Disclosure of Invention
The invention aims to provide a deep brain stimulation system with functions of real-time blood pressure monitoring, gesture sensing and intelligent reaction.
To achieve the above object, the present invention provides a deep brain stimulation system for treating orthostatic hypotension, comprising:
a data acquisition module for collecting physiological status information of a patient, comprising:
an implantable pressure sensor implanted in an artery for monitoring blood pressure data (BPt) in real time;
An electrocardiograph sensor integrated with the electronic wristwatch for acquiring electrocardiographic signals (ECGt) and heart rate data (HRt);
An attitude sensor implanted in a subcutaneous region of pectoral major muscles for detecting a posture state (Post) of a patient;
The data fusion module is implanted in the subcutaneous region of the pectoral large muscle and is used for receiving the multi-source physiological data output by the data acquisition module, integrating the multi-source physiological data and filtering noise to generate comprehensive physiological state information (St= { BPt,HRt,ECGt,Post }) of a patient;
a deep brain stimulation module comprising:
directional electrodes implanted in the patient's compartment Zhou Huizhi (PVG) and peri-aqueductal gray (PAG) areas for precise stimulation of the target brain region to enhance sympathetic nerve output;
The stimulation power supply component is implanted in the subcutaneous region of pectoral large muscle and is used for providing electric signals with adjustable voltage (Vt), frequency (ft) and pulse width (PWt) for the directional electrode, supporting segmentation and directional stimulation and avoiding interference to non-target regions;
The response module is integrated with the data fusion module in the subcutaneous region of pectoral large muscle and is used for analyzing physiological state information St, dynamically adjusting the voltage, frequency and pulse width of the deep brain stimulation module and optimizing and adjusting strategies in real time according to physiological feedback data of patients;
The risk prediction module is integrated with the data fusion module in the subcutaneous region of pectoral muscle and is used for processing the comprehensive physiological state history data { St−n,...,St }, predicting the future physiological state St+1 of the patient and transmitting the predicted value to the response module;
The reinforcement learning module is integrated with the data fusion module in the subcutaneous region of the pectoral large muscle and is used for dynamically updating the stimulation parameter adjustment strategy of the response module according to the physiological feedback data of the patient through a reinforcement learning algorithm.
Further, the data fusion module generates real-time comprehensive physiological state information (St= { BPt,HRt,ECGt,Post }) of the patient through synchronous processing, noise filtering, feature extraction and weighted fusion of information such as blood pressure data, heart rate data, electrocardiosignal, posture state and the like.
Furthermore, the directional electrode is composed of a plurality of independent electrode plates, each electrode plate can independently control the on-off state, and meanwhile, each electrode plate is accurately distributed in a chamber Zhou Huizhi (PVG) and a gray matter (PAG) area around a water guide pipe according to an anatomical structure and a treatment target, wherein the stimulation power supply assembly is used for avoiding the influence on a non-target area by independently controlling the on-off state of each electrode plate, improving the pertinence and the safety of stimulation and realizing the accurate stimulation on a specific target brain area (such as PVG or PAG).
Further, the stimulating power supply assembly consists of a single power supply and a parallel switch matrix;
The device comprises a single power supply, a voltage regulator, a pulse generator and a power control unit, wherein the single power supply is implanted in the subcutaneous region of pectoral large muscle of a patient, and is used for providing adjustable stimulation parameters for a directional electrode, including voltage (Vt), frequency (ft) and pulse width (PWt), the voltage regulator is used for meeting the requirements of different target brain regions on the stimulation intensity by adjusting the amplitude of output voltage (Vt), the pulse generator is used for generating the required stimulation frequency (ft) by controlling the repetition rate of pulses, the adjustment of the frequency (ft) influences the rhythm of a stimulation signal and the discharge response of target neurons, and meanwhile, the pulse generator is used for controlling pulse width (PWt) by adjusting the duration of single pulses, and the size of the pulse width (PWt) determines the coverage depth and the duration effect of each electrical stimulation signal;
The parallel switch matrix is positioned between a single power supply and the directional electrode and comprises a plurality of independent switches and a switch control logic circuit, wherein each switch is used for controlling the on-off state of one or a group of electrode slices, and the switch control logic circuit receives a switch instruction from the response module and decides which switches are on or off, so that the aim of configuring specific electrode slice activation combinations for segmented stimulation (activating specific areas) or directional stimulation (concentrating current to specific paths) is fulfilled.
Further, the response module sends the generated stimulation parameter instructions to the voltage regulator and the pulse generator of the single power supply, and simultaneously sends electrode activation instructions to the switch control logic circuit to designate electrode pieces to be activated.
