1. Background

In recent years, the deep integration of neuroscience and artificial intelligence has promoted the revolutionary upgrade of brain-computer interface technology. "Neuromodulation and brain-computer closed-loop interaction technology" has been listed as one of the important strategic directions of the future industry by the state. This project focuses on the treatment needs of closed-loop brain-computer interfaces for typical neurological diseases (such as Parkinson's disease, depression, etc.) and the application of intelligent rehabilitation. For example, the Adaptive Deep Brain Stimulation (DBS) system developed by Medtronic has been approved by the FDA to automatically adjust stimulation parameters based on the patient's real-time brain activity; For treatment-resistant depression, personalized closed-loop brain stimulation therapy has shown rapid and lasting symptom improvement in case studies. However, most of the existing closed-loop neuromodulation systems are based on the threshold feedback of a single physiological signal, lacking a deep understanding of the multimodal state and subjective purpose of patients, and the intelligent regulation strategy is relatively shallow, making it difficult to adapt to the complex and changeable brain state in time.

At the same time, the field of artificial intelligence is undergoing a paradigm shift from traditional "data-driven" to "intelligent self-knowledge". The Chinese research team took the lead in proposing the "Data-Information-Knowledge-Wisdom-Purpose" (DIKWP) artificial consciousness model, which adds a "Purpose/Purpose" layer on the basis of the classic DIKW framework to realize semantic collaboration and two-way feedback between various cognitive levels through a network structure, thus providing an innovative path for solving the "black box" problem of current AI models and improving the interpretability and controllability of AI decision-making process. This new cognitive system is an academic milestone and is regarded as an important "underlying code" to lead the safe, controllable, and explainable development of AI in the future, providing solid support for the move towards artificial general intelligence (AGI). Professor Yucong Duan, who proposed the model, has been authorized 114 invention patents, and the related core technical achievements have attracted worldwide attention. These preliminary accumulations have laid the foundation for the integration of artificial consciousness principles and brain-computer interface technology, and also provided strong support for the implementation of this project.

Based on the above background, this project is oriented to the guidance direction of the National Science and Technology Major Project "Research on Neuromodulation and Brain-Computer Closed-loop Interaction Technology", and proposes to construct a set of artificial consciousness neural interface system (DIKWP-NIS) with the closed-loop architecture of "multimodal perception-purpose generation-neuromodulation" as the core. The system will integrate the semantic perception and processing of multimodal information such as vision, hearing, touch, and proprioception, and use the "Purpose (Purpose) layer" of the DIKWP model as the center to generate meaningful regulatory targets for users, and use the Artificial Consciousness Processing Unit ([ACPU) to generate meaningful regulatory targets]{.underline} Realize adaptive and personalized control of neurostimulation devices. The project aims to break through the limitations of traditional brain-computer interfaces in passive response to a single signal, and develop a new paradigm of closed-loop regulation with autonomous cognitive ability, which can optimize and dynamically adjust stimulus parameters for scenarios such as Parkinson's disease tremor control and depression relief. While realizing the innovation of precision treatment and rehabilitation aids for major diseases, this project will explore a new model of brain-computer interaction in the 6G era, and play a leading role in the development of domestic autonomous neuromodulation chips and software platforms.

2. Theoretical basis

DIKWP Artificial Consciousness Model and Purpose Generation Principle: The DIKWP model abstracts the human cognitive process into five progressive elements: Data, Information, Knowledge, Wisdom, and Purpose. The model emphasizes the introduction of "purpose" driving in the process of intelligence generation from low to high: that is, the processing of each layer to the previous layer is guided by the main goal, so as to form a closed-loop cognitive architecture with autonomous purpose. Different from the traditional AI framework that only processes data-information-knowledge, the DIKWP model realizes the understanding and internalization of human purposes by machine intelligence by embedding the key high-order semantic layer of "Purpose", enabling AI systems to make autonomous decisions based on the goals given by humans. This opens up the possibility of establishing a common "cognitive language" between humans and machines, making every decision made by AI traceable and interpretable, and ensuring that its actions always serve human values and safety needs. From the perspective of consciousness science, the DIKWP model is in line with mainstream consciousness theories such as Global Workspace Theory (GWT) and Integrated Information Theory (IIT), which emphasizes the regulatory role of global purpose in decentralized information processing, and provides a feasible path for artificial systems to simulate the generation and self-regulation of purpose similar to human consciousness.

