, the depthwise separable convolutional neural network model is the TinyML model;

[0070] The wireless transmission module 7 adopts LoRaWAN low-power wireless transmission technology.

[0071] Optionally, after triggering the graded alarm via the wireless transmission module 7, step S5 further includes:

[0072] Step S6: Regularly collect system operation data, retrain and optimize the TinyML model to improve its adaptability and accuracy.

[0073] The installation and commissioning steps of the multimodal sensing constraint monitoring system provided in this embodiment include:

[0074] Hardware Installation: Securely clamp the medical clamping device 2, which integrates the MWJS01 accelerometer 3, gyroscope 4, and flexible tension sensor 5, onto the restraint strap 1, ensuring tight contact between the sensors and the restraint strap 1. Configure the communication parameters of the LoRaWAN wireless transmission module 7 to ensure stable data transmission.

[0075] Software debugging: After system startup, dynamic baseline calibration was performed to eliminate environmental noise. The TinyML model was deployed to optimize its storage and computational requirements. The system's real-time monitoring and alarm functions were verified by simulating patient activities and knot loosening.

[0076] Real-time monitoring: The MWJS01 accelerometer 3, gyroscope 4, and flexible tension sensor 5 collect real-time displacement, rotation, and pressure data of the constraint band 1. Environmental noise is eliminated through a dynamic baseline calibration algorithm, and wavelet transform is used to extract the frequency band features of the rope untying action.

[0077] Intelligent Analysis and Alarm: The pre-processed data is input into the TinyML model for behavior classification, intelligently distinguishing between normal patient activities and rope-untying behavior. If rope-untying behavior is detected, the system sends an alarm signal to the mobile terminals of medical staff and the central nursing station via the LoRaWAN module.

[0078] Model optimization: Regularly collect system operation data to retrain and optimize the TinyML model, thereby improving its adaptability and accuracy.

[0079] Hardware optimization: Based on actual usage, the materials and structure of the medical clamping device 2 were optimized to improve the comfort and reliability of the system.

[0080] System maintenance: Conduct daily system inspections to check the secure connection of sensors. Regularly back up system operating data, analyze system performance, and provide data support for future optimizations.

[0081] This invention also provides applications of the above-mentioned multimodal sensing constraint monitoring system in different scenarios such as psychiatric wards and home care.

[0082] Psychiatric ward: A multimodal sensor restraint monitoring system is used to monitor the restraint of patients at risk of self-harm or harm to others, ensuring their safety during restraint. Through digital restraint logs and risk warning functions, it helps healthcare staff optimize nursing procedures and improve nursing efficiency.

[0083] Home care: The multimodal sensing constraint monitoring system uses LoRaWAN low-power wireless transmission technology to transmit monitoring data in real time to the mobile terminal or remote monitoring platform of medical staff, ensuring the safety of patients in the home environment.

[0084] The above description is merely the preferred embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A multimodal sensing constraint monitoring system, characterized in that, Comprising: Restraint strap; Medical clamping device, which is equipped with an acceleration sensor and is used to clamp the restraint strap; Gyroscope, which is integrated on the medical clamping device and is used to capture limb rotation characteristics; Flexible tension sensor, which is integrated on the medical clamping device and contacts the restraint strap, and is used to provide real-time feedback on the restraint pressure; Edge computing module, which can implant a depthwise separable convolutional neural network model and is used to intelligently distinguish the rope-untying action from normal activities; Wireless transmission module, which is used to achieve multi-terminal linkage alarm within a certain range.

2. The multi-modal sensing restraint monitoring system according to claim 1, wherein: The medical clamping device is built-in with an anti-disassembly trigger mechanism and a temperature monitoring module.

3. The multi-modal sensing restraint monitoring system according to claim 1, wherein: The depthwise separable convolutional neural network model implanted in the edge computing module is a TinyML model, and the TinyML model performs real-time analysis on the collected multi-modal data through the depthwise separable convolutional neural network.

4. The multi-modal sensing restraint monitoring system according to any one of claims 1-3, wherein: The wireless transmission module adopts LoRaWAN low-power wireless transmission technology.

5. The multi-modal sensing restraint monitoring system according to claim 4, wherein: The acceleration sensor is a MWJS01 type acceleration sensor.

6. The multi-modal sensing restraint monitoring system according to claim 4, wherein: The medical clamping device is integrally formed with a medical-grade TPU material.

7. A multimodal sensing constraint monitoring method, employing the multimodal sensing constraint monitoring system as described in any one of claims 1-6, characterized in that, Including the following steps: Using a medical clamping device to clamp the restraint strap, monitoring the knot displacement through an acceleration sensor, capturing limb rotation characteristics through a gyroscope, and providing real-time feedback on the restraint pressure through a flexible tension sensor to obtain sensing data; Performing environmental noise elimination processing on the sensing data through a dynamic baseline calibration algorithm to ensure the accuracy of the sensor data; Combining wavelet transform to extract the frequency band characteristics of the rope-untying action from the sensing data, and inputting them into the depthwise separable convolutional neural network model implanted in the edge computing module for behavior classification; Performing real-time analysis on the collected multi-modal data through the depthwise separable convolutional neural network model to distinguish the normal activities of the patient from the rope-untying behavior; When the depthwise separable convolutional neural network model distinguishes the rope-untying behavior of the patient, triggering a hierarchical alarm through the wireless transmission module.

8. The multi-modal sensing restraint monitoring method according to claim 7, wherein: The depthwise separable convolutional neural network model is a TinyML model; The wireless transmission module adopts LoRaWAN low-power wireless transmission technology.

9. The multimodal sensing constraint monitoring method according to claim 8, characterized in that, After triggering the hierarchical alarm through the wireless transmission module, it further includes: Regularly collecting system operation data, retraining and optimizing the TinyML model to improve the adaptability and accuracy of the TinyML model.

Images