Further, a rule table is arranged in the response module, the rule table is based on the physiological state and the treatment target of the patient, and the activation strategy and the stimulation parameters under different states are defined by combining clinical data or personalized diagnosis results, wherein the control mode of the response module comprises the following steps:
s1, receiving current comprehensive physiological state data (St= { BPt,HRt,ECGt,Post }) from a data fusion module;
s2, judging whether the patient needs to adjust stimulation and the positions and the number of electrode plates to be activated according to the real-time data;
S3, matching the current physiological state with a preset rule table, if the current physiological state is not completely matched, selecting a most recently matched rule or using a default safety parameter, and determining a required activation area (PVG area, PAG area or both), the number of activations (the number and distribution of activated electrode slices), a stimulation parameter (voltage (Vt), frequency (ft) and pulse width (PWt));
s4, firstly, sending a stimulation parameter instruction to a single power supply, and generating a target electric stimulation signal by an internal component (a voltage regulator and a pulse generator) of the single power supply;
S5, selectively activating the directional electrode plates to realize segmentation and directional stimulation, and continuously monitoring physiological feedback (such as blood pressure BPt, heart rate HRt and electrocardiogram ECGt) of a patient through a data acquisition module;
And S6, if the feedback parameters do not reach the expected targets, re-matching the rule table, preferentially adjusting the stimulation parameters (such as voltage), if the stimulation parameters are still invalid, gradually increasing the number of activated electrode slices or adjusting the positions of the electrode slices until the physiological state of the patient is restored to the safe range.
Further, after the risk prediction module outputs the predicted physiological state St+1={BPt+1,HRt+1,ECGt+1,Post+1 at the next moment to the response module through the time sequence model, the decision logic of the response module is as follows:
Rule table matching:
the current state St is matched with a rule table, and an instant stimulation scheme (comprising electrode plate activation and stimulation parameters) is obtained;
if the predicted state St+1 is beyond the safety range, matching the rule corresponding to St+1 in the rule table;
Processing that the rule table cannot be matched:
If the current state St or the future state St+1 cannot match the rule table;
The response module automatically generates stimulation parameters, and the newly generated parameters and the corresponding states are recorded and used for dynamically updating the rule table;
decision output:
Considering the matching results of St and St+1 together, a final stimulation instruction is generated, including the activated electrode pad combination (PVG region, PAG region, or both), stimulation parameter voltage (Vt), frequency (ft), and pulse width (PWt).
Further, the step of automatically generating the stimulation parameters by the response module is as follows:
s1, acquiring a current state St through a data fusion module, acquiring a future state St+1 through a risk prediction module, and acquiring a blood pressure target value and a blood pressure deviation (delta BP) through system presetting, wherein the blood pressure target value (BP Target object) is defined as a systolic pressure target value=baseline systolic pressure-20 mmHg, and a diastolic pressure target value=baseline diastolic pressure-10 mmHg;
ΔBP=BP Target object−BPt
BPt is the current blood pressure monitored in real time;
Bias classification:
ΔBP >0, hypotension, requires an increase in stimulation intensity.
ΔBP is less than or equal to 0, and the blood pressure is normal or higher without increasing stimulation.
S2, generating a stimulation parameter according to the following rule based on the size of the delta BP:
The intensity of the voltage control stimulation signal directly determines the amplitude of the stimulation current, and the higher voltage can activate the target nerve region more strongly but possibly cause unnecessary side effects (such as overstimulation), the adjustment logic of the voltage control stimulation signal is used for calculating the current blood pressure deviation delta BP, the voltage is adjusted by Vt = V initial +kv x delta BP, the stimulation intensity is increased or reduced, wherein kv represents the voltage value required to be increased per unit blood pressure deviation, and the voltage is ensured not to exceed a safety range [ Vmin,Vmax ];
The frequency determines the rhythm of the stimulation signal, influences the frequency and efficiency of nerve excitation, and the higher frequency can quickly accumulate the stimulation effect, but possibly lead to nerve adaptation and reduce the curative effect, the adjusting logic of the frequency determines the current blood pressure deviation delta BP, the frequency is adjusted through ft=f Initial initiation+kff multiplied by delta BP, the stimulation rhythm is controlled, wherein kf represents the frequency which needs to be increased per unit blood pressure deviation, and meanwhile, the frequency needs to be ensured not to exceed the tolerable range [ fmin,fmax ] of the nerve;
The pulse width is defined as the duration of each stimulus, the accumulated intensity of the stimulus is influenced, the pulse width is prolonged, the stimulus effect of each pulse can be increased, but the power consumption can be increased, the adjustment logic is used for calculating the current blood pressure deviation delta BP, the pulse width is adjusted through PWt=PW Initial initiation+kp multiplied by delta BP, the single stimulus effect is prolonged, wherein kP represents the pulse width required to be increased per unit blood pressure deviation, and the frequency is ensured not to exceed the tolerable range of nerves [ PWmin,PWmax ];
Activating more electrode plates can expand the stimulation range and improve the treatment effect, and reasonably selecting the activated electrode plate combination according to the anatomy and the target area (such as PVG and PAG), wherein the adjustment logic is used for calculating the current blood pressure deviation delta BP; by a linear relationship n=n Initial initiation+(kn ×Δbp), increasing the stimulation coverage, wherein kn represents the number of electrode pads that need to be increased per unit blood pressure deviation, while also ensuring that the number of active electrode pads does not exceed the maximum supported number of devices Nmax;
S3, the response module generates stimulation parameters including an activated electrode plate combination (the number N and a target area (PVG, PAG or both)), a stimulation parameter voltage (Vt), a frequency (ft) and a pulse width (PWt) according to the adjustment rule.
