A neural system closed-loop regulation system based on transcranial photoacoustic cooperation

Abstract

The application provides a kind of nervous system closed loop control system based on transcranial photoacoustic cooperation, which is composed of EEG monitoring device, personal information, photoacoustic cooperation system, multi-modal large model and psychological disease database;For the transcranial infrared light and ultrasonic cooperative stimulation technology of patient, through optimizing stimulation parameters, realize the bidirectional regulation of photoacoustic, take multi-modal large model as the base, take multi-modal data fusion and reinforcement learning as the technical means, design closed loop photoacoustic cooperative stimulation system device and automatically optimize intervention parameters, realize the effective regulation method of psychological disease through non-invasive brain stimulation technology, realize the adaptive adjustment and closed loop feedback of parameters.Through real-time detection of patient's brain wave and collection of patient's personal information, evaluate the incoming system after integration, adjust intervention in real time according to different brain states, and use feedback to adjust the parameters of the next intervention and the precision regulation process of excitation.

Description

Technical Field

[0001] The present invention relates to the field of large language model medicine, and in particular to a closed-loop control system of the nervous system based on transcranial photoacoustic collaboration. Background Art

[0002] In recent years, light and sound stimulation have garnered increasing attention as a means of modulating neural activity and cognitive function. Transcranial light stimulation uses laser or LED light sources to penetrate the scalp and skull, inducing neurobiological changes in the brain. Compared to light of other frequencies, mid- and far-infrared light (3-15 μm) can modulate neural activity through a relatively specific, non-thermal mechanism. This mechanism involves the resonance of photon vibration frequencies with key chemical bonds in functional proteins, promoting or inhibiting neuronal action potential firing and glial cell differentiation, thereby modulating brain function or intervening in brain diseases. Transcranial ultrasound stimulation (TUS) uses low-intensity focused ultrasound, which has high penetrability and focus, allowing it to penetrate the skull and precisely target brain regions. It has been shown to elicit action potentials and promote synaptic vesicle release, and stimulating specific brain regions can trigger corresponding behaviors. Mental health issues are receiving increasing attention in China, particularly those related to depression, anxiety, and addiction. According to the "2023 China Mental Health and Well-being Blue Book," China's emphasis on mental health is increasing, investment is steadily increasing, and the market size is also expanding. National mental health issues are becoming more prominent and attention is increasing, but the supply of mental health services is relatively insufficient, there is an imbalance between supply and demand, service levels are uneven, and service quality needs to be improved.

[0003] Current treatment options include the following:

[0004] Transcranial magnetic stimulation: A non-invasive technology that uses strong magnetic fields to generate electric currents in the cerebral cortex to regulate neuronal activity, but it can only stimulate the surface of the brain and cannot effectively regulate deep structures.

[0005] Transcranial direct current stimulation (TDS): This technique alters the excitability of the brain's nerves by applying low-intensity direct current to the scalp. It is commonly used to improve cognitive function, memory, and mood regulation. However, its effectiveness is unstable, clinical research results are inconsistent, and the effects can vary significantly between patients.

[0006] Transcranial alternating current stimulation (TRAS): This procedure attempts to synchronize or modulate brainwave activity by applying alternating current to the brain. It aims to enhance or suppress specific brainwave frequencies. Some patients experience headaches, scalp irritation, or tingling.

[0007] Traditional open-loop stimulation, transcranial sound stimulation therapy, and transcranial light stimulation therapy have problems such as the inability to monitor and adjust interventions in real time, the inability to dynamically adjust the stimulation parameters of ultrasound and infrared light, the inability to perform transcranial light and sound synergistic stimulation, and the inability to identify and predict addictive behaviors by integrating multiple data types. Summary of the Invention

