Paster type thrombus monitoring and NO release closed-loop system for treating venous thrombosis

Abstract

The invention provides a patch type thrombus monitoring and NO release closed-loop system for treating venous thrombosis, which comprises a flexible patch substrate on which a sensor module is integrated, and the sensor module comprises a blood flow velocity sensor, a biological impedance sensor, an acceleration sensor and a temperature sensor, the acquisition units are respectively used for acquiring blood flow velocity v at a target vein, related parameters of vascular resistance R, limb movement states and skin temperature of a patch area; the NO release module is used for releasing NO to act on the venous thrombosis part; the control module comprises a signal processing unit and a control unit and is used for receiving the signal output by the sensor module and operating a closed-loop control algorithm to generate an NO release control instruction; and the power supply module provides electric energy for the sensor module, the control module and the NO release module. The invention provides a patch type closed-loop system which integrates thrombus multi-mode monitoring, decision making, NO release assisting and treatment mode self-adaptive switching.

Description

Technical Field

[0001] The present invention belongs to the field of venous thrombosis, and particularly relates to a patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombosis. Background Art

[0002] Venous thrombosis is a common disease with high morbidity and mortality globally. Especially deep vein thrombosis (DVT) of the lower extremities can not only cause pulmonary embolism but also lead to sequelae of chronic venous insufficiency, seriously affecting the quality of life of patients. Currently, the clinical treatment methods for venous thrombosis mainly include systemic anticoagulant therapy, thrombolytic therapy, and mechanical thrombectomy.

[0003] Systemic anticoagulant therapy (such as using low molecular weight heparin, warfarin, novel oral anticoagulants, etc.) is the most common method for treating DVT, aiming to prevent further growth of thrombus and formation of new thrombi. However, anticoagulant therapy usually relies on standardized dosing, lacks a mechanism to adjust the dose in real time according to the dynamic changes of the patient's thrombus, and has a significant bleeding risk, especially in elderly or patients with comorbidities.

[0004] Systemic thrombolytic therapy (such as intravenous injection of fibrinolytic drugs like rt-PA) can directly promote thrombus dissolution, but also faces problems such as a narrow treatment window and large side effects (high risk of bleeding, especially intracranial bleeding), and requires close monitoring of coagulation function, limiting its clinical application.

[0005] Mechanical thrombectomy techniques (such as catheter-assisted thrombus aspiration or fragmentation) can quickly remove thrombi, but are invasive operations with risks such as vascular injury, restenosis, and surgical-related complications. And currently, such means mainly focus on acute large thrombus situations and have insufficient management capabilities for chronic stages or microthrombi.

[0006] In response to the above problems, local drug release (local controlled release) strategies have received attention in recent years. By locally implanting or surface-attaching drug release systems, it is expected to enhance the local treatment effect while reducing the side effects of systemic exposure. For example, some studies have reported local release of antithrombotic drugs through drug-coated stents, drug release catheters, or skin patches containing NO donors.

[0007] However, the existing local drug delivery systems have the following general deficiencies: Lack the ability to monitor the real-time thrombus status. Currently, most local release systems are passive releases (such as time-controlled release, single-environment-responsive release), and do not dynamically adjust the release amount according to the actual changes of the thrombus. The formation, growth, and dissolution of thrombi during the treatment process are dynamically changing, and a single preset release rate cannot adapt to this dynamic situation, easily resulting in treatment lag or over-treatment.

[0008] The NO release technology has insufficient control precision. As an important molecule for local anti-thrombosis and vasodilation, nitric oxide (NO) has a very short half-life (less than 1 second), requiring the release system to have extremely high spatio-temporal precision. However, traditional NO release systems, such as NO donor coatings or passive drug delivery patches, usually cannot accurately regulate the NO release amount and release time, let alone adaptively adjust according to the feedback of thrombus evolution, posing risks of insufficient dosage or local overexposure.

[0009] A complete closed-loop feedback regulation mechanism has not been established. Among existing local treatment devices, there is no published literature on a closed-loop feedback system composed of real-time thrombus monitoring (such as blood flow blockage detection, thrombus imaging analysis) and precise local NO release. Most existing systems only continuously release until the NO donor is exhausted and cannot reduce the dosage or stop the drug in real time according to the actual dissolution progress of the thrombus, easily leading to over-treatment or side effects.

[0010] The lack of strategy switching between the acute thrombus stage and the chronic prevention and treatment stage. Thrombi have different NO requirements at different stages of formation (acute congestion stage, organization and chronic stage). Existing systems cannot automatically distinguish between the acute and chronic stages according to the change in thrombus load and dynamically adjust the NO release intensity and mode according to actual needs, resulting in a single treatment strategy and limited curative effect.

[0011] Based on the above deficiencies of the existing technologies, there is an urgent need for a treatment system that can: monitor the dynamic evolution of thrombi in real time and multimodally (such as blood flow velocity, thrombus density, elastic modulus, etc.); establish a mathematical model (such as a state space model) based on the thrombus state and make optimal decisions; precisely intervene in thrombus formation and dissolution through local NO release; form a sensing-decision-drug release closed-loop feedback control; automatically identify the acute / chronic stage of the thrombus and switch treatment strategies; and at the same time have the characteristics of flexibility, miniaturization, high integration, and being convenient for long-term wearing and use by patients.

[0012] The present invention is precisely proposed in response to the above pain points of the existing technologies, aiming to provide a patch-type closed-loop system integrating thrombus multimodal monitoring, decision-making, assisted NO release, and adaptive switching of treatment modes, providing a new technical means for the precise, safe, and efficient treatment of venous thrombosis. Summary of the Invention

[0013] The purpose of the present invention is to provide a patch-type closed-loop system integrating thrombus multimodal monitoring, decision-making, assisted NO release, and adaptive switching of treatment modes.

[0014] To achieve the above purpose, the present invention provides the following technical solution: A patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombosis, comprising: a flexible patch substrate, and integrated on the patch substrate are: The sensor module includes a blood flow velocity sensor, a bioimpedance sensor, an acceleration sensor, and a temperature sensor, which are respectively used to obtain the blood flow velocity v at the target vein, parameters related to vascular resistance R, the limb movement state, and the skin temperature of the patch area; The NO release module releases NO to act on the venous thrombosis site; The control module includes a signal processing unit and a control unit, receives the signals output by the sensor module, and runs a closed-loop control algorithm to generate NO release control instructions; among them, the control module calculates the thrombus load index x according to the collected blood flow velocity v, blood flow resistance R, temperature, and patient limb movement data 1 ; At the same time, according to x 1 Adjust the NO release amount of the NO release module; The power supply module provides electrical energy for the sensor module, the control module, and the NO release module.

[0015] Furthermore, x 1 Is calculated according to the following formula: Among them, v 0 And R 0 Are the reference blood flow velocity and the reference blood flow resistance respectively, α, β, and γ are weight coefficients, and f(other factors) is used to represent the influence of the auxiliary parameters detected by the accelerometer and the temperature sensor on the thrombus risk.

[0016] Furthermore, the NO release module includes: The drug reservoir is used to store the donor drug that can release NO; The surface acoustic wave driving unit is arranged near the drug reservoir and is used to generate surface acoustic waves under the drive of the control module to promote the release of NO from the donor drug; The microneedle array is connected to the drug reservoir and extends out of the patch substrate to contact the skin, and is used to deliver the released NO or the NO-containing medium to the subcutaneous target area; Among them, the surface acoustic wave driving unit adopts a piezoelectric substrate and an interdigital transducer structure, the drug reservoir is in the form of a porous matrix or a hydrogel, and the microneedle array is a hollow needle structure, so as to achieve efficient NO output and transmission.

[0017] Furthermore, the control module is configured to execute a model predictive control algorithm, including: internally storing a state space model that depicts thrombus evolution and the action of NO, and an objective function for evaluating the control effect; Among them, the state space model is defined by the following dynamic equations:

[0018] x 1 Is the thrombus load index, x 2$C_{local}$ is the local NO concentration, $u$ is the control input for the NO release rate, and $\alpha$, $\beta$, and $\gamma$ are all positive parameters; After obtaining the signal from the sensor module, the control module predicts the change of the thrombus in the future time domain based on the state space model, and determines the optimal control output of the NO release module by solving the following optimization objective function:

[0019] where $x$ 1 ref is the target thrombus volume value, $Q$ 1 is the control weight for the thrombus volume (or thrombus burden), and $R$ is the penalty weight for the NO release amount; through the above optimization, the thrombus index tends to the predetermined target value and the NO dosage is controlled, so as to inhibit the excessive NO release while ensuring the thrombolysis or thrombosis prevention effect.