ACPU Architecture and Interpretable Neural Modulation Operating System: In order to realize the artificial consciousness function of the DIKWP model in engineering, the Artificial Consciousness Processing Unit (ACPU) architecture is introduced in this project. ACPU can be regarded as a combination of dedicated intelligent processor and operating system that integrates cognitive computing and neural regulation, and its core is an interpretable, adaptive, and self-regulating neuromodulation "consciousness operating system". The system draws on the concept of "semantic operating system" proposed by Professor Yucong Duan's team, and decomposes the complex decision-making process into five monitorable links according to the DIKWP framework: data, information, knowledge, wisdom, and purpose. Each link has a well-defined semantic state and algorithmic logic to ensure that the entire process from perceptual input to stimulus output is under intelligible and intervenable control. Specifically, the ACPU architecture includes: a multimodal semantic perception module (responsible for converting various signals from the brain-computer-body into DIKWP structured semantic representations); P-layer Purpose decision-making module (simulates the execution center functions of the prefrontal lobe of the brain, and generates the regulatory purpose of the current moment according to the perceptual semantics and built-in goals); and the Neuromodulation Execution Module, which includes semantic stimulus coding and [neurotransmitter regulation models described below]{.underline}, translating Purpose into specific neurostimulation parameters. Through this modular architecture, ACPU is able to adaptively adjust strategies for different patients and situations, enabling closed-loop optimal control of neurostimulation devices. In addition, due to the transparent and traceable decision-making link, the operating system is easy to integrate clinical knowledge and safety constraints, making the regulatory decision-making of Parkinson's disease, depression and other diseases both efficient and in line with medical norms.

Semantic Stimulus Coding and Neurotransmitter Regulation Modeling: How to map high-level Purpose to low-level bioelectric/chemical stimuli in neuromodulation circuits is a core scientific question. In this project, we propose a mechanism of "semantic stimulus coding", which translates the regulatory purpose generated by the P layer into the corresponding neurostimulation mode based on the DIKWP semantic space. For example, when the Purpose is "Tremor Relief", the system will select an electrical stimulation parameter mode that adjusts the excitability of a specific neural circuit based on this semantic goal; When the Purpose is "improving mood", a magnetic stimulation sequence that affects the reward/emotion circuitry in the brain or a chemical transmitter release regimen that targets a specific neural nucleus may be employed. To this end, we will construct a mapping relationship library between DIKWP semantics and neurophysiological parameters, and train it through machine learning combined with brain science knowledge. Furthermore, the project proposes a joint modeling of "semantic neurotransmitter regulation": considering that neurotransmitters are the internal mediators of the brain to achieve higher-order semantic functions such as emotion and motivation, we will establish a correlation model between different semantic states and the activities of specific neurotransmitter systems. For example, elevated dopamine levels usually correspond to increased motivation and reward expectations, and "purpose"-related volitional behaviors often involve regulation of the midbrain-limbic dopamine pathway; The release of serotonin is closely related to mood stability and depression relief. Through literature research and experimental data analysis, we will map semantic factors (such as "concentration", "pleasure", "nervousness", etc.) and key neurotransmitters (such as dopamine, 5-HT, norepinephrine, etc.) under the DIKWP framework, as an important basis for designing intervention strategies. This semantic-neurotransmitter mapping structure will support the system in selecting the most effective biological intervention pathway according to P-layer targets (e.g., locating specific brain regions to induce the release of desired transmitters) to achieve precise regulation based on individual differences.

[Purpose-transmitter resonance model]{.underline} and closed-loop regulation strategy: In order to achieve precise control of the dynamic brain, this project introduces the "purpose-transmitter resonance" theoretical model. The model assumes that in the closed loop of the artificial consciousness system (ACPU) and the biological brain, there is an optimal matching point for the user's purpose state and the brain's neurochemical state to reinforce each other—that is, the "resonant" state. When the P-layer purpose and the corresponding neurotransmitter activity rhythm in the user's brain reach a coordinated resonance, the regulatory effect will be most significant and stable. Accordingly, we will refer to the adaptive oscillator model in control theory to establish a coupling equation between the purpose signal and the concentration of neurotransmitters and the excitability of the brain network, and analyze the resonance characteristics of the brain network in different purpose modes. The model will be used to guide the design of a closed-loop control strategy: that is, the ACPU continuously adjusts the stimulation parameters to approximate the Purpose-transmitter resonance condition according to the monitored brain feedback, so as to achieve dynamic optimization of regulatory efficiency. When the user's state changes, such as mood fluctuations or sports purpose changes, the system can predict new resonance points through the model and quickly adjust the parameters to ensure that the closed-loop adjustment always resonates with the user's internal needs. The introduction of the Purpose - Transmitter resonance model has improved our closed-loop system from simple passive feedback control to model-driven active predictive control, which greatly improves the response speed and stability of the system in complex environments.