[0008] The present invention combines transcranial ultrasound stimulation, transcranial infrared light stimulation, electroencephalogram (EEG), large language model, RAG and other technologies, and mainly uses transcranial infrared light and ultrasound synergistic stimulation technology for patients with mental illnesses such as addictive diseases, anxiety disorders, depression, etc. By optimizing the stimulation parameters, photoacoustic bidirectional regulation is achieved to achieve the best regulation effect. With a multimodal large model as the basis, and multimodal data fusion and reinforcement learning as technical means, an efficient and real-time closed-loop photoacoustic synergistic stimulation system device is designed and intervention parameters are automatically optimized. Through non-invasive brain stimulation technology, an effective regulation method for mental illness is achieved, and adaptive adjustment of parameters and closed-loop feedback are achieved. By real-time detection of the patient's brain waves and collection of the patient's personal information, the multimodal large model is integrated into the system after evaluation, and the multi-dimensional information of the patient is used to diagnose and provide a treatment plan. After implementation, the patient's brain condition is monitored, and the intervention is adjusted in real time according to different brain states. Feedback is used to adjust the parameters of the next intervention and the precise regulation process of stimulation, i.e., closed-loop stimulation.

[0009] The present invention provides a closed-loop control system for the nervous system based on transcranial photoacoustic collaboration, which is composed of EEG monitoring equipment, personal information data, photoacoustic collaboration system, multimodal large model and mental disease database; wherein,

[0010] EEG monitoring equipment uses electrophysiological indicators to record brain activity, monitoring an individual's delta, theta, alpha, and beta waves in real time, and transmits the monitored data to a multimodal large model for analysis;

[0011] Personal information provided by users, including their medical history, imaging data, and psychological data, is passed to the multimodal large model for analysis;

[0012] The photoacoustic synergy system consists of two devices: an ultrasonic transducer and an infrared light source. The ultrasonic transducer emits ultrasonic pulses of varying frequencies to stimulate and treat patients with mental illnesses. The infrared light source emits infrared light of varying wavelengths to suppress background neural activity in the excitatory cortex, increasing the gain of the ultrasound stimulation effect. The photoacoustic synergy system is dynamically controlled by a large multimodal model, continuously adjusting intervention parameters to find the most suitable personalized treatment plan.

[0013] The multimodal large model is used to process and evaluate multidimensional data including brain activity data and personal information, formulate corresponding treatment plans based on individual conditions, make decisions and generate corresponding intervention parameters, and dynamically adjust the photoacoustic collaborative system in real time;

[0014] The mental illness database is a database that centrally stores knowledge about mental illnesses, making up for the fact that the multimodal large model does not have a deep enough understanding of certain mental illnesses. It also stores previous user data and simulates a reinforcement learning mechanism, allowing the large model to continuously learn and optimize, automatically guide execution, and continuously track feedback.

[0015] The present invention has the following beneficial technical effects:

[0016] This invention utilizes an ultrasonic transducer device, which converts electrical energy into acoustic energy (specifically, ultrasonic waves). In this application, the ultrasonic transducer generates ultrasonic pulses of a specific frequency, with adjustable frequency parameters, to achieve dynamic brain neuromodulation. This invention utilizes transcranial ultrasound stimulation (TUS), which offers advantages over other neuromodulation technologies, such as being non-invasive, providing a deep stimulation depth, and high spatial resolution.

[0017] The present invention utilizes an infrared light source device: a device that emits infrared light. In this application, the infrared light source device is used to generate infrared light of a specific wavelength, which can be used to suppress or excite. The wavelength parameters are adjustable to dynamically regulate background neural activity in the brain's cortex and increase the gain of ultrasound stimulation. This invention utilizes transcranial infrared light stimulation technology, which offers advantages such as non-invasiveness, bidirectional regulation, and deep penetration.

[0018] This invention implements a highly efficient closed-loop neural feedback control circuit from perception to decision-making to intervention, building a "perception-assessment-decision-intervention" platform. Continuous data feedback enables continuous learning and optimization of the large model, adjusting treatment plans in real time to ensure precision and personalization, while also automatically guiding execution and tracking feedback. Real-time monitoring of patient clinical manifestations aids understanding of the clinical markers and mechanisms of addictive disorders, thereby improving the safety and effectiveness of treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a system framework diagram of the present invention;

[0020] Figure 2 Schematic diagram of modulation mode optimization and neural regulation of photoacoustic synergy system;

[0021] Figure 3 Schematic diagram of neural regulation by photoacoustic synergy;

[0022] Figure 4 This is a functional diagram of the large model. DETAILED DESCRIPTION

[0023] In order to make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other. To achieve the above-mentioned objectives, the present invention adopts the following technical solutions.