[0020] Furthermore, there are two control modes in the control module, namely the acute thrombolysis mode and the chronic maintenance mode, and the mode is switched according to the comparison result between the thrombus volume $x$ 1 and the thrombus volume threshold $x$ th , where: When $x$ 1 >$x$ th , it is switched to the acute thrombolysis mode, and the control weight $Q$ for the thrombus volume in the objective function is increased 1 and the upper limit of the NO release rate is increased to increase the NO release dose for rapid thrombolysis; When $x$ 1 ≤$x$ th , it is switched to the chronic maintenance mode, and the control weight $Q$ for the thrombus volume in the objective function is decreased 1 and the upper limit of the NO release rate is decreased to release NO periodically at a low dose to maintain the thrombus stability and prevent recurrence; moreover, the control module can judge and automatically switch between the above two modes in real time according to the estimated value of the thrombus volume feedback by the sensor module.

[0021] Furthermore, the flexible patch substrate adopts a rigid-flexible combined structure, with the control module and the power module installed in the rigid area, and the sensor module and the NO release module installed in the flexible area; The rigid area and the flexible area are connected by flexible interconnection wires, and the back of the patch substrate has a medical adhesive layer for fixing the system on the skin surface and ensuring reliable and unbroken electrical connection between the modules during the movement of the patient's limb.

[0022] Furthermore, the power module includes a micro rechargeable battery and a power management circuit. The power management circuit has overcharge and over-discharge protection functions and integrates a wireless charging receiving coil, allowing the system to be charged by an external wireless charging device; the system also includes a wireless communication module that uses low-power Bluetooth wireless communication technology to send monitoring data and system status to an external terminal device and receive remote control instructions or parameter configurations to achieve remote monitoring and adjustment of the closed-loop system.

[0023] Furthermore, the blood flow velocity sensor in the sensor module is used to detect the venous blood flow velocity v and calculate it through the following formula: where L is the fixed distance between the two optoelectronic probes on the patch, and Δt is the transmission time difference of the pulse wave from the upstream sensor to the downstream sensor; the bioimpedance sensor is used to estimate vascular patency and infer vascular resistance R; the acceleration sensor is used to detect the movement intensity and posture of the target limb; the temperature sensor is used to collect the target skin temperature information; the control module fuses and calculates the thrombus load index x 1 and determines the NO release rate u through a model predictive control algorithm to control the thrombus load within a safe range Furthermore, a control method for a patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombosis includes the following steps: S1. The patch system is attached to the skin of the patient at the thrombus or thrombogenic site, and the sensor module is activated to monitor the thrombus state; S2. The control module obtains the sensor data and estimates the thrombus volume x 1 or the thrombus load index, and compares it with the preset thrombus volume threshold x th ; S3. In the acute thrombus mode, the model predictive control algorithm is executed to calculate the NO release amount in real time according to the state space model and the optimization objective function to quickly dissolve the thrombus; S4. In the chronic maintenance mode, NO is released periodically at a low dose to maintain the thrombus load stable at a safe level; S5. According to the real-time monitoring results, the above monitoring and control process is cycled until the thrombus load is reduced and stably maintained within the target range.

[0024] Furthermore, a safety constraint mechanism is introduced into the model predictive control algorithm, including: limiting the maximum NO release rate to avoid over-dosing; real-time monitoring of the patient's vital sign signals and setting safety interlock conditions. When abnormal vital signs or abnormal system operation are detected, immediately reduce or stop the NO release to ensure patient safety; and when the control algorithm is abnormal or the sensor fails, automatically switch to a preset conservative fixed-dose administration mode to ensure compliance with medical safety specification requirements during the operation of the closed-loop system.

[0025] In view of the clinical actual needs in the treatment of lower extremity deep vein thrombosis (DVT), the present invention provides a flexible integrated, closed-loop intelligent, and individualized regulated nitric oxide (NO) releasing patch system, which has the following specific clinical beneficial effects: Improve the onset speed and accuracy of acute thrombosis treatment. By using multi-modal sensors to continuously monitor thrombus burden (such as blood flow velocity, blood oxygen change, local tissue swelling and other indicators), the present invention can quickly identify the lesion area in the early stage of thrombus formation, and combine state space modeling and MPC intelligent optimization algorithm to timely adjust the NO release dose and frequency, achieve rapid thrombolysis in the acute stage, shorten the treatment window, and reduce the risk of serious complications such as thrombus detachment and pulmonary embolism.

[0026] Improve the effectiveness of preventing recurrence management in the chronic stage of thrombosis. For the chronic recovery stage after thrombus treatment, the present invention can automatically switch to a low-dose maintenance mode according to the thrombus burden index, periodically release NO in trace amounts, keep the blood vessels unobstructed, prevent re-embolism, meet the clinical requirements of long-term preventive management for chronic DVT patients, and avoid the lagged response of traditional passive monitoring means to thrombus recurrence.

[0027] Reduce the risk of complications such as bleeding during treatment. Compared with traditional systemic thrombolysis (such as intravenous injection of rt-PA), the present invention controls the precise release of local NO, which only acts locally on the thrombus and does not cause an increase in systemic fibrinolytic activity. Therefore, the bleeding risk is significantly reduced. By introducing the upper limit constraint of NO release rate and the interlocking mechanism of real-time vital sign monitoring, the patient safety during the treatment process is further ensured, especially suitable for high bleeding risk populations (such as the elderly, postoperative patients) to apply.

[0028] Achieve closed-loop treatment and reduce the frequency of medical staff intervention. The present invention adopts a closed-loop MPC control system, which does not rely on manual intervention to adjust NO release, and can automatically adjust the treatment strategy according to the thrombus evolution. Compared with the traditional treatment plan that requires doctors to repeatedly monitor and adjust the dose, it greatly reduces the burden on medical staff, improves the automation level of the treatment process, and is applicable to the daily management of hospitals and autonomous treatment in community and home environments.

[0029] Improve patient compliance and comfort, suitable for long-term wearable treatment. By adopting a flexible and bendable, rigid-flexible combined patch structure design, combined with a micro battery, wireless charging and low-power Bluetooth communication module, the patch of the present invention is light and comfortable to wear, can continuously cover the treatment area for a long time, adapt to the limb movement of patients, does not affect daily life activities, improves the treatment compliance of chronic patients, and helps to improve the long-term management effect of DVT.

[0030] The ability of individualized treatment and early intervention. By continuously monitoring the change of thrombus burden and dynamically comparing it with the individualized set target, the present invention can achieve a truly individualized treatment strategy. Especially in the initial stage of thrombosis with slight abnormalities (slight decrease in blood flow velocity, local oxygenation decrease), it can trigger the local release of micro-dose NO for intervention, and is expected to prophylactically block the further development of thrombus before the clinical symptoms of thrombus appear, realizing "ultra-early intervention".

[0031] Promote the transformation of DVT treatment from a centralized medical model to remote and home-based treatment. Thanks to the intelligent and portable design of the system of the present invention, patients do not need to be hospitalized for a long time and can continuously receive closed-loop management of thrombus in a home environment under the remote monitoring of doctors. This not only reduces the burden on medical resources, but also reduces the economic and psychological pressure brought by hospitalization to patients, conforming to the trend of the modern medical system towards chronic disease management and home medical care.

[0032] In summary, the patch-type thrombus monitoring and NO release closed-loop system of the present invention closely meets the current clinical treatment needs of venous thrombus. It can not only improve the efficacy and safety of DVT treatment, but also optimize the allocation of medical resources, improve the quality of life of patients, and has extremely high clinical application value and promotion prospects. Brief Description of the Drawings

[0033] Figure 1 is a schematic diagram of the overall structure of the system of the present invention; Figure 2 is the signal processing and control instruction output process of the control module of the present invention; Figure 3 is the closed-loop working process of the model predictive control (MPC) of the present invention. Detailed Description of the Specific Embodiment

[0034] The following further describes the technical solutions of the present invention in detail with reference to the drawings and specific embodiments. The described embodiments are for the purpose of illustrating the present invention and are not intended to limit the protection scope of the present invention. Figure See Figure 1 This embodiment provides: A patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombus. The overall structure of the system is as shown in Figure 1As shown in the figure, it includes: a sensor module, a control module, a NO release module, and a power module. Each module forms a closed-loop control circuit through electrical connection and signal interaction. Among them, the sensor module is used to collect physiological signals reflecting the DVT risk. The control module processes the signals and runs a model predictive control algorithm to generate control instructions. The NO release module precisely releases NO according to the control instructions to implement intervention, and the power module provides stable electrical energy for each part of the system. The following combines the attached drawings to explain the structural composition, functions of each module and their specific implementation in DVT prevention one by one, and elaborates on the system signal acquisition and processing flow, the model predictive control algorithm, and the closed-loop control workflow between modules.