3. the overall goal and innovation points

Overall Objective: This project aims to develop a multimodal closed-loop brain-computer interaction system (DIKWP-NIS) based on the DIKWP artificial consciousness model, and realize the precise and adaptive regulation of the nervous system by introducing the semantic understanding and purpose decision-making mechanism of artificial consciousness. The specific objectives include: 1) constructing a multimodal brain-machine-body integrated semantic perception and purpose generation mechanism, so that the system can integrate and process multi-source inputs such as vision, hearing, touch, and electromyography, and extract the high-level semantic state related to the user's current goal; 2) Design a neuromodulation operating system under the architecture of the Artificial Consciousness Processing Unit (ACPU) to realize the self-optimization and closed-loop regulation of stimulus parameters for typical neurological diseases such as Parkinson's and depression, so as to ensure that the system decision-making process is explainable, intervenable, safe and reliable; 3) Establish a joint modeling method of "semantic stimulus coding" and "neurotransmitter regulation", explore the mapping relationship from semantic goals to specific electrical/magnetic/chemical stimuli, and develop individual-oriented precise neuromodulation strategies; 4) The theory of "Purpose-transmitter resonance" was proposed, and a closed-loop control algorithm based on model prediction was developed to achieve real-time tracking and optimal regulation of dynamic brain states. 5) Develop a variety of application prototypes, including brain-controlled rehabilitation training system, artificial consciousness assisted sensory compensation system, multimodal brain stimulation platform for cognitive impairment, etc., to verify the versatility and effectiveness of the technology of this project.

Innovation: This project has obvious innovations in the following aspects:

• *Innovation 1: Multimodal semantic collaboration and purpose-driven closed-loop architecture. *Breaking the traditional brain-computer interface model that relies on a single signal, the DIKWP artificial consciousness model is introduced for the first time to realize the semantic integration of multimodal information such as vision, hearing, and somatosensory, and the "Purpose" layer is used as the center to make decisions and control neurostimulation, so as to build a semantic closed-loop interaction of the "brain-machine-body" trinity.

• *Innovation 2: ACPU Artificial Consciousness Neuromodulation Operating System. *This paper proposes an artificial consciousness processing unit architecture for brain-computer closed-loop applications, and designs a dedicated operating system kernel with interpretability and autonomous adaptability. The OS can adjust stimulation parameters in real time according to the patient's status and make the AI decision-making process transparent to meet the needs of medical safety and personalized treatment, which is a paradigm upgrade of traditional neuromodulation devices.

• *Innovation 3: Joint modeling of "semantic stimulus coding" and "neurotransmitter regulation". *He pioneered the mapping of semantic information and neuromodulation parameters, and proposed a dual modeling method that considers both external stimulus patterns and internal neurotransmitter environments. Through semantic-guided neurostimulation coding and the regulation of intrinsic chemical signals in the brain, this project is expected to achieve more refined and humanized interventions.

• *Innovation 4: Closed-loop regulatory strategy of "Purpose-transmitter resonance". *A new strategy combining control theory and brain science is proposed: to find the resonance conditions between artificial purpose and brain transmitter activity through model simulation, and to guide stimulus regulation. This strategy is different from the traditional passive feedback control, but actively predicts and induces the system to enter the optimal resonance state, which significantly improves the efficiency and robustness of closed-loop control.

• *Innovation 5: A universal artificial consciousness neural interface platform across diseases and scenarios. *The results of the project will be verified in a number of typical scenarios in the form of a prototype system, including brain-like intelligent rehabilitation training, sensory impairment assistance, and closed-loop regulation of cognitive impairment, reflecting the general scalability of the technology. This cross-domain application capability stems from the general architecture of artificial consciousness that we introduced, which lays the foundation for expanding to more human-computer interaction scenarios (such as intelligent education and training, human-machine collaboration, and assisted driving brain control) in the future.

• *Innovation 6: Forward-looking layout for 6G communication and domestic ecology. *The project closely connects with the new requirements of the ultra-high-speed and low-latency network environment for brain-computer interface in the future 6G era, and reserves a cloud collaboration interface and a secure encryption channel in the system design to support the efficient transmission and remote control of brain-computer data. At the same time, priority will be given to the use of domestically developed chips and software platforms (including customized neuromodulation SoCs and real-time operating systems) to create an independent and controllable artificial consciousness brain-computer interface technology chain to help enhance China's international competitiveness in this field.