Further, when the rule table is incompletely matched or the prediction deviation of the risk prediction module is larger, the reinforcement learning module optimizes the stimulation parameter adjustment strategy through a feedback mechanism, and the reinforcement learning module works as follows:
s1, receiving a current state St from a data fusion module, acquiring a predicted state St+1 from a risk prediction module, and initializing a current Q (St,At) value;
S2, selecting an action based on a current Q-value function, adopting an epsilon-greedy strategy, selecting random actions (exploration) by epsilon-probability, and selecting a current optimal action (utilization) by 1-epsilon-probability, wherein the actions comprise voltage adjustment, frequency adjustment, pulse width adjustment and electrode plate activation;
S3, activating a single power supply and a parallel switch matrix through a response module according to the selected action At, and monitoring the feedback state St' and the rewards Rt after stimulation in real time:
bonus function:
s4, updating a Q-value by using a Q-Learning algorithm:
wherein, alpha learning rate, control step length updated each time, gamma is discount factor, measure the importance of future rewards;
S5, if the Q (St,At) value of a certain state-action combination is stable and good in performance for a long time, the Q value is updated into a rule table as a new rule.
The deep brain stimulation system for treating the postural hypotension provided by the invention has the following remarkable beneficial effects by combining a deep brain stimulation technology and a modern intelligent algorithm:
1. Accurate stimulation target area, improving treatment effect
The directional electrode can specifically stimulate a gray matter (PAG) area around a chamber Zhou Huizhi (PVG) and a water guide pipe through the accurate control of the independent electrode slice, obviously enhance the output of sympathetic nerves, improve the blood pressure regulating effect, avoid the interference to non-target areas and reduce side effects.
And the activation state of the electrode plates is dynamically controlled through the parallel switch matrix to support the segmented stimulation of a specific area and the directional concentration of current paths, so that the stimulation accuracy and the treatment effect are further improved.
2. Real-time monitoring and dynamic adjustment, ensuring safety
And the data acquisition module monitors the blood pressure, heart rate, electrocardiosignal and posture state of the patient in real time and ensures that the physiological information of the patient is comprehensively mastered.
And the data fusion module is used for carrying out fusion and noise filtration on the multi-source physiological data to generate real-time comprehensive physiological state information St and ensure the accuracy and reliability of the data.
And the response module dynamically adjusts the stimulation parameters (such as voltage, frequency and pulse width) according to the real-time physiological state St and the predicted state St+1, ensures that the stimulation effect reaches the target range, and simultaneously keeps the physiological state of the patient within the safe range.
3. Intelligent control, improved adaptability and self-learning ability
Rule table matching, namely, a rule table is built in a response module, and proper stimulation parameters and electrode slice activation schemes are quickly matched according to the individual physiological states of patients, so that the requirement of manual intervention is reduced.
And the risk prediction module is used for realizing prospective control on the adjustment of the stimulation parameters by predicting the future physiological state St+1 so as to avoid possible blood pressure fluctuation in advance.
Reinforcement learning module:
And under the condition that the rule table is incompletely matched or the time sequence prediction deviation is large, dynamically optimizing the stimulation strategy through a feedback mechanism.
The rule table can be gradually updated and supplemented through the reinforcement learning algorithm, so that the continuous optimization and self-adaption capability of the system are realized, and the method particularly shows strong advantages under complex or emergency conditions.
4. Personalized treatment, adapting to different patients
The system can adjust the rule table and the initial values of the stimulation parameters according to the individual characteristics (such as age, body type, medical history and the like) of the patient, so as to realize personalized treatment.