[0035] The sensor module is mainly composed of multiple types of physiological parameter sensors, which are used to collect various signals reflecting the risk of deep vein thrombosis and send them to the control module for processing. Preferably, the sensor module is integrated on a flexible patch and can be attached to the skin surface of the DVT-prone area of the patient's lower limb (such as the calf). The sensor module includes, but is not limited to, the following sensor elements: Blood flow velocity sensor: used to measure the blood flow velocity v in the target vein. One implementation method is to use a photoplethysmogram (PPG) sensor array: for example, two optoelectronic sensors (light-emitting diodes and photodiodes) are arranged along the flow direction on the superficial vein path on the skin surface, and the time difference Δt of the pulse wave reaching the upstream and downstream points is recorded. The blood flow velocity is calculated according to the known distance L. Specifically, the formula can be used to estimate the blood flow velocity, where L is the fixed distance between the two sensors and Δt is the time difference of the pulse wave propagating from the upstream sensor to the downstream sensor. Another method is to integrate a micro ultrasonic Doppler sensor in the patch and deduce the blood flow velocity through the Doppler frequency shift formula. However, considering the wearability and power consumption of the system, it is preferred to use a non-invasive optical method or electrical method to obtain the blood flow velocity. The blood flow velocity is a key indicator for evaluating whether venous blood is stagnant: when the blood flow velocity is significantly reduced, it means that the blood stays for a long time, resulting in blood stasis and an increased DVT risk.

[0036] Pressure / strain sensor: used to detect local pressure changes or deformations of the limb to reflect the degree of venous filling and tissue swelling. The sensor can be a flexible strain gauge or a pressure sensor, which is arranged at the position where the patch contacts the limb. When a deep vein thrombosis occurs, the distal venous pressure increases, and the limb may show mild swelling or muscle tension changes. The strain / pressure sensor can sense the subtle changes in the surrounding tension on the skin surface. For example, a flexible strain gauge is arranged around the calf to monitor the change trend of the calf circumference; if there is a continuous mild swelling, it may indicate venous return obstruction. The sensor can also sense the squeezing pressure generated by the patient's muscle contraction to judge the muscle pump activity and correspondingly understand the venous emptying efficiency.

[0037] Bioimpedance Sensor: It is used to measure the impedance of local tissues to infer changes in blood volume and vascular patency. A bioimpedance sensor generally consists of one or more pairs of electrodes (conductive electrode patches are set on the patch). By injecting a weak alternating current between tissues and measuring the voltage, the complex impedance value of the site can be obtained. Blood is a conductive liquid medium, and its impedance value changes with the increase or decrease of blood content: when venous filling increases (blood content rises), tissue impedance decreases, and conversely, impedance rises when blood decreases. Therefore, this system uses impedance sensing to obtain information on changes in venous blood volume. For example, after each heartbeat pumps blood, the blood in the calf veins gradually returns to the heart, and the impedance will increase accordingly; if a thrombus causes partial obstruction of the vein, the return will slow down, and the impedance will remain at a low level for a longer time. By analyzing the changes in the impedance waveform with the cardiac cycle and the baseline value of the impedance at rest, the degree of blood flow blockage can be inferred. Bioimpedance measurement can also be used to calculate blood flow resistance (see below), because impedance is inversely proportional to the cross-sectional area of blood vessels and can reflect changes in blood vessel radius.

[0038] Temperature Sensor: It is used to monitor the temperature of the skin in the patch area. When DVT occurs, local inflammatory reactions or circulatory changes may occur, leading to changes in the skin temperature of the affected limb. For example, when a thrombus causes local inflammation, the skin temperature may rise; conversely, if local blood circulation is poor, the temperature of the limb end may also decrease. The miniature temperature sensor integrated on the patch (such as a thermistor or semiconductor thermometer) can accurately sense the subtle changes and trends of skin temperature and serve as an auxiliary judgment index. The temperature sensor is small in size and low in power consumption and can be placed at the position in the center of the patch that contacts the skin to continuously provide environmental and skin temperature data. Combining with other sensor data, abnormal changes in temperature can be used as a hint of thrombus risk (for example, if the skin temperature rises accompanied by swelling, it may indicate the formation of deep vein inflammatory thrombus).

[0039] Acceleration Sensor: It is used to detect the motion state of the limb. An accelerometer (such as a three-axis MEMS accelerometer) can be mounted on the flexible patch to sense the motion intensity, posture changes, and vibration signals of the calf. One of the main risk factors for DVT is prolonged inactivity, so it is very important to monitor whether the patient's limb is in a stationary state for a long time. When the acceleration sensor detects that there is almost no obvious movement in the lower limb for a long time (for example, there is almost no acceleration change for dozens of minutes), the control module can use this information to improve the thrombus risk assessment result (increase the risk index x 1 ) and trigger a more active NO release strategy. On the contrary, if the accelerometer detects regular activities or muscle contractions (for example, the patient is walking or performing dorsiflexion of the foot), it indicates that there is sufficient power for blood flow, and the system can reduce the intervention intensity. The accelerometer can also judge the relative position of the limb (such as raising the leg helps venous drainage) and incorporate these factors into the closed-loop control adjustment.

[0040] The above-mentioned sensor elements are integrally arranged on the flexible patch. Each sensor is reasonably distributed according to its function: for example, the photoelectric sensor is arranged along the direction of the superficial vein, the bio-impedance electrodes are attached in pairs on both sides of the calf to measure the impedance across the limb, the temperature sensor is located in the center of the patch in contact with the skin, the pressure / strain sensors are placed circumferentially around the limb, and the accelerometer is fixed on the patch to sense the overall movement. The sensor module also includes primary signal conditioning circuits such as amplifiers, filters, and multiplexers to ensure the signal quality of each sensor. In terms of selection, each sensor needs to meet the requirements of high sensitivity, fast response speed, and having flexible packaging, low power consumption, etc. For example, the PPG sensor preferably uses an infrared-wavelength LED to transmit deeper tissues and reduce ambient light interference; the impedance measurement electrodes use medical conductive materials (such as Ag / AgCl) to reduce contact impedance and avoid skin irritation; the accelerometer selects a three-axis device with low noise, which can detect tiny vibrations in a stationary state; the temperature sensor requires an accuracy within 0.1 °C and a short thermal response time. The sensor module is integrated through a medical flexible substrate, and the patch needs to have good biocompatibility and adhesiveness to ensure that long-term application will not cause skin discomfort or sensor displacement.

[0041] In the application of DVT prevention, the sensor module is optimally arranged to capture the early signs of DVT formation. The fusion of multiple sensor information enables the system to comprehensively monitor the venous circulation status. For example, when the patient's lower limb is stationary, the accelerometer reading is close to zero and remains so for a long time. At the same time, the photo / impedance sensing shows a decrease in blood flow velocity, slower venous emptying, and a gradual increase in temperature. Then these signals together indicate a high-risk situation of deep vein thrombosis formation. Compared with a single indicator, the multi-modal monitoring of the sensor module greatly improves the reliability and sensitivity of risk detection, providing sufficient data support for subsequent control decisions.

[0042] The control module consists of two main parts: an electronic hardware circuit and a control algorithm software. It is used to receive and process the signals collected by the sensor module, calculate various parameters and fusion indicators characterizing the DVT risk, and generate control instructions for the NO release module based on the model predictive control strategy to achieve closed-loop regulation. The control module is generally composed of a microcontroller or an embedded processor, combined with necessary analog-to-digital conversion, power management, and communication interface circuits, and is encapsulated in the patch or in a micro control unit connected to the patch through wires.

[0043] In terms of hardware, the control module integrates an analog-to-digital converter (ADC) to collect analog signals from the sensor module. For example, the multi-channel ADC sequentially reads the PPG photodiode output, electrode impedance measurement voltage, strain gauge resistance value, etc., and converts them into digital quantities. The control module has a built-in memory for storing measurement calibration parameters, model parameters, as well as the program code and data required for running the control algorithm. To ensure real-time performance, the selected microcontroller needs to have sufficient computing speed and accuracy (for example, having a hardware floating-point arithmetic unit to perform MPC optimization calculations in real time). In addition, the control module may include a wireless communication interface or a wired communication port for communicating with external devices (such as transmitting data to a doctor monitoring system or receiving parameter adjustment instructions), but these are not essential parts of the closed-loop control and can be added according to application requirements. The control module is powered by a power module, and has a voltage stabilization circuit inside to ensure a stable voltage for analog-to-digital conversion, the electric processor core, etc. Its circuit design meets the low-power requirements, dynamically manages power consumption between intermittent operation of the MPC algorithm and continuous sampling to adapt to the battery-powered characteristics.