4. the technical route

Figure 1: The overall architecture of the DIKWP-NIS artificial consciousness neural interface system is shown Purpose. The system takes multi-modal semantic perception and purpose decision-making as the closed-loop center, realizes the adaptive control of the stimulation device through the ACPU operating system, and forms a closed-loop regulation link from environmental/physical signals to neurostimulation to physiological feedback.

Figure 1 shows the overall architecture of the DIKWP-NIS system to be developed in this project. The input of the system comes from the multimodal semantic perception module, which integrates various signals from the external environment and the user's body (such as camera field of view, microphone sound, tactile sensor, EEG/EMG signal, physiological parameters, etc.), and converts these raw data into structured semantic representations in the DIKWP model (corresponding to D, I, K, W levels) through semantic co-processing. On this basis, the P-layer Purpose Generation Module is entered, which is equivalent to the brain center of artificial consciousness, and combines the current contextual semantics and the system's built-in objective functions (such as mitigation goals for specific symptoms) to generate Purpose decisions for the next step of neural regulation. P-layer purpose can be derived from the understanding of the user's brain state (e.g., detected motor purpose or emotional demand), or it can be initiated by the system (e.g., guiding the user towards a training goal based on a rehabilitation strategy).

The Purpose decision is then passed on to the Neural Modulation Operating System kernel based on the ACPU architecture. The kernel contains two key functional modules: one is the semantic stimulus coding module, which selects or generates the corresponding stimulus parameter sequences (such as electrode combination, pulse frequency/intensity mode of electrical stimulation, coil position and frequency of magnetic stimulation) from the pre-established semantic-stimulus mapping library according to the P-layer purpose, so as to realize the translation of high-level semantic targets to low-level neurostimulation instructions; The second is the neurotransmitter modulation module, which is used to predict and evaluate the impact of the current purpose and stimulation regimen on the user's neurotransmitter level and overall brain network status, and can adjust the stimulation regimen if needed to maintain homeostasis or avoid side effects. The two modules work together under the scheduling of the ACPU operating system to continuously iteratively optimize the stimulus output. The generated stimulus control instructions are sent to a stimulation actuator, which can be an implanted deep brain electrode, a transcranial magnetic stimulation coil, a surface electrical stimulation electrode, or a future minimally invasive neurotransmitter release device. The stimulation actuator acts on the corresponding area of the patient's brain according to the instructions to implement physical or chemical regulatory interventions.

At the other end of the closed loop, the patient's brain and body respond to stimuli, and this response is transmitted back to the system input through a multi-source feedback pathway. Feedback signals include neurophysiological signals (such as EEG, local field potentials and other indicators that reflect the state of the brain), muscle movement or behavior performance (detected by wearable sensors), autonomic responses (heart rate, skin electroderm, etc.), and patients' subjective feelings and reports (obtained through human-computer interaction interfaces). The multi-source feedback is processed by the semantic perception module and upgraded to a high-level semantic representation, which is used by the P-layer Purpose module to evaluate the current regulation effect and decide the subsequent Purpose update. In this way, a closed-loop loop of "perception-purpose-stimulus-feedback" is formed: the system continuously adjusts the semantic understanding and regulation strategies of the environment and users according to the feedback, so that the whole interaction process has adaptive learning ability.

The above technical route can be summarized as the following implementation steps: firstly, through the collection and annotation of brain-computer-body multimodal data, the knowledge base and semantic parsing algorithm supported by the DIKWP model are established to ensure that the system can accurately "read" the user state and environmental information; Secondly, the Purpose reasoning engine was developed, which draws on cognitive reasoning and reinforcement learning methods, so that the system can make regulatory decisions that meet the long-term interests of users in complex situations. Thirdly, the stimulus coding and transmitter regulation algorithm were designed and optimized, and the model parameters were adjusted through biofeedback experiments to form an optimal stimulus output strategy driven by artificial consciousness. Fourth, develop ACPU software and supporting hardware prototypes to open up real-time communication in sensing, computing and stimulation to achieve closed-loop system integration; Finally, the effectiveness and safety of closed-loop control are verified through the simulation platform and some function demonstrations, and the performance of each module is gradually iteratively improved on this basis. The above route will be closely coordinated by theoretical research, algorithm development, prototype implementation and experimental verification to ensure the smooth realization of the project goals.