The risk prediction module can optimize the prediction model according to the long-term monitoring data of the patient, and the reinforcement learning module can learn the optimal treatment scheme in long-term feedback.
5. The system design is efficient and energy-saving, and the running stability is improved
Single power supply and parallel switch matrix:
the adjustable voltage, frequency and pulse width are provided by a single power supply, so that the hardware complexity is reduced, and the multi-channel output is supported to meet the requirements of electrode slice segmentation and directional stimulation.
The parallel switch matrix dynamically activates the target electrode plate through the high-efficiency switch logic, so that the energy consumption is reduced and the service life of the equipment is prolonged.
6. Reducing side effects and improving comfort of patients
Accurate control, namely, by optimizing the combination of the stimulation parameters and the activated electrode plates, the over stimulation or unnecessary energy waste is avoided, and the possible side effects are reduced.
Dynamic monitoring and adjustment, wherein the system can quickly respond to the physiological state change of a patient, avoid dizziness, hypodynamia or faint caused by hypotension and improve the daily life quality of the patient.
7. Data-driven long-term efficacy optimization
And (3) recording and analyzing the physiological data, namely recording the real-time physiological data and feedback results of the patient by the system, and providing a long-term treatment effect evaluation basis for doctors.
And the rule table is dynamically updated, namely the risk prediction module and the reinforcement learning module work together, and the rule table is continuously optimized through long-term feedback data, so that the system can adapt to the long-term treatment requirement of a patient.
8. Innovative and clinical application potential
The system integrates the modern biological sensing technology, the intelligent control algorithm (rule table matching, time sequence prediction and reinforcement learning) and the efficient power management technology, and provides an innovative comprehensive solution for the treatment of the postural hypotension. By means of accurate, real-time and personalized treatment modes, the system is most likely to become standard equipment for treating the postural hypotension in the future, and a safe and effective long-term treatment option is provided for patients.
The invention combines the deep brain stimulation technology with the intelligent algorithm through the multi-module synergistic effect, and solves the problems of poor effect and large side effect in the prior art for treating the postural hypotension. The system has obvious advantages in the aspects of accuracy, instantaneity, safety and personalized adaptation, and has extremely high clinical application value and popularization prospect.
Detailed Description
The invention aims to provide a deep brain stimulation system for treating postural hypotension, which is provided with real-time blood pressure monitoring, posture sensing and intelligent response mechanisms.
The deep brain stimulation system for treating the postural hypotension comprises a data acquisition module, a data fusion module, a deep brain stimulation module, a response module, a risk prediction module and a reinforcement learning module, wherein the data acquisition module is used for collecting physiological state information of a patient and comprises an implanted micro pressure sensor implanted into an upper limb artery of the patient, the implanted micro pressure sensor senses blood flow pressure change through a capacitance diaphragm structure, a pressure value is output in real time through an LC resonance circuit integrated on an electronic watch, and the deep brain stimulation system has the working characteristic of high sensitivity and monitors blood pressure data BPt in real time. Integrated into an electronic wristwatch, an electrode type single-lead electrocardiograph sensor supporting QRS wave detection is adopted to collect electrocardiogram signals (ECGt) and heart rate data (HRt). The device is implanted in the subcutaneous region of pectoral large muscle, and is internally provided with a 6-axis IMU (inertial measurement unit), and an attitude sensor integrating an accelerometer and a gyroscope is used for detecting the posture state (Post) of a patient.
In such a closed-loop regulation system, BPt is the "main target variable" that directly reflects the severity of OH, which is the system control, HRt is the auxiliary assessment of autonomic balance and cardiac compensation, ECGt is the provision of more comprehensive cardiac state information for safety monitoring and stimulation strategy fine tuning, post is the accurate detection of posture changes, which is the key signal for OH event triggering, and can also be used as a prospective reference for risk prediction.
Wherein, in order to facilitate understanding of the relationship of the three dependent variables of BPt、HRt、ECGt, the detailed role of BPt、HRt、ECGt in the overall system is explained in detail herein;
BPt (blood pressure data) is the most direct indicator for determining whether a patient has postural hypotension (OH), and is also the main reference for adjusting stimulation strategies (especially voltage, frequency, pulse width).
Usage scenario in the system:
Whether or not priming is required is determined by the system not performing or performing only the lowest intensity stimulation when the BPt is within a safe range or slightly lower, and by the rules table or risk prediction module triggering an action of "boost stimulation" or "change electrode activation mode" when the BPt is below a certain threshold.
Real-time closed loop feedback, namely after the stimulation parameters are adjusted, the change of BPt is a key for judging whether the stimulation is effective or not, and if the blood pressure is not recovered, the stimulation is further increased or the electrode plate is adjusted and activated.