[0044] In terms of software and algorithms, the control module executes the following main functional processes Figure 2 as shown in the signal processing and control signal output process of the control module]: (1) Signal preprocessing and feature extraction: The control module filters, calibrates, and calculates features for the collected raw sensor data. First, digital filtering algorithms are applied to each channel signal to remove noise and interference. For example, a band-pass filter is used for the PPG signal to extract the pulse wave component and filter out DC and high-frequency noise; motion artifact correction is performed on the acceleration signal to avoid motion interfering with the baselines of the photoelectric and impedance signals. Then, combined with the system's initial calibration parameters (such as the zero points and sensitivities of each sensor), the raw readings are converted into physical quantities (such as converting ADC counts into voltage, and then the voltage into impedance values or temperature values). On this basis, key feature parameters are extracted: Calculation of blood flow velocity v: The control module calculates the blood flow velocity v based on the pulse wave time delay measurement provided by the sensor module. For example, when the patch contains two PPG sensors, the control module detects the pulse waveforms of the upstream sensor and the downstream sensor, calculates the time difference Δt of their characteristic points (such as the starting rising edge of the waveform), and then combines the center distance L between the two sensors to calculate it. In addition, if a single-probe Doppler sensor is used, the control module calculates the flow velocity according to the received Doppler frequency shift Δf using the Doppler formula (where c is the propagation speed of ultrasonic waves in tissue, f 0 is the transmitted ultrasonic frequency, and θ is the angle between the ultrasonic wave and the blood flow direction). Regardless of the measurement method used, the obtained blood flow velocity v (which can be the average flow velocity or the peak flow velocity) is passed as an important feature to the subsequent algorithm.

[0045] Calculation of blood flow resistance R: Blood flow resistance is an important indicator reflecting the patency of blood vessels. It usually cannot be directly measured but is inferred through relevant parameters. The control module can estimate the value or change trend of R through various means. One method is to use Ohm's law analogy: If the pressure difference ΔP across a certain blood vessel segment (which can be approximated by inferring venous pressure or measured by a sensor) and the blood flow Q of this segment (obtained by multiplying the flow velocity v by the cross-sectional area A, Q = A·v) are obtained, then the blood flow resistance R can be defined as ΔP / Q. In the absence of direct pressure measurement, the control module can estimate R based on a vascular mechanics model. According to Poiseuille's law, under laminar flow conditions, the vascular resistance is related to blood viscosity and vascular geometry as follows: where η is the blood viscosity coefficient (the approximate value of blood viscosity at 37°C can be taken as about 4 times the water viscosity), L is the length of the measured blood vessel segment, and r is the inner radius of the blood vessel. This formula reflects the strong influence of blood vessel radius on resistance (a decrease in r will cause a significant increase in R). Therefore, the control module can infer R by using the change in blood vessel radius estimated by a bioimpedance sensor: within the limb segment covered by the patch, the impedance value is inversely proportional to the cross-sectional area of the blood vessel, and thus approximately inversely proportional to r 2 When the impedance measurement shows a reduction in blood vessel volume, it can be inferred that r decreases, and thus the increase amplitude of R relative to the reference can be calculated according to the above formula. The control module can pre-store the reference blood flow resistance R 0 under normal physiological conditions (which can be estimated from normal blood vessel radius and viscosity), and quantify the change rate of blood flow resistance by comparing the currently estimated R with R 0 . Additionally, if a pressure sensor is configured to monitor superficial venous pressure or tissue tension, the resistance value in the form of Ohm's law can also be obtained by combining the flow velocity. Whether using model inference or data fusion, the obtained blood flow resistance index (which can be denoted as R or R~) will be used for subsequent thrombus risk assessment.

[0046] Other characteristic parameters: The control module will also record the statistical results of the accelerometer (such as the integral of exercise intensity over a recent period of time, the stationary time t immobilet , etc.), the temperature trend (the change rate of ΔT), and the change value of the strain gauge (reflecting the degree of limb micro-swelling), etc. These parameters are not directly used in the formula for calculating hemodynamics, but can be input as risk factors into the fusion algorithm. For example, if t immobilet exceeds a preset threshold (such as no obvious movement for 1 hour), the system can increase the risk estimate; if the skin temperature rises by more than a certain amplitude compared to the baseline, it can also be counted as an abnormal signal.

[0047] (2) Formation of the thrombus load fusion index x 1 The control module fuses and calculates the above multiple sensing characteristics to form a comprehensive index x 1, for quantifying the current deep vein thrombosis burden or the risk of occurrence. The design algorithm of this index can be formulated based on medical experience and data-driven models. For example, it can adopt the form of weighted linear combination: where v 0 and R 0 are the reference blood flow velocity and blood flow resistance under healthy conditions respectively, v and R are the current real-time calculated values, α and β are weight coefficients used to adjust the relative contribution of velocity and resistance to the risk; the third term f(⋅) represents the combined effect of other factors (such as exercise, temperature, etc.), which can be obtained by mapping the stationary time of the accelerometer, the temperature deviation from the normal value, etc. through a certain function, and γ is its weight. If the influence of other factors is minor, it can also be ignored and only the two terms of v and R are fused. In the above model, when the blood flow velocity v is lower than the reference (i.e., the blood flow slows down), or the blood flow resistance R is higher than the reference (i.e., the vascular resistance increases), then the difference between the two terms is positive, making x 1 increase; conversely, if the blood flow is unobstructed (v is close to normal and R is low), then x 1 is close to 0. By selecting appropriate α and β (and γ), x 1 can map the level of thrombosis risk. When x 1 exceeds a certain threshold, it can be considered that there is a high risk of DVT or even a hint that thrombosis may have already formed. The control module can dynamically adjust the values of α and β to adapt to individual differences (for example, adjust the proportion of each item in the total risk according to the patient's baseline signs or clinical risk score).

[0048] It should be noted that x 1 as the "thrombosis burden fusion index" can not only be calculated instantaneously, but also be designed as an index with memory accumulation characteristics. For example, the control module can perform time integration or smoothing on x 1 to simulate the characteristic that thrombosis formation is a gradual accumulation process: when the blood flow is continuously poor for a period of time, then x 1 gradually accumulates and increases, and conversely, a short-term improvement may only cause x 1 to slightly decline but still maintain a certain warning level. This dynamic behavior can be achieved by introducing state variables in discrete time steps (see the following state space model setting), or by adding an integral term to the fusion formula. Regardless of the specific implementation details, the real-time updated x 1 index will be used as the key state / output of the MPC control algorithm to judge the NO release demand.

[0049] (3) Model Predictive Control Algorithm: The control module adopts the Model Predictive Control (MPC) strategy to perform real-time optimization control on NO release. This is the core algorithm for the system to achieve closed-loop fine adjustment. MPC builds a model for the controlled object, predicts the future behavior of the system at each control moment, and solves the optimal control action, thereby realizing the optimal control of multiple constraints and multiple objectives. The following elaborates in detail on the settings and working processes of MPC in this system: State Space Model Setting: First, the control module establishes a discrete-time state space model for the local venous hemodynamics of the human body. The fusion index x 1 can be regarded as the main state or output of the system. At the same time, other state variables are introduced as needed to describe the dynamic characteristics of the system. For example, in a simplified model, the state vector is defined as , where x 1 represents the aforementioned thrombus load index, and x 2 can represent another state related to blood flow (such as blood flow velocity deviation, blood vessel radius change amount, or muscle pump activation state, etc., abstract quantities). The control input of the system is defined as u, representing the NO release rate or dose (which can be represented by a continuous variable, such as the concentration of NO release or the driving current magnitude of the electrochemical generator). The output variable y can be taken as a measurement quantity related to the state, such as (that is, the main output concerned by the system is the risk index). Based on medical knowledge and experimental data, we assume that under a small sampling period Δt \D (for example, calculating control every Δt seconds), the state evolution of the system can be approximately described by a linear model: where k is the discrete-time step index, A is the state transition matrix, B is the control influence matrix, C is the output matrix, w(k) represents external disturbances or unmodeled dynamics, and E is the disturbance influence matrix (which can be simplified and ignored). For this system, the model matrices can be set according to the understanding of the role of NO: The influence of NO release on the state x 1 should be to reduce it, and the influence on x 2 (such as blood flow velocity) is to increase it positively. Therefore, the matrix B takes a negative value in the first row accordingly (indicating that an increase in u will cause x 1 to decrease), and a positive value in the second row (an increase in u promotes x 2 to rise). When there is no NO input (u = 0), the system may cause x 1 to gradually increase due to slow blood flow, which can be reflected by the element (1,1) in the A matrix being slightly greater than 1 (indicating that without intervention, the risk index will accumulate over time); at the same time, x 2It may decline slowly without intervention or exhibit attenuation characteristics in A. The above is only a qualitative description, and the actual values can be obtained through empirical approximation of physiological processes or offline experimental calibration. For example, according to clinical observations, if there is no activity for a long time, the DVT risk shows a gradually increasing trend, which can be fitted to x 1 The proportion of increase per minute; NO administration can make this increase zero or reverse to a decrease. The control module can pre-store a set of model parameters A, B, C and fine-tune them according to individual responses during operation to improve the model accuracy. Although the process of human thrombus formation is complex and non-linear, the model required by MPC does not require 100% accuracy. It provides an approximation for guiding predictions and can still achieve good control effects under closed-loop correction.