[0024] The present invention provides a closed-loop control system framework for the nervous system based on transcranial photoacoustic synergy. Figure 1 As shown, it consists of EEG monitoring equipment, personal information data, photoacoustic collaborative system, multimodal large model and mental illness database.

[0025] EEG monitoring devices use electrophysiological indicators to record brain activity, monitoring an individual's delta, theta, alpha, and beta waves in real time and transmitting this data to a multimodal large-scale model for analysis. EEG monitoring devices monitor electroencephalograms (EEGs), a technology that records the brain's electrical activity. Electroencephalograms (EEGs) capture electrical signals from the brain using electrodes placed on the scalp. These electrodes are attached to the scalp and analyzed using Fourier transforms to identify the signal's spectral components, such as alpha, beta, theta, and delta waves.

[0026] Personal information: Personal information provided by users, including their medical history, imaging data, psychological data, etc., and the user's personal information is passed to the multimodal large model for analysis.

[0027] Photoacoustic synergistic system: It consists of two major devices, namely ultrasonic transducer equipment and infrared light source equipment. The ultrasonic transducer equipment adopts transcranial ultrasound stimulation technology (TUS), which uses low-intensity, low-frequency focused ultrasound to stimulate specific nerves or functional areas in the brain. The ultrasonic transducer equipment can emit ultrasound pulses of different frequencies to stimulate and treat patients with mental illness. The infrared light source equipment adopts transcranial infrared light stimulation technology, which uses low-intensity infrared light to regulate brain neural activity. By focusing infrared light of a specific wavelength on the cerebral cortex or deeper brain tissue, it can effectively achieve the excitation or inhibition of neurons, and then regulate the activity of specific functional areas. The infrared light source equipment suppresses / excites cortical background neural activity by emitting infrared light of different wavelengths, thereby increasing the gain of the ultrasonic stimulation effect. The photoacoustic synergistic system is dynamically controlled by a large multimodal model, and the intervention parameters are continuously adjusted to find the most suitable personalized treatment plan.

[0028] Ultrasonic transducer: used to convert electrical energy into acoustic energy (especially ultrasonic waves). In this application, the ultrasonic transducer is used to generate ultrasonic pulses of a specific frequency, and the frequency parameters are adjustable to achieve dynamic brain neuromodulation.

[0029] Infrared light source device: A device that emits infrared light. In this application, the infrared light source device is used to generate infrared light of a specific wavelength to suppress or excite, and the wavelength parameters are adjustable to achieve dynamic brain neuromodulation of cortical background neural activity and improve the gain of the ultrasound stimulation effect.

[0030] Multimodal large model: The core of the system framework of the present invention is used to process and evaluate multidimensional data such as brain activity data and personal information, formulate corresponding treatment plans based on individual conditions, make decisions to generate corresponding intervention parameters, and dynamically adjust the photoacoustic collaborative system in real time.

[0031] RAG is a natural language processing technology that combines information retrieval and text generation. It retrieves documents or snippets highly relevant to user queries from large knowledge bases and combines this information with generative models to generate more accurate, fact-based, and relevant answers or content. This approach not only improves the quality and reliability of generated content but also expands the model's knowledge coverage. It is widely used in intelligent question-answering systems, conversational bots, and content creation, effectively addressing the limitations of traditional generative models in knowledge depth and accuracy. RAG combines large models with databases.

[0032] Mental Illness Database: This vector database centrally stores knowledge about mental illnesses, addressing the potential for hallucinations caused by multimodal large models due to insufficient understanding of certain mental illnesses. It also stores previous user data and simulates a reinforcement learning mechanism, enabling the large model to continuously learn and optimize, automatically guide execution, and continuously track feedback.