5. Task decomposition

In order to achieve the above objectives, the research content of the project is divided into the following six interrelated tasks:

1. Research on DIKWP Multimodal Semantic Perception and Purpose Generation Mechanism: Build a multi-channel data collection and annotation platform covering vision, hearing, touch, physiological signals, etc., and study the semantic representation methods and fusion algorithms of each modal information. A semantic parsing engine was developed based on the DIKWP model to realize entity recognition, event extraction, and context understanding of multi-source data. On this basis, a P-layer Purpose generation algorithm is developed to explore a computational model for inferring the user's Purpose/Target from the multimodal semantic state, solve the problem of Purpose recognition and uncertainty processing under noise interference, and form a mechanism prototype of brain-machine-body semantic collaboration.

2. ACPU architecture and neuromodulation operating system development: Design the system architecture of the Artificial Consciousness Processing Unit (ACPU), including software and hardware interface specifications, key module function definitions, and runtime scheduling mechanisms. Based on this architecture, an interpretable neural regulation operating system kernel is developed to realize the mapping and coordination of the processing processes of each layer of the DIKWP model in the system. Key research: multi-task real-time scheduling strategy to ensure low-latency response in the closed-loop of perception-decision-stimulation; The visual monitoring tool of the decision-making process realizes the transparency and traceability of the AI decision-making link. The safety isolation and emergency intervention mechanism ensures that the system can automatically switch to safe mode or remind manual intervention when abnormal conditions occur.

3. Semantic stimulus encoding and personalized parameter optimization: Construct a mapping library from semantic target to stimulus parameters, and preset candidate stimulus regimens (including electrical stimulation waveform parameters, magnetic stimulation protocols, drug doses, etc.) for typical semantic purposes (such as "reducing tremor amplitude", "improving attention", "relieving anxiety", etc.). In this study, the parameter adaptive adjustment method based on reinforcement learning and optimization algorithm is studied, so that the system can continuously fine-tune the stimulation parameters to adapt to individual differences and state changes under closed-loop operation conditions. This task will preliminarily verify the effectiveness of semantic stimulus coding on the in vitro neuronal culture and animal experiment platform, and form a software module for parameter optimization algorithms.

4. Semantic-neurotransmitter joint regulation model and algorithm: Carry out research on the mechanism of neurotransmitter monitoring and regulation, select key transmitters closely related to emotion and motivation (such as 5-HT, DA, etc.), and establish a quantitative correlation model between them and brain activity patterns and behavioral indicators. Develop algorithms that can estimate the user's transmitter status in real time (with the help of EEG, brain microdialysis sensing, or metabolic indicators, etc.), and design transmitter regulation strategies (such as indirect transmitter modulation methods such as transcranial electrical/magnetic stimulation). On this basis, a purpose-transmitter resonance control algorithm was proposed to realize the target-guided dynamic adjustment of transmitter level. The algorithm is verified and the parameter sensitivity analysis is carried out by simulation tools such as Matlab/Simulink, which provides theoretical support for closed-loop control.

5. Prototype system integration and application verification: The above modules are integrated to build a DIKWP-NIS prototype system, including sensor interface, ACPU main control board, stimulation output interface and host computer monitoring software. The user-friendly interface is designed to allow researchers and clinicians to access the internal status of the system and adjust parameters. The functional test and verification of the prototype system were carried out in typical scenarios, such as the construction of "brain-controlled rehabilitation training" experiment (the patient controls the virtual reality training task through the brain-computer interface, and the system adjusts the brain stimulation enhancement training effect in real time), and the "emotional intervention" experiment (monitors the emotional changes of the subjects and gives stimulation intervention in real time) to evaluate the effectiveness, safety and robustness indicators of the system under different applications.

6. Application demonstration and optimization: In cooperation with clinical and industrial partners, the prototype of this system is deployed and trialized in the actual application environment, and the initial demonstration and application of Parkinson's disease patients in motor symptom regulation, emotional counseling in depressed patients, and memory training in patients with cognitive impairment are carried out. Collect clinical feedback and user experience data, and iteratively optimize the system's algorithm parameters, human-computer interaction design, and hardware reliability. Formulate corresponding technical specifications and operating guidelines to lay the foundation for subsequent large-scale application. At the same time, combined with the demonstration results, the technical limitations and risks of this project are analyzed, and the next research and improvement direction is proposed.