[0050] Optimization objective function definition: The control module sets an optimized objective function J for control decisions. This objective function reflects the performance indicators that the system hopes to achieve within the prediction time range. In the DVT prevention scenario, the main goal is to keep the thrombus risk index x 1 at a low level while avoiding excessive release of NO to prevent side effects and save resources. Therefore, an objective function of the following form can be constructed: where N p is the number of steps in the prediction horizon (e.g., predicting the behavior in the next N p discrete periods), represents the thrombus index value at the i-th step in the future, is the corresponding control input value; Q x and R u are non-negative weight coefficients used to balance the relative importance of the risk index and control input. The above J is a typical quadratic performance index. The first part of the sum term (x 1 ) 2 is expected to be as small as possible (the lower the risk, the better), and the second part (u) 2 represents the cost of NO release (or change) and is also expected to be as small as possible (meaning achieving the goal with as little NO as possible). By adjusting the ratio of Q x and R u , the aggressiveness of control can be weighed: for example, assigning a larger weight to Q x will pay more attention to suppressing x 1 , even if a larger dose of NO is required; while a larger R u will tend to save NO and administer as little drug as possible if the risk is not high. In addition, constraint conditions can be added to the optimization, such as: NO release upper limit constraint , ensuring that the calculated optimal control sequence does not violate the safe dose (to avoid side effects such as systemic hypotension caused by excessive NO); state / output constraints such as (The risk index should not exceed a certain safety upper limit), etc., which are set as needed. MPC comprehensively considers these objectives and constraints, and solves for the future control action sequence that satisfies the constraints and minimizes J. .

[0051] Prediction and optimization solution: At each control time k, the control module obtains the current system state x(k) (i.e., the just-calculated x 1 and the possible x 2 and other states, which are estimated by an observer if necessary). Next, using the above state-space model, predict the trajectories of the system state and output changing with the control sequence within the next N p steps. Specifically, the control module will simulate the model equations in the computer: starting from the current x(k), applying a set of candidate control sequences , recursively calculate and the corresponding . Since there are countless combinations of possible control sequences, the control module uses an optimization algorithm (such as a quadratic programming QP solver) to efficiently search for the optimal sequence, that is, the control sequence that minimizes the objective function J and satisfies the constraints . In a typical MPC implementation, since J is quadratic and the model is linear (or linearized), this optimization is a convex quadratic programming problem and can be solved in a relatively short time. The processor of the control module completes this solution process in each cycle by calling a pre-written and tuned optimization library.

[0052] Applying control and rolling horizon: After the optimization solution, MPC obtains the optimal control sequence for a future period of time. However, for real-time and robustness considerations, only the first control quantity u*(k) in the sequence is applied to the current system. That is, the control module immediately sends a control instruction to the NO release module according to the magnitude of u*(k) (see the NO release module section below for details) to perform the current NO release action. Thus, the control decision at the current moment is completed. Subsequently, the system enters the next discrete time k + 1, the sensor module collects and updates the data again, the control module recalculates the new x(k + 1) and rolls the prediction horizon forward (rolling optimization). It will re-plan the optimal control for the next N p steps based on the latest state. This process repeats in a loop, forming a "prediction-optimization-applying control-rolling" closed-loop control process. This rolling horizon MPC strategy ensures that the control strategy can be continuously adjusted according to the latest system feedback. Even if there are certain deviations in the model or changes in the environment, MPC can ensure the control effect through continuous feedback correction.

[0053] The working process of the above MPC algorithm can be summarized as follows [cooperating with Figure 3 the flowchart shown]. State Measurement and Estimation: Read the current sensor data and calculate the state x(k) (mainly the current x 1 , and other state variables).

[0054] Model Prediction: Based on the system state space model, predict the behavior of the system within the next N p steps starting from x(k).

[0055] Construct and Minimize the Objective Function: Optimize the future control sequence according to the defined objective function J and constraints, and solve for the sequence that minimizes J (the optimal sequence).

[0056] Output Control Command: Take the first step u*(k) of the optimal sequence as the current control command and send it to the actuator (NO release module) to actually apply control.

[0057] Receding Horizon Optimization: Advance the time to k + 1, feedback the new state, and repeat steps 1 - 4 to perform prediction and optimization again on the next time domain window.

[0058] Through the above MPC closed-loop control, the system can actively and continuously adjust the NO release to maintain a good venous flow state. For example, if the sensor detects that the patient has been stationary for a long time and the blood flow is gradually slowing down, the control module will predict that x 1 will rise to a dangerous value in the future, so it will increase the NO release in advance (increase u) to promote vasodilation and accelerate blood flow, thus preventing problems before they occur and avoiding x 1 from actually reaching the dangerous threshold. If the patient starts to move later (e.g., performs foot exercises), the sensor will sense the improvement in blood flow, and x 1 will decrease accordingly. The MPC will then reduce the NO dose accordingly to avoid unnecessary drug release. Compared with simple threshold control or open-loop timed drug delivery, this predictive and receding horizon optimization control method can more precisely balance efficacy and safety, ensuring that the DVT risk is always under control while minimizing the NO dosage and power consumption and improving system efficiency.

[0059] It should be noted that while implementing the above complex algorithm, the control module can also take into account some safety and auxiliary functions. For example, the control module can set a dual threshold mechanism: when x 1 continues to rise close to the dangerous upper limit, even if the MPC is still gradually increasing NO, it can trigger an alarm or force the system into a safe mode (e.g., lock u at the maximum safe value u max and continuously release it, and send an alarm signal to prompt medical staff to intervene). Another example is that the control module can record historical sensing data and control behaviors into internal storage or upload them regularly for medical staff to analyze and adjust the MPC parameters (such as adjusting Q x or R uweight to control more actively or conservatively). These measures ensure the safety and reliability of the system operation under various conditions.

[0060] In summary, through high-speed data processing and model prediction optimization, the control module realizes the intelligent control of NO release in DVT prevention. It takes the thrombus load index x 1 as the core, constructs a closed-loop feedback, and converts multi-sensing signals into optimal intervention measures. The model predictive control algorithm enables the system to have a forward-looking adjustment ability, and can actively intervene according to the development trend of DVT risk, preventing thrombus formation more effectively than traditional passive responses.

[0061] The NO release module is the actuator module of this system, which is used to release a certain dose of nitric oxide (NO) to the local tissue of the patient as needed according to the instructions output by the control module, so as to play the role of vasodilation and anti-thrombosis. The structure of this module can be implemented in various forms, preferably a miniaturized and controllable NO supply device, integrated on the flexible patch, close to the target action area. The NO release module needs to meet the requirements of real-time controllability, accurate dosage, safety and portability, etc.

[0062] In a specific embodiment, the NO release module adopts the design of an electrochemical NO generation unit. It includes a small liquid storage cavity containing a NO precursor chemical substance (such as nitrite solution or S-nitrosoglutathione and other compounds that can release NO), and a pair of microelectrodes arranged in the liquid storage cavity. When the control module applies current through the electrodes, an electrochemical reaction occurs to decompose the precursor substance to generate NO gas. The generated NO gas is slowly released through the micropores or permeable membrane on the module and penetrates through the skin into the local tissue. The control module can precisely control the generation rate of NO by adjusting the magnitude and duration of the electrode current: increasing the current increases the NO generation rate (the release dosage increases), and decreasing the current decreases the NO generation rate or even stops (at this time, only a trace amount of background release occurs). The materials of the electrochemical NO generator are selected to be non-toxic to the products and corrosion-resistant for the electrodes, such as platinum or carbon-based electrodes, and the liquid storage and membrane materials are selected as medical polymer materials to ensure biocompatibility for long-term contact with the skin. The entire generation unit is small in size (can be on the order of square centimeters) and can be embedded in the patch without significantly increasing the thickness.

[0063] As another implementation, the NO release module can also adopt a microfluidic pump-based drug delivery mechanism. In this solution, the patch integrates a micro drug reservoir pre-filled with a NO donor drug (such as nitroglycerin preparations or NO-rich gels). The module is built with a micro pump and an electronically controlled valve. When NO release is required, the control module sends an instruction to open the valve and start the micro pump, quantitatively pushing out a very small volume of the drug from the reservoir, which is released through a diffusion medium (such as a hydrophilic gel pad or a porous membrane) at the patch-skin contact area. Once the drug reaches the gel pad, a chemical reaction or volatilization occurs, releasing NO gas that penetrates into the skin. By controlling the pumping time length or frequency, the released dose can be accurately controlled. This method is different from the electrochemical generation principle, but it can also achieve closed-loop modulation of NO output. Its advantage is that the drug concentration is preset and the release amount is easy to linearly control. The disadvantage is that the drug reservoir needs to be replenished or replaced regularly (suitable for short-term application or replaceable patches).