[0033] The system operates as follows: A patient with a mental illness first uploads their personal information, including their medical history, imaging data, and psychological data, to a multimodal large-scale model. They then wear an EEG monitor to monitor brain activity in real time. The detected EEG waves are transmitted to the multimodal large-scale model, which comprehensively evaluates the user's multidimensional data and makes decisions. Intervention parameters are first generated, and then the photoacoustic synergy system is activated. The photoacoustic synergy system consists of an ultrasonic transducer and an infrared light source. The ultrasonic transducer is placed at a selected location on the head, coated with ultrasonic coupling gel to enhance ultrasound transmission. The infrared light source is positioned on either side of the ultrasonic coupling gel, bordering the ultrasonic transducer, to increase the gain of the ultrasound stimulation. Brain activity in addicts fluctuates depending on their state. For example, the reward system typically exhibits low function (requiring continuous improvement), but experiences a transient, strong surge (requiring instantaneous suppression) when exposed to relevant cues. The ultrasonic transducer device emits pulses of a specific frequency, and the infrared light source device emits infrared light of a specific wavelength. The two work together to stimulate the patient's brain. After the stimulation, the EEG monitoring device performs real-time detection and transmits the data generated after the brain stimulation to the multimodal large model. The large model makes decisions after evaluation of the intervention, regenerates the photoacoustic intervention parameters, and dynamically adjusts the mid- and far-infrared and ultrasonic frequency emission and control components to form a real-time data feedback loop.

[0034] Among them, the modulation mode optimization of the photoacoustic collaborative system and its neural regulation are as follows: Figure 2 As shown, the efficient and real-time closed-loop neural feedback control loop is a multi-stage process encompassing multimodal data application, real-time monitoring, feedback, and precise control. At the large-scale model representation and decision-making stage, a multimodal large model is utilized to process data based on core physiological markers, and deep learning algorithms are used to analyze patient status and condition changes in real time. Simultaneously, RAG technology is integrated with the addiction case database and knowledge base to construct precise neural intervention plans, enhancing the evidence-based interpretation and medical evaluation capabilities of photoacoustic parameter decision-making.

[0035] The multidimensional clinical data sensor continuously collects patient data (such as brain activity, behavior, and physiological state changes) and connects to the large-scale model to make photoacoustic parameter decisions and execute interventions, forming a real-time data feedback loop. The photoacoustic collaborative control intervention terminal autonomously generates and adjusts intervention parameters based on neural oscillation mechanisms, customizing treatment plans and directly impacting the patient's nervous system, alleviating symptoms or promoting recovery. When the data meets the trigger conditions, the large-scale model reassesses the patient's condition in real time and updates the photoacoustic stimulation parameters, ensuring the timely and precise intervention.

[0036] From perception to decision-making to intervention, an efficient closed-loop neural feedback control circuit is formed, building a "perception-decision-intervention" platform. Continuous data feedback enables the large model to continuously learn and optimize, adjusting treatment plans in real time to ensure precision and personalization, while automatically guiding execution and tracking feedback. Real-time monitoring of patient clinical manifestations helps understand the clinical markers and mechanisms of addictive disorders, improving treatment safety and effectiveness.

[0037] Among them, photoacoustic coordinated neural regulation Figure 3 Several studies have used transcranial light and sound stimulation to intervene in patients with addiction. For example, transcranial ultrasound targeting the nucleus accumbens in addicted patients can safely and effectively modulate reward circuitry, reducing craving and relapse rates. Transcranial light stimulation targeting the dorsolateral prefrontal cortex can alleviate craving symptoms. The mechanisms of addictive disorders are complex, involving cortical regions such as the dorsolateral prefrontal cortex and the nucleus accumbens, as well as deep nuclei. This requires neuromodulatory techniques that can produce strong stimulation effects on both large superficial cortical regions and smaller deep nuclei, and that can bidirectionally modulate (inhibit or enhance) neural activity in specific brain regions. By synergizing transcranial ultrasound with transcranial infrared light stimulation, infrared light can suppress background cortical neural activity, enhancing the gain of transcranial ultrasound stimulation on the cortex. Simultaneously, transcranial ultrasound can also inhibit or enhance neural activity in deep nuclei. This photoacoustic synergistic stimulation technique holds promise as a new approach for intervention in addictive disorders.