HRt (heart rate data) is often closely related to regulation of blood pressure and autonomic nerve status, and particularly, in OH attacks, heart rate compensation elevation (sympathetic excitation) may occur in some patients, and heart rate changes may occur in an asynchronous manner with blood pressure fluctuations, which may be indicative of the current autonomic nerve functional status of the patient.
Usage scenario in the system:
Auxiliary judgment of sympathetic/parasympathetic balance, if the blood pressure is reduced but the heart rate is not significantly increased, the autonomic nervous response is possibly prompted to be insufficient, and stronger DBS stimulation is needed to promote the sympathetic output;
And early warning of abnormal conditions, namely whether the heart rate is too fast, too slow or an ectopic heart rate appears, and prompting the DBS safety boundary.
ECGt (electrocardiogram) can reflect more of the electrophysiological characteristics of the heart, such as whether abnormalities such as extra-systole, atrial fibrillation, ventricular premature beat, etc., and cardiac arrhythmias may sometimes be complicated in patients with orthostatic hypotension.
Usage scenario in the system:
Safety monitoring, namely if the ECGt shows severe arrhythmia, the system is cautious when adjusting DBS parameters, so that the stimulation frequency or the excessive current is avoided to increase the heart burden;
Optimizing stimulation strategies PVG/PAG stimulation may, in some cases, affect the vagus nerve or other neural circuitry, potentially affecting heart rhythm, and ECGt may capture these changes in time and assist the system in making appropriate corrections.
The data fusion module is implanted in the subcutaneous region of the pectoral large muscle and is used for receiving the multi-source physiological data output by the data acquisition module, integrating the multi-source physiological data and filtering noise to generate comprehensive physiological state information of a patient (St={BPt,HRt,ECGt,Post);
in the two modules, the specific working steps are as follows
S1, firstly, recording acceleration and angular velocity data of a patient in real time by a 6-axis IMU sensor (inertial measurement unit) implanted by the system.
The body position change is determined by the following analysis steps:
and (3) static posture, namely detecting the current acceleration direction of the patient through an accelerometer, and distinguishing standing, lying and sitting postures.
Dynamic posture, namely recording angular velocity through a gyroscope, and judging posture transition (such as transition from lying to standing).
Attitude transition rate analysis:
calculating the acceleration value and time interval delta t of the body position change, and evaluating the change rate: If the rate v > v Threshold value (e.g., fast up), a possible hypotension triggering event is marked.
Combining the posture state (standing) and the speed (fast standing), the system marks the potential hypotension event preliminarily and enters a further data acquisition link.
S2, an implanted pressure sensor monitors blood pressure data in real time (BPt), an optical heart rate sensor collects heart rate data (HRt), an electrocardio sensor collects electrocardiogram signals (ECGt), and the comprehensive data form the current state:
The data fusion module integrates the attitude status and the real-time physiological data to generate a current comprehensive status St={BPt,HRt,ECGt,Post }
S3, the data fusion module receives { BPt,HRt,ECGt,Post }:
synchronous sampling, namely aligning time stamps of all signals to ensure the consistency of data.
Noise filtering, namely eliminating fluctuation noise of the pressure sensor through a Kalman filter.
And extracting features such as blood pressure change rate delta BP, heart rate change trend and the like, and enhancing state evaluation.
Weighting processing, namely setting weights according to the importance of the data, wherein the embodiment sets the weights of the blood pressure data to be higher than the posture state:
S4, simultaneously, carrying out hypotension risk assessment by a calculation module on the data fusion module;
Firstly judging whether the current blood pressure BP is lower than a blood pressure target value (BP target), wherein the BP target is defined as a systolic pressure target value=baseline systolic pressure-20 mmHg, a diastolic pressure target value=baseline diastolic pressure-10 mmHg, if the real-time systolic pressure of a patient is smaller than the baseline systolic pressure-20 mmHg, the system is marked as a hypotension risk, if the heart rate HRt is larger than 100bpm, the auxiliary judgment is a compensatory hypotension (compensated but insufficient), the posture state Post =standing is an important condition for triggering the risk, the condition belongs to low-intensity stimulation, if the patient is required, the system is marked as a hypotension risk, if the heart rate HRt is smaller than 100bpm, the auxiliary judgment is a non-compensatory hypotension (heart non-compensated), and the posture state Post =standing is an important condition for triggering the risk, and the condition belongs to high-intensity stimulation.