[0064] Regardless of the specific structure adopted, the NO release module generally includes a release interface, that is, a medium layer that contacts the skin and transmits NO. A common design is to set a microporous membrane or a polymer gel on the skin contact surface of the patch. This layer of medium ensures that NO diffuses into the skin evenly and slowly on the one hand, avoiding local irritation caused by overly concentrated NO, and on the other hand, it also plays a sealing role to prevent the leakage of NO precursor substances or NO gas when release is not required. For example, a polytetrafluoroethylene (PTFE) microporous membrane or a silicone gel layer can be used, which has a certain transmittance to NO and a blocking effect on liquid precursors. After the electrode generates NO, the NO molecules can pass through the membrane and enter the skin, while the precursor liquid remains in the reservoir and does not directly contact the skin.

[0065] The NO release module is connected to the control module circuit through a wire and receives the control signals output by the control module (such as voltage, current, or start timing signal). In a typical case, a DAC or PWM port of the control module is connected to the NO generating electrode through a drive circuit to adjust the current; if the micro pump scheme is adopted, the control module controls the opening of the micro valve and drives the motor of the micro pump through a signal line. The entire NO release unit consumes a certain amount of power during operation (electrochemical electrolysis consumes electrical energy, and the micro pump requires driving power), so the power supply module needs to provide corresponding short-term high-power supply capabilities. However, due to the intermittent and very small-dose release, the overall energy consumption is still within a controllable range.

[0066] In one embodiment, there are such installation layout requirements that the NO release module should be located on the patch near the deep vein. For example, for calf DVT prevention, the NO release unit can be placed in the center of the patch on the back of the calf, corresponding to the area near the gastrocnemius muscle - there are soleus venous plexuses and others under the skin in this area, which communicate with the deep vein, and the NO released here can affect the dilation of the deep vein through microcirculation diffusion. If the patch is used for the thigh, it can be placed on the skin surface in the projection area of the femoral vein. By anatomical positioning, the position of NO release is optimized to make the released NO reach the deep vein endothelium as soon as possible. There is literature indicating that the diffusion radius of NO molecules in biological tissues is limited and the half-life is also short (in seconds), so local release and nearby action are important features of the present invention. The patch delivers NO directly to the vicinity of the target area, acting on the venous blood vessels at a local high concentration without significantly entering the systemic circulation and causing adverse reactions such as hypotension.

[0067] The NO release module undertakes the function of performing treatment in the closed-loop system, and its performance is directly related to the curative effect and safety. The following points need to be particularly noted in the selection and design: Response speed: The module must be able to quickly respond to control instructions and adjust the NO release amount within a few seconds. This means that the electrochemical reaction or pumping mechanism has small hysteresis and inertia. By miniaturizing the electrodes and the micro-pump cavity, the response time can be controlled within the second level. Quick response ensures that the control actions calculated by the MPC can timely affect the physiological state, otherwise excessive delay will reduce the effect of closed-loop control. Dose accuracy: The module should have linearly adjustable output within the entire working range (0 to the maximum dose u max ). The NO production rate of the electrochemical device is usually proportional to the current, and the control module can achieve precise control through the current calibration curve. For the micro-pump, the pumping volume is linearly related to the opening time and is also easy to measure. The system can calibrate the NO output before leaving the factory or during use. For example, by measuring the NO concentration by colorimetry, the correspondence between the control signal and the actual release amount is established and stored in the control module for reference.

[0068] Safety control: The module must prevent accidental large-dose release or leakage. Therefore, redundant shutdown measures are designed, such as the electrochemical circuit presenting an open-circuit zero-current state without a control signal, and the micro-pump automatically closing the valve when powered off. In addition, a simple NO concentration monitoring feedback (such as a metal oxide NO gas sensor) can be integrated near the release port to detect whether the output is within the expected range, and once deviated, the control module is notified to adjust or stop. Capacity and continuous working ability: The reserve of NO precursors or drugs in the NO release module should be able to meet the needs of at least one typical preventive course. For example, the designed capacity allows the device to continuously operate for dozens of hours with intermittent release without exhaustion. If longer preventive time is required, it can be designed as a replaceable patch or cartridge. In the application of DVT prevention, the NO release module achieves optimized intervention for the deep vein thrombosis mechanism through the closed-loop control: NO is an endogenous potent vasodilator and antiplatelet aggregator. When NO is locally released, it diffuses to the smooth muscle of the venous blood vessels, causing vasodilation and an increase in the vessel diameter r. Then, according to the aforementioned relationship, the blood flow resistance R is significantly reduced, and the venous flow Q and velocity v increase accordingly. This effectively alleviates the state of blood stasis and reduces the promoting factors for thrombosis. In addition, the action of NO on the vascular endothelium can also inhibit the adhesion and aggregation of platelets, reduce the coagulation activity, and further directly counteract another important link in thrombosis formation. Since the half-life of NO is extremely short and it only acts locally, the system can release NO in a high-frequency, low-dose pulsed manner to continuously maintain the effect without accumulation and with minimal systemic side effects. For example, in postoperative patients who are bedridden for a long time, the blood flow in the calves is slow. The system can release a small dose of NO every few minutes to simulate the vasodilation effect during muscle activity, thereby keeping the veins unobstructed; when the patient moves briefly and the system detects an improvement in blood flow, the release frequency is reduced to avoid overdose. Therefore, under closed-loop control, the NO release module achieves precise treatment of "drug delivery on demand" at the target site, dynamically adjusts the output according to the DVT risk, and makes the prevention process efficient and individualized.

[0069] The power supply module is used to provide a stable working power supply for the above-mentioned sensor module, control module, and NO release module. Since this system is in the form of a wearable patch, a small rechargeable battery is generally used for power supply, supplemented by a power management circuit to meet the voltage and current requirements of each part. The design focus of the power supply module lies in energy density, safety, and the stability of power supply to each load.

[0070] In this embodiment, the power module uses a flexible thin-film lithium-ion battery as the main power source. For example, a lithium polymer battery with a capacity of several hundred milliamperes per hour can be selected, with a thickness of less than 1 mm, which can be bent and attached to the patch substrate. The nominal voltage provided by this battery is about 3.7V, and different voltage levels required by each module are converted through a power management circuit. The control module usually requires a digital circuit voltage of about 3.3V. Some sensors (such as operational amplifiers and bridges) may require ±5V or other bias voltages, while the NO release module may require a relatively high driving voltage or current pulse (such as above 5V or a current level of dozens of milliamperes) during electrolysis. Therefore, the power module integrates a DC-DC step-up / step-down converter and a low dropout regulator (LDO). For example, a boost converter is set to boost the battery voltage to 5V to supply the NO release electrode drive circuit, and another buck regulator outputs 3.3V to supply the microcontroller and various digital sensors. Some low-power sensors (such as accelerometers) can be directly powered by the battery voltage, but for the sake of simplifying the design, they are usually uniformly powered by the regulated output to reduce noise.

[0071] The power module is equipped with a charging and protection circuit. The lithium battery is charged with voltage and current limited through a charging management chip (it can be powered by a USB interface or connected to an inductive wireless charging coil to achieve wireless charging of the patch when the patient is inconvenient to disassemble). The protection circuit (such as a protection IC against overcharging and over-discharging) ensures that the battery voltage does not exceed the safe range and avoids over-discharging to damage the battery. In addition, the power module includes a power monitoring function. The control module can read the battery voltage through an analog-to-digital converter or obtain the remaining battery percentage through a dedicated fuel gauge chip. This allows the system to give an indication (such as an LED flashing to indicate that charging is required) or take energy-saving measures when the power is low. For long-term prevention (such as postoperative patients who need to be continuously guarded for several days), the capacity of the power module should be designed to support the operation of the device for at least one predetermined period (such as 24 hours), and then it can be replaced / charged when the patient is resting or being cared for without affecting continuous monitoring.

[0072] The power module is usually placed in the edge area of the patch to disperse the weight and avoid affecting the softness of the sensing area. For example, a flat battery is placed at the upper end of the patch (in the thigh direction), the control module circuit is in the middle, and the NO release and sensing units are at the lower end. In this way, the weight is concentrated at the proximal end of the body, the patient feels comfortable, and no pressure is exerted on the measurement point. All power connections are distributed to each module through a flexible printed circuit and are insulated in multiple layers to ensure the safety of the user. The power module itself is also sealed and waterproofed to prevent erosion by human sweat.