[0038] Single-target intervention: Synergistic stimulation of infrared light and unfocused ultrasound at a single target in the cortex. Transcranial light stimulation has a wide range of action and a regulatory effect of "the weak become weaker and the strong become stronger". Unfocused ultrasound is used to stimulate the cerebral cortex, and transcranial light stimulation is performed on the cerebral cortex simultaneously to enhance the excitatory or inhibitory effect of transcranial ultrasound stimulation, achieving an effect stronger than that of transcranial light or sound stimulation alone.

[0039] Multi-target intervention: Based on the circuit-based bidirectional photoacoustic synergistic stimulation of multiple targets in the cortex and deep brain areas, focused ultrasound is used to stimulate the deep target brain areas in the addiction-related circuit, and "light + unfocused ultrasound" is used to stimulate cortical targets to inhibit the overly excited brain areas in the addiction circuit and enhance the underexcitability of the brain areas in the addiction circuit.

[0040] The functions of the large model are as follows: Figure 4As shown: The multimodal addiction model will combine the patient's medical history, imaging data, and mental health data to perceive clinical parameter information, quickly integrate neural activity, behavioral information, and clinical manifestations, and transmit them to the large model for data fusion analysis. Analyze the patient's status in real time, generate and optimize the photoacoustic intervention plan. The intervention after the decision is implemented by the photoacoustic control device, which adjusts the treatment parameters according to the neural oscillation mechanism, directly acting on the patient's nervous system to alleviate symptoms or promote recovery. After the large model intervenes, it will conduct a post-intervention evaluation based on the patient's withdrawal status, degree of addiction, and psychological state, and store the evaluation results in the vector database. Based on the evaluation results, the current photoacoustic synergistic stimulation plan will be dynamically updated and improved to find the optimal intervention plan and implement it.

Claims

1. A closed-loop control system for the nervous system based on transcranial photoacoustic collaboration, characterized in that: It consists of EEG monitoring equipment, photoacoustic collaborative system, multimodal large model and mental illness database; Among them, EEG monitoring equipment uses electrophysiological indicators to record brain activity, which is used to monitor the individual's delta wave, theta wave, alpha wave, and beta wave in real time, and transmits the monitored data to the multimodal large model for analysis; The photoacoustic synergy system consists of two devices: an ultrasonic transducer and an infrared light source. The ultrasonic transducer emits ultrasonic pulses of varying frequencies to stimulate and treat patients with mental illnesses. The infrared light source emits infrared light of varying wavelengths to suppress background neural activity in the excitatory cortex, increasing the gain of the ultrasound stimulation effect. The photoacoustic synergy system is dynamically controlled by a large multimodal model, continuously adjusting intervention parameters to find the most suitable personalized treatment plan. The multimodal large model is used to process and evaluate multidimensional data including brain activity data and personal information, formulate corresponding treatment plans based on individual conditions, make decisions and generate corresponding intervention parameters, and dynamically adjust the photoacoustic collaborative system in real time; The Mental Illness Database is a centralized database for storing mental illness knowledge, which makes up for the lack of in-depth understanding of certain mental illnesses in the multimodal large model. It also stores previous user data and simulates a reinforcement learning mechanism, allowing the large model to continuously learn and optimize, automatically guide execution, and continuously track feedback. Closed-loop control is a multi-stage process, including the large-scale model representation decision-making stage, the multi-dimensional clinical data perception stage, and the optoelectronic coordinated control intervention stage. At the large-scale model characterization and decision-making end, a multimodal large model is used to process data based on physiological markers, and deep learning algorithms are used to analyze patient status and condition changes in real time. Continuously collect patient data at the multi-dimensional clinical data sensing end, including changes in brain activity, behavior, and physiological state, and connect to the large model to make photoacoustic parameter decisions and perform interventions, forming a real-time data feedback loop; The optoelectronic coordinated control intervention end autonomously generates and adjusts intervention parameters based on the neural oscillation mechanism, personalizes treatment plans, directly affects the patient's nervous system, alleviates symptoms or promotes recovery; when the data reaches the trigger condition, the large model re-evaluates the patient's status in real time and updates the photoacoustic stimulation parameters to ensure the timeliness and accuracy of the intervention.