6. Feasibility analysis

The implementation of the project is based on a solid research foundation and mature technical conditions. From the perspective of theoretical foundation, Professor Yucong Duan's team has been working hard in the field of basic theories of cognitive computing and artificial intelligence for many years, and the DIKWP artificial consciousness model and supporting patent achievements have constructed a new cognitive language and architecture from "black box" to "white box". These early accumulations provide a unique theoretical advantage for this project, which enables us to take the lead in introducing the principle of artificial consciousness into the closed-loop regulation of brain-computer interfaces. In terms of algorithms and software, the research team has rich experience in semantic understanding, machine learning, intelligent control, etc., and the relevant research results have been published in high-level journals at home and abroad, and there are successful application cases in the fields of medical big data analysis and affective computing, which will help the development and verification of key algorithms of the project.

From the perspective of hardware and experimental conditions, the current technology platform of neuromodulation and brain-computer interface is becoming more and more complete. The existing implantable brain stimulator supports the closed-loop operation of real-time signal acquisition and stimulation output, which provides a mature hardware foundation for the development of new control algorithms in this project. The laboratory of the team is equipped with high-density EEG/EMG acquisition equipment, transcranial magnetic stimulator (TMS), EEG-stimulation synchronous control platform, etc., which can meet the needs of multi-modal data acquisition and closed-loop simulation experiments. In addition, we have established partnerships with a number of hospitals and rehabilitation centers to obtain the actual data and clinical needs of target user groups such as Parkinson's disease and depression, and ensure that the research protocol is suitable for real-world application scenarios.

As far as the external environment for the implementation of the project is concerned, the national policy strongly supports the cross-innovation of the new generation of artificial intelligence and brain science, and the relevant scientific research plans and capital investment continue to increase, which provides a strong guarantee for the smooth development of the project. At the same time, the forward-looking layout of 6G communication technology has created the necessary conditions for large-scale brain-computer interaction systems—ultra-high-speed and low-latency networks can enable brain-computer interface devices and cloud AI models to achieve real-time connection and collaborative computing. This means that the system developed in this project has good potential for expansion, and can be connected to the cloud knowledge base and computing resources to enhance functions in the future. In addition, the rapid development of domestic high-performance computing chips and smart sensors enables us to use domestic devices to build an efficient ACPU hardware platform, so as to avoid dependence on imported technology, reduce costs and improve the safety and controllability of the system.

In terms of risks, this project involves the intersection of medicine, artificial intelligence, electronic engineering and other disciplines, and may face challenges in the research and development process, such as complex algorithm models that are difficult to converge and insufficient access to human experimental data. In this regard, we have set up a joint research team composed of artificial intelligence experts, brain scientists, clinicians and engineering technicians to ensure that each key link has professional support in the corresponding fields. In the project plan, links such as algorithm simulation verification and small-sample pre-experiments are reserved to find problems and adjust the scheme in a timely manner. Considering the ethical and safety requirements of human trials, we will mainly use simulation and ex vivo experiments in the prototype verification phase, and work closely with the ethics committees of partner hospitals to gradually carry out limited human trials while ensuring safety. In summary, the project has a solid upfront foundation and controllable implementation risks, and is expected to complete the set goals as planned.

7. Stage objectives

The project is developed in three phases, each with clear goals and milestones:

• Phase 1 (initial year of the project, approx. 1 year): Completion of basic theory and key technology verification. The specific objectives include: the DIKWP multimodal semantic perception and purpose inference prototype system has been preliminarily built, and its effectiveness for simple tasks has been verified in the laboratory environment; The core framework of the ACPU architecture and the neural regulation operating system has been built to realize the basic joint debugging of each module. The semantic stimulus coding and neurotransmitter regulation model was preliminarily tested in the simulation environment, and the initial parameter set for Parkinson's tremor control and depression emotion regulation was obtained. Phase Milestone: A phase review meeting was held to present the results of the multimodal Purpose identification demonstration and closed-loop control simulation experiments, and the first phase goals were evaluated and confirmed by the expert group.

• Phase 2 (mid-project, approx. 2-3 years): Completion of system integration and pilot validation. The specific goals include: the software and hardware integration of the DIKWP-NIS prototype system has been completed, and the ACPU operating system has the ability to operate stably; Experiments in a closed environment were carried out on a small sample of subjects (or clinical cases) to verify the closed-loop regulation effect of the system on Parkinson's disease symptoms (such as limb tremor) and depressive mood, which showed advantages over traditional open-loop stimulation. The semantic-stimulus mapping library and parameter adaptive algorithm were optimized to improve the adaptability of the system to individual differences. Phase Milestones: Formation of a complete prototype device that enables closed-loop regulatory demonstration of at least two typical functions (motor control and emotion regulation); Submit an interim research report, including detailed experimental data and effect analysis, and pass the mid-term inspection and review.