A deep brain stimulation module comprising:
directional electrodes implanted in the patient's compartment Zhou Huizhi (PVG) and peri-aqueductal gray (PAG) areas for precise stimulation of the target brain region to enhance sympathetic nerve output;
A stimulation power supply assembly implanted in the subcutaneous region of pectoral large muscle for providing adjustable voltage (Vt), frequency (ft) and pulse width (PWt) to the directional electrode, supporting segmentation and directional stimulation, avoiding interference with non-target regions;
The response module is integrated with the data fusion module in the subcutaneous region of pectoral large muscle and is used for analyzing physiological state information St, dynamically adjusting the voltage, frequency and pulse width of the deep brain stimulation module and optimizing and adjusting strategies in real time according to physiological feedback data of patients;
The risk prediction module is integrated with the data fusion module in the subcutaneous region of pectoral muscle and is used for processing comprehensive physiological state history data { St−n,...,St }, predicting the future physiological state St+1 of the patient, and transmitting the predicted value to the response module for prospective adjustment of deep brain stimulation parameters;
and the reinforcement learning module is integrated with the data fusion module in the subcutaneous region of the pectoral large muscle and is used for dynamically updating the stimulation parameter adjustment strategy of the response module according to the physiological feedback data of the patient through a reinforcement learning algorithm.
In one embodiment, the directional electrode is composed of a plurality of independent electrode plates, each electrode plate can independently control the on-off state, meanwhile, each electrode plate is accurately distributed in a chamber Zhou Huizhi (PVG) and a gray matter (PAG) area around a water guide pipe according to an anatomical structure and a treatment target, wherein a stimulation power supply component is used for avoiding the influence on a non-target area by independently controlling the on-off state of each electrode plate, improving the pertinence and the safety of stimulation and realizing the accurate stimulation on a specific target brain area (such as PVG or PAG).
The stimulation power supply assembly consists of a single power supply and a parallel switch matrix, wherein the single power supply is implanted into a subcutaneous region of pectoral muscle of a patient, a voltage regulator, a pulse generator and a power control unit are further arranged in the single power supply, the adjustable stimulation parameters comprising voltage (Vt), frequency (ft) and pulse width (PWt) are provided for a directional electrode, the voltage regulator is used for meeting the requirements of different target brain regions on stimulation intensity by adjusting the amplitude of output voltage (Vt), the pulse generator is used for generating required stimulation frequency (ft) by controlling the repetition rate of pulses, the adjustment of the frequency (ft) influences the rhythm of a stimulation signal and the discharge response of target neurons, the pulse generator is used for controlling the pulse width (PWt) by adjusting the duration of a single pulse, the coverage depth and the duration effect of each electrical stimulation signal are determined by the pulse width (PWt), and the power control unit is further arranged in the single power supply, and is used for realizing the power distribution and protection functions of the power control unit, monitoring the output power and preventing or short circuit and ensuring the stable operation of the system.
The parallel switch matrix is positioned between a single power supply and the directional electrode and comprises a plurality of independent switches and a switch control logic circuit, wherein each switch is used for controlling the on-off state of one or a group of electrode slices, and the switch control logic circuit receives a switch instruction from the response module and decides which switches are on or off, so that the aim of configuring specific electrode slice activation combinations for segmented stimulation (activating specific areas) or directional stimulation (concentrating current to specific paths) is fulfilled.
In one embodiment, the response module sends the generated stimulation parameter instructions to the voltage regulator and the pulse generator of the single power supply, and simultaneously sends electrode activation instructions to the switch control logic circuit to designate the electrode pads that need to be activated.
The response module is internally provided with a rule table which is based on the physiological state and the treatment target of the patient and combines clinical data or individual diagnosis results to define the activation strategy and the stimulation parameters under different states, wherein the establishment process of the rule table is shown in detail:
The initial rule table establishment process is usually performed by an "initialization" or "calibration" process before the system is formally used or implanted. The typical flow is as follows:
S1, data acquisition and baseline measurement
Collecting basic values of resting blood pressure BPt, heart rate HRt, electrocardiograph ECGt and body position Post (lying position, sitting position, standing position and the like) of a patient under different daily conditions;
The patient's blood pressure and heart rate fluctuations upon transition from lying to standing position were collected under "no activated DBS stimulation" conditions, and the patient's OH severity, rate of occurrence, and tolerance were assessed.
S2, body position transformation induction test
Under a controllable environment (such as a hospital monitoring condition), the patient is subjected to posture change (lying position, sitting position or lying position, standing position), and fluctuation of blood pressure BPt, heart rate HRt and electrocardio ECGt is measured in real time;
According to the amplitude of blood pressure drop and the threshold value of symptoms such as dizziness or syncope, the required stimulation voltage, frequency and pulse width range are preliminarily determined;
The minimum effective stimulation value (stimulation parameter that can initially combat blood pressure drop) and the maximum safety threshold (avoid overstimulation) are determined.