[0073] The energy consumption of each module in this system has been optimized: Most of the sensor modules are passive components (such as impedance, strain gauges) or low-power devices (the working current of the accelerometer is in the microampere range). The control module performs simple operations most of the time and only briefly increases the computing power consumption when solving for optimization. Although the NO release module consumes a relatively large current during operation, its operation is intermittent and the duty cycle is controllable. Therefore, the average power consumption can be made very low. For example, sensor acquisition and control calculations are performed once per second. Most of the time, the microcontroller is in the energy-saving mode and is only active during the sampling instant and control calculations (a few milliseconds). The annualized current may only be a few milliamperes. Assuming that the NO release only pulses for a few seconds per minute, even if its current reaches dozens of milliamperes, the average value converted is only a few milliamperes. Therefore, a battery with a capacity of 300 mAh can theoretically supply power continuously for several days. In practice, to ensure reliability, the device will be charged once a day or every 48 hours.

[0074] The power supply module involves energy storage and special attention should be paid to safety. The lithium battery used should comply with medical-grade specifications and have over-temperature protection. The patch material plays a certain role in flame retardancy and heat insulation for the battery, preventing damage to the patient in the event of a battery failure in extreme cases. In the circuit design, isolation and current limiting should be done well for each voltage rail. For example, the NO release electrode drive circuit is isolated from the sensor circuit to avoid interference with the weak sensing signal when the electrode drives a large current. At the same time, once the NO module is short-circuited or overloaded, the power supply should limit the current to protect the circuit and the battery. The control module can also monitor the power supply status. Once abnormal power consumption or a sudden voltage drop is detected, it can automatically turn off non-critical modules (such as pausing NO release) and issue an alarm to remind the user to check the device.

[0075] In summary, the power supply module provides reliable energy support for the system, enabling the entire patch system to work independently in a clinical environment without relying on bulky external devices. This greatly facilitates the implementation of DVT prevention: The patient only needs to apply the device without connecting wires or tubes and can move freely or rest in bed. At the same time, thanks to reasonable power consumption management and charging design, the device can operate continuously for a long time, ensuring continuous monitoring and intervention of DVT risks without interruption.

[0076] Each functional module is assembled into a complete system through circuit connections. Its signal flow and control flow are as Figure 1 and Figure 2 shown. The following details the closed-loop control workflow of this system in combination with the signal transmission path between modules: (a) Module connection architecture: Figure 1In the overall system structure block diagram shown, the sensor module, control module, NO release module, and power module are interconnected to form a closed loop. Among them, the power module provides the required electrical energy for all other modules and is connected to the sensor module (to power the active sensor), control module (to power the core processing unit), and NO release module (to power the drive unit) through power lines. The sensor module and the control module are connected by multiple signal lines, and the collected analog signals are transmitted to the ADC input port of the control module; the control module and the NO release module are connected by control signal lines (such as PWM signal lines, electrode drive lines, etc.) to send control instructions and drive the NO release device to act. In addition, if the control module communicates with external devices (not necessary), it can also be connected through a wireless module. However, within the closed-loop system, the main signal path circulates among the above three functional modules. The entire connection is encapsulated inside the patch using flexible printed circuit board and micro-interface technology to ensure reliable connection while maintaining the softness and comfort of the patch.

[0077] (b) Signal acquisition and processing flow: When the system starts working (for example, when the patient applies the device and turns on the power), the closed-loop process is initialized and enters a loop. First, the sensor module acquires signals: the optical sensor detects the pulse wave, the impedance electrode applies an excitation to measure the impedance value, the strain gauge and thermometer read the values, the accelerometer senses the movement, etc. These raw signals are provided in the form of analog voltage, resistance change, or digital readings. Subsequently, these signals enter the input interface of the control module along the connection lines. The control module polls or triggers an interrupt to read the signals of each sensor according to a preset sampling period (for example, several times per second). The analog signals are converted through analog-to-digital conversion and are converted into digital quantities, and the discrete sensing data is stored in the memory of the control module.

[0078] Then, the control module executes signal processing and feature extraction algorithms as described: filter and correct the raw data, extract feature parameters such as blood flow velocity v and blood flow resistance R, and fuse and calculate the thrombus load index x 1 . This process is equivalent to condensing a large amount of low-level signals into highly summarized state information x 1 , as well as auxiliary quantities such as the current motion state. Figure 3 The flow diagram of this series of signal processing steps is given: starting from the sensor signal, passing through the filter module and feature calculation module, and finally outputting the risk index x 1 .

[0079] (c) Control decision-making and execution: When the control module obtains x at the current moment 1After reaching the relevant state, it enters the closed-loop control decision-making stage. The MPC algorithm built into the control module uses the latest state, combines with the state-space model, predicts the development trend of future risk indicators, and calculates the NO release control quantity u required for this control cycle. As described above, in Section 2, the control module has derived the optimal u∗(k) based on x 1 and so on. At this time, the control module sends this control instruction to the NO release module. In terms of hardware implementation, the control module may directly represent the magnitude of u by adjusting the voltage / current of one of its output pins. For example, if the NO release module is of the electrochemical type, the control module's DAC outputs an analog voltage corresponding to u to the electrode drive circuit, and the drive current flows through the electrode according to this voltage setting value, thereby releasing the corresponding dose of NO; if it is of the micropump type, the control module outputs a PWM signal to control the opening duration or frequency of the pump, which is proportional to the required u. After risk assessment, the control module transfers the calculated control quantity to the NO release module, and this action closes the loop from sensing to execution.

[0080] (d) Physiological response and feedback: After receiving the control instruction, the NO release module immediately performs the corresponding NO release operation (such as releasing a certain volume of NO gas into the tissue). NO begins to exert its pharmacological effect locally: the vascular smooth muscle relaxes, the venous diameter expands, and the local hemodynamic parameters begin to change. This change is detected again by the sensor: as the blood flow accelerates, both the optoelectronic and impedance signals of the sensor module will reflect phenomena such as increased flow rate and increased volume (for example, a smaller increase in impedance indicates that more blood remains in the blood vessels and is quickly carried away; the amplitude and rising steepness of the PPG pulse waveform may increase, indicating increased blood perfusion); if there was mild edema caused by congestion before, the blood flow improvement may now gradually subside, and the size detected by the strain gauge may slightly decrease; at the same time, the accelerometer may still not detect movement, but the chemical intervention alone has partially compensated for the impact of insufficient movement. All these changes cause the risk indicator x 1 to start to decline. In this way, in the next sampling cycle, the new data fed back by the sensor module to the control module will reflect the state after the action of NO - usually a trend of reduced risk. The control module will compare the updated x 1 (k + 1) with the previous prediction, and the MPC algorithm will correct the model error and consider the achieved effects in subsequent optimizations.

[0081] The above completes a cycle of the closed loop. Starting from the next cycle, the process returns to step (b), sensor->control->execution->feedback, and iterates continuously. Typically, the control module executes a closed loop at a fixed frequency, such as executing a complete decision every 1 second or every few tens of seconds (the specific frequency depends on the system requirements: DVT formation is a relatively slow process, and a control cycle of several seconds is sufficient; but in order to capture transient changes in blood flow and provide smooth control, a rate of about 1Hz may be selected). Through small adjustments in high frequency, the system can gradually maintain the patient's venous status within a safe range.

[0082] (e) DVT optimization characteristics of closed-loop strategy: This modular closed-loop system is optimized for deep vein thrombosis prevention. All links work together to achieve efficient prevention and control based on the pathological characteristics of DVT: First, the multi-point monitoring of the sensor module meets the factors related to thrombosis in Virchow's three elements: blood stasis (reflected by blood flow velocity v and resistance R), vascular wall status (NO improves endothelial function, which is not directly measured but inferred through the effect), and hypercoagulable state (cannot be directly sensed, but can be indirectly inferred through long-term immobility and slow blood flow). In particular, long-term immobility is a common factor inducing DVT. The system closely monitors through the accelerometer and incorporates this factor into the x 1 Calculation, so that even if the immobility itself has not yet caused the blood flow to slow down, it will increase x in advance 1 , prompting the system to take action in advance. This embodies a feedforward control idea, which, combined with the feedback MPC, enables the system to respond to risks more promptly.

[0083] Secondly, the MPC strategy of the control module enables the system to proactively combat risk accumulation. Unlike traditional feedback control, MPC uses models to predict the future, even if the current x 1 If the patient is still safe, but if it is predicted to rise after a period of time, the system will increase NO release in advance to suppress this rising trend. This is especially important for a gradual process such as thrombosis: early minor interventions are often more effective and safer than large-dose treatments afterwards. For example, if a patient has just entered the anesthesia recovery room and has not yet had a thrombosis, but the sensor shows that his limbs are immobile and blood flow is gradually slowing down, MPC will 1 When it starts to rise, it gradually increases u, so that x may never 1 It can be said that MPC gives the system the ability to "prepare for a rainy day".