2. A closed-loop control system for the nervous system based on transcranial photoacoustic collaboration according to claim 1, characterized in that: The user's personal information is passed to the multimodal large model, and the personal information includes the user's medical history, imaging data, and psychological data.

3. The closed-loop control system for the nervous system based on transcranial photoacoustic collaboration according to claim 1, characterized in that: The user first uploads their personal information to the multimodal big model, including patient medical history, imaging data, and psychological data. Then, they wear an EEG monitoring device to detect brain activity in real time and transmit the detected EEG bands to the multimodal big model. The multimodal big model comprehensively evaluates the user's multidimensional data and makes decisions. First, it generates intervention parameters and then turns on the photoacoustic collaborative system. The photoacoustic collaborative system contains an ultrasonic transducer device and an infrared light source device. The ultrasonic transducer device is set at a selected position on the head, which is coated with ultrasonic coupling agent, and the infrared light source device is set on both sides of the ultrasonic coupling agent bounded by the ultrasonic transducer device to increase the gain of the ultrasonic stimulation effect. The ultrasonic transducer device emits pulses of a specific frequency, and the infrared light source device emits infrared light of a specific wavelength. The two work together to stimulate the patient's brain. After receiving the stimulation, the EEG monitoring device performs real-time detection and transmits the data generated after the brain stimulation to the multimodal big model. The big model makes a decision after evaluation of the intervention, generates photoacoustic intervention parameters again, and dynamically adjusts the mid- and far-infrared and ultrasonic frequency emission and control components to form a real-time data feedback loop.

4. The closed-loop control system for the nervous system based on transcranial photoacoustic collaboration according to claim 1, characterized in that: Among them, photoacoustic synergistic stimulation includes: single-target photoacoustic synergistic intervention, including synergistic stimulation of infrared light and unfocused ultrasound at a single target in the cortex. Transcranial photostimulation has a wide range of action and a regulatory effect of "the weak becomes weaker and the strong becomes stronger". Unfocused ultrasound is used to stimulate the cerebral cortex, and transcranial photostimulation is performed on the cerebral cortex simultaneously to enhance the excitatory or inhibitory effect of transcranial ultrasound stimulation, achieving an effect stronger than that of transcranial light or sound stimulation alone.

5. The closed-loop control system for the nervous system based on transcranial photoacoustic collaboration according to claim 1, characterized in that: Among them, photoacoustic synergistic stimulation also includes multi-target photoacoustic synergistic intervention, including: circuit-based multi-target bidirectional photoacoustic synergistic stimulation of the cortex and deep brain areas, the use of focused ultrasound to stimulate deep target brain areas in addiction-related circuits, and the use of light + unfocused ultrasound to stimulate cortical targets, inhibiting overly excited brain areas in addiction circuits and enhancing underexcitable brain areas in addiction circuits.

6. The closed-loop control system for the nervous system based on transcranial photoacoustic collaboration according to claim 1, characterized in that: The functions of the large model include: combining the patient's medical history, imaging data, and mental health data, perceiving clinical parameter information, integrating neural activity, behavioral information, and clinical manifestations, and transmitting them to the large model for data fusion analysis, real-time analysis of the patient's status, and generating and optimizing photoacoustic intervention plans. The intervention after the decision is implemented by the photoacoustic collaborative system, which personalizes the treatment parameters according to the neural oscillation mechanism, directly acting on the patient's nervous system to alleviate symptoms or promote recovery.

7. The closed-loop control system for the nervous system based on transcranial photoacoustic collaboration according to claim 6, characterized in that: After the large model intervention, the patient's withdrawal status, addiction level and psychological state are evaluated after the intervention, and the evaluation results are stored in the vector database. According to the evaluation results, the current photoacoustic synergistic stimulation program is dynamically updated and improved to find the optimal intervention program and implement it.

8. The closed-loop control system for the nervous system based on transcranial photoacoustic collaboration according to claim 1, characterized in that: RAG technology is used to integrate large models and databases; RAG technology specifically retrieves documents or fragments that are highly relevant to user queries from the data, and combines this information with the large model to generate answers or content that are fact-based and more relevant.

Images