Segmented/directional electrode activation test
In the adjustable range, the electrode pads of PVG (chamber Zhou Huizhi) and PAG (ash around the water guide pipe) are activated sequentially/in combination, and which activation combination is most effective for blood pressure elevation and has the least side effect is observed;
based on the test results, the "most effective brain region stimulation combination" was recorded as the base rule.
S3, forming an initial rule table
Summarizing the data, and defining a plurality of rules (Rule) by combining experience and safety limit of a clinician, wherein each Rule corresponds to a mapping relation of 'current state/predicted state to target stimulation scheme';
For example, if the blood pressure is too low (BPt < X1) and the heart rate is high (HRt > Y1), the low voltage + high frequency + PVG region electrodes are activated, if the blood pressure is very low (BPt < X2), pulse widths can be superimposed, more electrode pads can be activated, etc. These rules support most of the decisions of the system during the startup phase.
The control mode of the response module comprises the following steps:
s1, receiving current comprehensive physiological state data from a data fusion module (St={BPt,HRt,ECGt,Post);
s2, judging whether the patient needs to adjust stimulation and the positions and the number of electrode plates to be activated according to the real-time data;
S3, matching the current physiological state with a preset rule table, if the current physiological state is not completely matched, selecting a most recently matched rule or using a default safety parameter, and determining a required activation area (PVG area, PAG area or both), the number of activations (the number and distribution of activated electrode slices), the stimulation parameters (voltage (Vt), frequency (ft) and pulse width (Pwt));
s4, firstly, sending a stimulation parameter instruction to a single power supply, and generating a target electric stimulation signal by an internal component (a voltage regulator and a pulse generator) of the single power supply;
S5, selectively activating the directional electrode plates to realize segmentation and directional stimulation, and continuously monitoring physiological feedback (such as blood pressure BPt, heart rate HRt and the like) of a patient through a data acquisition module;
And S6, if the feedback parameters do not reach the expected targets, re-matching the rule table, preferentially adjusting the stimulation parameters (such as voltage), if the stimulation parameters are still invalid, gradually increasing the number of activated electrode slices or adjusting the positions of the electrode slices until the physiological state of the patient is restored to the safe range.
In one embodiment, after the risk prediction module outputs the predicted physiological state St+1={BPt+1,HRt+1,ECGt+1,Post+1 at the next time to the response module through the time series model, the decision logic of the response module is as follows:
Rule table matching:
The current state St is matched with a rule table to acquire an instant stimulation scheme (comprising electrode plate activation and stimulation parameters);
If the predicted state St+1 exceeds the safety range, preferentially matching St+1 with the corresponding rule;
Processing that the rule table cannot be matched:
If the current state St or the future state St+1 cannot match the rule table;
The risk prediction module automatically generates stimulation parameters, and the newly generated parameters and the corresponding states are recorded and used for dynamically updating the rule table;
decision output:
Considering the matching results of St and St+1 together, a final stimulation instruction is generated, including the activated electrode pad combination (PVG region, PAG region, or both), stimulation parameter voltage (Vt), frequency (ft), and pulse width (PWt).
When St and St+1 are considered together, different weights are set, and the final parameter=α×the current state parameter+β×the predicted state parameter, where α and β are weight coefficients and can be adjusted according to the actual situation.
The new parameters and the corresponding states thereof generated automatically each time are recorded in a rule table for subsequent rapid matching, thereby improving the response speed and efficiency of the system.
Further, the step of automatically generating the stimulation parameters by the response module is as follows:
S1, acquiring a current state St through a data acquisition module, acquiring a future state St+1 through a risk prediction module, and simultaneously acquiring a blood pressure target value and a blood pressure deviation (delta BP) through system presetting, wherein the blood pressure target value (BP Target object) is defined as a systolic pressure target value = baseline systolic pressure-20 mmHg, and a diastolic pressure target value = baseline diastolic pressure-10 mmHg;
ΔBP=BP Target object−BPt
BPt is the current blood pressure monitored in real time.
Bias classification:
ΔBP >0, hypotension, requires an increase in stimulation intensity.
ΔBP is less than or equal to 0, and the blood pressure is normal or higher without increasing stimulation.