• Phase 3 (late stage of the project, approx. 4–5 years): Completion of application demonstration and finalization of results. Specific objectives include: conducting application tests with larger samples in partner hospitals or research institutions, such as trials of more than 50 patients with Parkinson's disease or depression, to evaluate the clinical effectiveness and safety of the system; According to the test feedback, the friendliness and reliability of the human-computer interaction of the system are further improved, and a product specification can be formed for clinical or industrial reference; At the same time, exploratory application verification was carried out in extended scenarios such as sensory impairment rehabilitation and cognitive training, and multi-domain performance data was collected. Stage milestones: release the final project report and technical specification documents, and complete the finalization of system software and hardware; Apply for no less than 5 relevant invention patents and software copyrights, and publish papers in authoritative academic journals to explain core innovations; During the acceptance of the project, a series of application cases and complete technical data can be displayed to prove that the expected goals of the project have been fully achieved.

8. Expected accomplishments

Upon completion, the project will have the following expected outcomes and impacts:

• Core technology and patents: A set of original core technologies of artificial consciousness neural interface has been formed, including DIKWP-NIS system architecture design, ACPU operating system implementation, Purpose-transmitter closed-loop control algorithm, etc. It is planned to apply for no less than 5 invention patents, covering key innovation points such as semantic stimulation coding methods, neurotransmitter regulation devices, and artificial consciousness control chip architecture, so as to consolidate the advantages of independent intellectual property rights.

• Prototype system and platform: One set of DIKWP-NIS prototype equipment was developed, including multi-modal sensing module, ACPU processing module and neurostimulation module, which can demonstrate Parkinson's tremor inhibition and depression intervention in the experimental environment. Supporting the development of an open software platform, the core algorithm is encapsulated into a callable API interface, which is convenient for subsequent researchers and developers to carry out functional expansion and secondary development based on this system. The completion of the prototype system and platform will lay the foundation for future industrialization.

• Academic Papers and Standards: Published no less than 8 papers in high-level academic journals or conferences in the fields of artificial intelligence, brain-computer interface, biomedical engineering, etc., systematically expounding the theoretical innovation and experimental results of the project, and enhancing China's international academic influence in the field of artificial consciousness and brain-computer interaction. Combined with project experience, he participated in the formulation of relevant industry standards or guidelines, such as "Technical Requirements for Artificial Intelligence Closed-loop Neuromodulation System" or "Specification for Clinical Application of Brain-Computer Interface", etc., and worked with domestic counterparts to promote the standardization of this emerging field.

• Talent training and team development: Through the implementation of the project, no less than 5 interdisciplinary interdisciplinary scientific research talents (postdoctoral fellows, doctoral students, etc.) will be cultivated, so that they can master cutting-edge knowledge such as artificial consciousness models, brain-computer interface technology and medical AI applications, and reserve high-level R&D strength for related fields in China. The project team will grow into one of the leading teams in this interdisciplinary field, with the ability to undertake large-scale national projects and international collaborative research, laying the organizational foundation for continuous innovation.

• Socio-economic benefits: This project provides a new approach to the treatment of neurological diseases and rehabilitation aids, and is expected to achieve significant benefits in the management of chronic diseases such as Parkinson's disease and depression. Closed-loop neuromodulation technology is expected to reduce patients' dependence on drugs and their side effects, improve quality of life, and reduce long-term medical burden. Economically, if the technology is further transformed, it can give birth to new medical devices and intelligent rehabilitation products, forming a potential industrial growth point. As an innovative practice of combining artificial intelligence and brain science, this project will also improve China's international competitiveness in this frontier field, and is in line with national strategic goals such as "Healthy China" and "Self-reliance and Self-reliance in Science and Technology".

9. Application demonstration and promotion plan

In the later stage of project development and after the completion of the project, we will actively promote the demonstration and application of the technology in practical scenarios, and formulate a comprehensive promotion strategy:

1. Demonstration of typical application scenarios:

• Active Medicine Brain-Controlled Rehabilitation System: In cooperation with rehabilitation hospitals, DIKWP-NIS was applied to limb function training after stroke. Through the multi-modal monitoring of the system, the patient's motor purpose and fatigue degree are identified, and the parameter rhythm of transcranial electrical stimulation or functional electrical stimulation is adjusted in real time to assist the patient to complete limb movement training. It is planned to recruit a certain number of hemiplegic patients for a controlled trial to evaluate the effect of this system in improving the efficiency of motor function recovery and shortening the rehabilitation cycle compared with traditional rehabilitation therapy.