[0084] Furthermore, the closed-loop control ensures individualized adjustment. Since the baseline blood flow and vascular conditions vary among patients, a unified open-loop drug administration plan may result in overdose for some and underdose for others. However, this system automatically adjusts through feedback. For those with slower blood flow and higher resistance, the device releases more NO on their patches; conversely, for patients at low risk, the device rarely takes action. This adaptability not only improves the therapeutic effect but also reduces unnecessary interventions, reflecting the trend of personalized medicine.

[0085] Finally, the entire closed-loop system operates without manual intervention, reducing the medical burden and monitoring costs. Medical staff only need to set the parameters when the device is initially placed, and then the system can continuously monitor and fine-tune the treatment 24 / 7. In the prevention of DVT, time and persistence are crucial: missing a few hours of protection may be sufficient to form a blood clot. With the closed-loop system, the protection can be carried out seamlessly. For example, when there is no one to observe the patient at night, the system is still automatically sensing the blood flow in his legs and releasing NO appropriately. In the morning, medical staff can also check the night situation from the records. Compared with devices such as intermittent pneumatic pumps that can only act mechanically periodically and cannot be used when the patient walks, the patch-type NO closed-loop system can work continuously in any posture, filling the gap of other prevention methods.

[0086] In summary, through the signal flow and control process as described above, the various modules of the present invention constitute a closed-loop control system for the prevention of deep vein thrombosis. The sensor module obtains rich information related to DVT, the control module makes decisions using advanced model prediction algorithms, the NO release module precisely executes interventions, and the power module provides lasting power. The four work in a closed-loop and cooperate with each other. Figure 1 The block diagram shown intuitively reflects this collaborative relationship: the sensor module sends the monitoring signal to the control module, the control module analyzes it and outputs a control signal to the NO release module, the NO release module acts on the patient's venous system and forms a feedback through the sensor, and so on in a cycle. Functionally, this closed-loop system maintains the hemodynamic stability and anticoagulant state of the patient's lower limb veins in real time, significantly reducing the incidence of venous thrombosis. Its technical solution integrates biosensing, intelligent control, and targeted drug delivery, providing a brand-new prevention idea in the clinical scenario of deep vein thrombosis: replacing part of the manual care with an intelligent patch to achieve active, continuous, and personalized DVT prevention and control, and escorting the safety of patients.

Claims

1. A patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombosis, comprising: A flexible patch substrate, characterized in that the patch substrate is integrated with: The sensor module includes a blood flow velocity sensor, a bioimpedance sensor, an acceleration sensor, and a temperature sensor, which are used to obtain the blood flow velocity v at the target vein, the parameters related to the vascular resistance R, the limb movement state, and the skin temperature of the patch area respectively; NO release module, which releases NO to act on the site of venous thrombosis; A control module, including a signal processing unit and a control unit, receives the signal output by the sensor module and runs a closed-loop control algorithm to generate a NO release control instruction; wherein the control module calculates a thrombus load index x1 according to the collected blood flow velocity v, blood flow resistance R, temperature and patient limb activity data; At the same time, adjusting the NO release amount of the NO release module according to the x1; The power module provides power to the sensor module, the control module and the NO release module.

2. A patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombosis according to claim 1, characterized in that: The x1 is calculated as follows: Among them, v0 and R0 are the baseline blood flow velocity and baseline blood flow resistance, respectively, α, β and γ are weight coefficients, and f (other factors) is used to represent the influence of auxiliary parameters detected by accelerometer and temperature sensor on thrombosis risk.

3. A patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombosis according to claim 1, characterized in that: The NO release module comprises: A drug reservoir for storing NO-releasing donor drugs; A surface acoustic wave driving unit, disposed near the drug reservoir, for generating surface acoustic waves under the drive of the control module to prompt the donor drug to release NO; a microneedle array, connected to the drug reservoir and extending out of the patch substrate to contact the skin, for delivering the released NO or NO-containing medium to a subcutaneous target area; Among them, the surface acoustic wave driving unit adopts a piezoelectric substrate and an interdigital transducer structure, the drug reservoir is in the form of a porous matrix or a hydrogel, and the microneedle array is a hollow needle structure, thereby achieving efficient NO output and transmission.

4. A patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombosis according to claim 1, characterized in that: The control module is configured to execute a model predictive control algorithm, including: internally storing a state space model describing thrombus evolution and NO action, and an objective function for evaluating control effects; The state space model defines the following dynamic equations: ； The x1 is the thrombus load index, x2 is the local NO concentration, u is the control input of NO release rate, and α, β, and γ are all positive parameters; After acquiring the signal from the sensor module, the control module predicts the change of the thrombus in the future time domain based on the state space model, and determines the optimal control output of the NO release module by solving the following optimization objective function: ； Among them, x1 ref is the target thrombus volume value, Q1 is the thrombus volume (or thrombus load) control weight, and R is the NO release penalty weight. Through the above optimization, the thrombus index tends to the predetermined target value and the NO dosage is controlled, thereby ensuring the thrombolytic or thrombosis prevention effect while inhibiting excessive NO release.

5. A patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombosis according to claim 4, characterized in that: The control module is also provided with two control modes: acute thrombus dissolution mode and chronic maintenance mode, and the thrombus volume x1 and the thrombus volume threshold x1 are used to determine the thrombus volume. th The comparison result is used to switch the mode, where: When x1>x th When , switch to the acute thrombus dissolution mode, increase the thrombus volume control weight Q1 in the objective function and increase the upper limit of NO release rate to increase the NO release dose for rapid thrombolysis; When x1≤x th When the thrombus volume is estimated, the control module switches to the chronic maintenance mode, reduces the thrombus volume control weight Q1 in the objective function and reduces the upper limit of the NO release rate, so as to release NO at a periodic low dose to maintain the stability of the thrombus and prevent recurrence; and the control module can judge and automatically switch between the above two modes in real time according to the thrombus volume estimation value fed back by the sensor module.

6. A patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombosis according to claim 1, characterized in that: The flexible patch substrate adopts a rigid-flexible structure, the control module and the power module are installed in the rigid area, and the sensor module and the NO release module are installed in the flexible area; The rigid area and the flexible area are connected by flexible interconnecting wires. The back of the patch substrate has a medical adhesive layer for fixing the system on the skin surface and ensuring reliable and unbroken electrical connection between modules during the patient's limb movement.

7. A patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombosis according to claim 1, characterized in that: The power module includes a micro rechargeable battery and a power management circuit. The power management circuit has overcharge and over-discharge protection functions and integrates a wireless charging receiving coil, allowing the system to be charged through an external wireless charging device; the system also includes a wireless communication module, which uses low-power Bluetooth wireless communication technology to send monitoring data and system status to an external terminal device, and receive remote control instructions or parameter configurations to achieve remote monitoring and adjustment of the closed-loop system.

8. A patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombosis according to any one of claims 1 to 7, characterized in that: The blood flow velocity sensor in the sensor module is used to detect the venous blood flow velocity v, and is calculated by the following formula: Wherein L is the fixed distance between the front and rear photoelectric probes on the patch, Δt is the transmission time difference of the pulse wave from the upstream sensor to the downstream sensor; the bioimpedance sensor is used to estimate the vascular patency and infer the vascular resistance R; the acceleration sensor is used to detect the target limb movement intensity and posture; the temperature sensor is used to collect the target skin temperature information; the control module is based on the above measurement data, fuses and calculates the thrombus load index x1, and determines the NO release rate u through the model predictive control algorithm, so that the thrombus load is controlled within a safe range.

9. A control method for a patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombosis, comprising the following steps: S1. The patch system is attached to the skin of the patient's thrombus or thrombosis-prone area, and the sensor module is activated to monitor the thrombus status; S2, the control module obtains the sensor data and estimates the thrombus volume x1 or thrombus load index, and compares it with the preset thrombus volume threshold x1. th Compare; S3, in the acute thrombosis mode, executing the model predictive control algorithm, and calculating the NO release amount in real time according to the state space model and the optimization objective function to quickly dissolve the thrombus; S4, in chronic maintenance mode, NO is released periodically at low doses to maintain the thrombus burden at a safe level; S5. The above monitoring and control process is repeated according to the real-time monitoring results until the thrombus load is reduced and stably maintained within the target range.

10. The control method of the patch-type thrombus monitoring and NO release closed-loop system for treating venous thrombosis according to claim 9, characterized in that: A safety constraint mechanism is introduced into the model predictive control algorithm, including: limiting the maximum release rate of NO to avoid overdose; real-time monitoring of the patient's vital signs and setting safety interlock conditions, and immediately reducing or stopping NO release to ensure patient safety when abnormal vital signs or system operation is detected; and automatically switching to a preset conservative fixed-dose administration mode when the control algorithm is abnormal or the sensor fails, to ensure that the closed-loop system complies with medical safety regulations during operation.

Images