S2, generating a stimulation parameter according to the following rule based on the size of the delta BP:
The intensity of the voltage control stimulation signal directly determines the amplitude of the stimulation current, and the higher voltage can activate the target nerve region more strongly but possibly cause unnecessary side effects (such as overstimulation), the adjustment logic of the voltage control stimulation signal is used for calculating the current blood pressure deviation delta BP, the voltage is adjusted by Vt = V initial +kv x delta BP, the stimulation intensity is increased or reduced, wherein kv represents the voltage value required to be increased per unit blood pressure deviation, and the voltage is ensured not to exceed a safety range [ Vmin,Vmax ];
The frequency determines the rhythm of the stimulation signal, influences the frequency and efficiency of nerve excitation, and the higher frequency can quickly accumulate the stimulation effect, but possibly lead to nerve adaptation and reduce the curative effect, the adjusting logic of the frequency determines the current blood pressure deviation delta BP, the frequency is adjusted through ft=f Initial initiation+kff multiplied by delta BP, the stimulation rhythm is controlled, wherein kf represents the frequency which needs to be increased per unit blood pressure deviation, and meanwhile, the frequency needs to be ensured not to exceed the tolerable range [ fmin,fmax ] of the nerve;
The pulse width is defined as the duration of each stimulus, the accumulated intensity of the stimulus is influenced, the pulse width is prolonged, the stimulus effect of each pulse can be increased, but the power consumption can be increased, the adjustment logic is used for calculating the current blood pressure deviation delta BP, the pulse width is adjusted through PWt=PW Initial initiation+kp multiplied by delta BP, the single stimulus effect is prolonged, wherein kP represents the pulse width required to be increased per unit blood pressure deviation, and the frequency is ensured not to exceed the tolerable range of nerves [ PWmin,PWmax ];
Activating more electrode plates can expand the stimulation range and improve the treatment effect, and reasonably selecting the activated electrode plate combination according to the anatomy and the target area (such as PVG and PAG), wherein the adjustment logic is used for calculating the current blood pressure deviation delta BP; by a linear relationship n=n Initial initiation+(kn ×Δbp), increasing the stimulation coverage, wherein kn represents the number of electrode pads that need to be increased per unit blood pressure deviation, while also ensuring that the number of active electrode pads does not exceed the maximum supported number of devices Nmax;
S3, the response module generates stimulation parameters including an activated electrode plate combination (the number N and a target area (PVG, PAG or both)), a stimulation parameter voltage (Vt), a frequency (ft) and a pulse width (PWt) according to the adjustment rule.
In one embodiment, when the rule table is incompletely matched or the prediction deviation of the risk prediction module is large, the reinforcement learning module optimizes the stimulation parameter adjustment strategy through a feedback mechanism, and the reinforcement learning module works as follows:
s1, receiving a current state St from a data fusion module, acquiring a predicted state St+1 from a risk prediction module, and initializing a current Q (St,At) value;
S2, selecting an action based on a current Q-value function, adopting an epsilon-greedy strategy, selecting random actions (exploration) by epsilon-probability, and selecting a current optimal action (utilization) by 1-epsilon-probability, wherein the actions comprise voltage adjustment, frequency adjustment, pulse width adjustment and electrode plate activation;
S3, activating a single power supply and a parallel switch matrix through a response module according to the selected action At, and monitoring the feedback state St' and the rewards Rt after stimulation in real time:
bonus function:
s4, updating a Q-value by using a Q-Learning algorithm:
wherein, alpha learning rate, control step length updated each time, gamma is discount factor, measure the importance of future rewards;
S5, if the Q (St,At) value of a certain state-action combination is stable and good in performance for a long time, the Q value is updated into a rule table as a new rule.
The rule table is not invariable in the long-term operation process of the system, but can be dynamically updated along with the continuous feedback of the reinforcement learning module and the risk prediction module:
S1, reinforcement learning module feedback
When the system finds that the existing rule table cannot solve certain extreme or emergency situations well, a new stimulation strategy is tried, and if the new strategy achieves good effect (for example, blood pressure quickly rises and has no side effect), the new strategy is written into the rule table. For a long time, rule tables will be more rich and personalized.
S2, risk prediction deviation correction
If the risk prediction module (based on the time series model) deviates too much from the predicted value and the actual value of the future blood pressure, the system records the rule triggered at the time and the execution effect thereof, and is used for updating the prediction model and correcting or refining the corresponding rule.
4. Example structure of rule Table
For a more visual understanding, a simplified example rule table (for reference only, the actual system may be more complex) is given below:
The "suggested activated electrode" may correspond to a variety of electrode pad combinations, only roughly described in the table.
The circuits and control involved in the present invention are all of the prior art, and are not described in detail herein.
The foregoing description is only illustrative of the present invention and is not intended to limit the scope of the invention, and all equivalent structures or equivalent processes or direct or indirect application in other related technical fields are included in the scope of the present invention.