• Artificial Consciousness Assisted Sensory Impairment Intervention: This system is deployed in special education or disability assistance institutions for perceptual enhancement and information assistance for people with visual and auditory impairments. For example, for blind people, the system uses a visual camera to obtain environmental information, and after semantic analysis, it communicates the key points of the scene to the user through auditory or tactile feedback. At the same time, it monitors its EEG to determine the focus of attention, and intelligently adjusts the way information is presented to match the user's needs. Through a series of case verifications, the effect of the system on improving the environmental understanding ability and self-care ability of people with perceptual impairment was evaluated.

• Closed-loop Regulatory Platform for Cognitive Impairment: Trialing the cognitive intervention function of this system in a group of patients with memory impairment or Alzheimer's disease. Transcranial magnetic stimulation (TMS) combined with cognitive training tasks was used to regulate the closed-loop stimulation of the patient's brain memory network: when attention is distracted or memory retrieval difficulty is detected in the cognitive task, the system automatically applies electromagnetic stimulation to specific brain regions to promote neural network activation. Through a period of intervention trial, the differences in cognitive assessment results between the intervention group and the control group were compared, and the effect of the system on the improvement of symptoms of different subtypes of cognitive impairment was verified.

2. Promotion and industrial transformation path:

• Clinical Collaboration and Validation: After the initial results of the demonstration application, we will work closely with large tertiary hospitals and rehabilitation centers to conduct larger-scale clinical trials and multi-center studies to obtain the safety and efficacy data required for regulatory approval. This will lay the foundation for the subsequent application for medical device registration with the State Food and Drug Administration.

• Industrial cooperation and incubation: Actively connect with medical device manufacturers and neuromodulation equipment companies, and explore the possibility of jointly establishing industrialization projects or incubating start-ups. Taking advantage of the company's advantages in manufacturing and market channels, we will accelerate the upgrading of the prototype of this project into a commercial product that meets medical standards. In particular, we will promote the tape-out trial production of domestic neuromodulation chips based on ACPU architecture, and combine them with the software platform of this system to create the first set of artificial consciousness neuromodulation equipment that can be applied on a large scale in China.

• Standard formulation and policy docking: Based on the results of the project, we will work with relevant units in the industry to submit draft industry technical standards, such as "Technical Requirements for Intelligent Brain-Computer Interface Systems" or "Guidelines for Clinical Application of Closed-loop Neuromodulation", to improve the standardization and recognition of this technology in the industry. At the same time, we will strive to incorporate the technology of this project into the national and local key science and technology promotion plans (such as the Wisdom Medical Demonstration Project, the Digital Life and Health Project, etc.), expand the influence with the help of policy support, and accelerate the transformation of achievements.

• Academic and public outreach: Promote the concept and results of the project and promote more opportunities for cross-field cooperation by hosting academic seminars and participating in international conferences on artificial intelligence and brain science. At the public level, the prospect of artificial consciousness neural interface in the field of medical rehabilitation is introduced through mainstream media reports, popular science articles and technology displays, so as to improve the society's awareness and acceptance of the innovative technology and create a good public opinion environment for the promotion and application of the product in the future.

In summary, this project will ensure that the research results can be "researched and useful" through well-designed demonstration applications and multi-faceted promotion measures. We expect that within a few years after the completion of the project, we will gradually push the DIKWP-NIS system into practical clinical application, provide new diagnosis and treatment intervention methods for the majority of patients, and lead the vigorous development of the domestic brain-computer interface and artificial intelligence integration industry.

References:

1. Yucong Duan et al. DIKWP Model of Artificial Consciousness: Theory, Design and Simulation Chinese Journal of Artificial Intelligence, 2024

2. Yucong Duan et al. Artificial Intelligence vs. Artificial Consciousness: A DIKWP Perspective Proceedings of an International Conference, 2025

3. Medtronic press release. The world's first adaptive DBS system was approved, 2023.

4. Scangos et al. Efficacy of Closed-loop Brain Stimulation on Depression . Nature Medicine, 2021.

5. NIH Research. Is Dopamine Influencing Brain Decision-Making Worth the Effort, 2020.

6. Harvard Health. Serotonin: A Natural Mood Booster, 2019.

7. A communication laboratory. Real-time connection between brain-computer interface and cloud AI model in 6G environment, 2025.