1. Background and significance

Cognitive impairment neurodegenerative diseases (such as Alzheimer's disease, frontotemporal dementia, etc.) are one of the major medical and social challenges facing the world and China. These diseases are characterized by high incidence, irreversible progression of the disease course, and heavy burden of care. According to statistics, there are currently more than 16 million dementia patients in China, and it is expected to increase to about 115 million in 2050, ranking first in the world. Alzheimer's disease (AD) is the most common form of cognitive impairment dementia, characterized by progressive memory loss and deterioration of overall cognitive function. Frontotemporal dementia (FTD) is more common in younger people, and is prominently manifested by personality and behavior changes, language impairment, etc. These diseases not only seriously affect the quality of life of patients, but also bring a heavy burden to families and society, so it is of great significance to study their pathogenesis and explore effective diagnosis and treatment strategies.

The pathological mechanisms of cognitive impairment neurodegenerative diseases are extremely complex. Taking AD as an example, the pathogenesis of AD is thought to be the result of the long-term accumulation of a variety of harmful factors, including amyloid plaque deposition, abnormal phosphorylation of Tau protein to form neurofibrillary tangles, synaptic and neuronal loss, neuroinflammatory responses, and brain metabolic disorders. At the cellular and molecular levels, the neuronal microenvironment of AD brain has problems such as abnormal energy metabolism, decreased glucose utilization, and lipid metabolism disorders, resulting in impaired neural network activity, decreased plasticity, and accelerated deposition of pathological products. The loss of synaptic connections is the direct anatomical basis of cognitive impairment in AD, and the extent of which is highly correlated with dementia severity. Studies have shown that synaptic loss is an early feature of many neurodegenerative diseases, including AD, and is the main pathological change associated with cognitive dysfunction. The occurrence of synaptic dysfunction is closely related to the abnormal activity of glial cells (microglia and astrocytes): activated glial cells mediate excessive synaptic phagocytosis and pruning, resulting in an imbalance in excitatory/inhibitory neurotransmission and exacerbating network dysfunction. In addition, the metabolic imbalance between the brain and the periphery is also an important factor, such as the significant reduction of glucose and oxygen consumption in the brain and the impaired insulin signaling in AD patients, which is vividly called "type III diabetes of the brain". Systemic metabolic disorders (such as abnormal glucose and lipid metabolism, iron ion imbalance, etc.) can affect the central nervous system through the blood-brain barrier and neuroendocrine pathways, accelerating neurodegenerative changes. In contrast, the pathological mechanism of FTD varies according to different pathological types (such as Tau proteinopathy, TDP-43 proteinopathy, etc.), showing a high degree of genetic and pathological heterogeneity. The clinical manifestations of FTD patients are varied, ranging from loss of behavioral inhibition and impulsivity to progressive decline in language function, and even members of the same family with the same mutation may have very different clinical manifestations. Due to the complex diversity of phenotypes and mechanisms, the pathogenesis of cognitive degenerative diseases is far from being fully elucidated.

In clinical practice, the current diagnosis and treatment methods for cognitive degenerative diseases are still very limited. There is no cure for AD and FTD, and existing drugs (such as cholinesterase inhibitors, donepezil, memantine, etc.) can only alleviate symptoms to a certain extent, but cannot prevent disease progression. Even newly approved immunotherapies for AD amyloid (e.g., monoclonal antibodies) only delay the course of the disease, have limited effect on cognitive impairment that has already occurred, and have safety concerns. A common difficulty in these disorders is the insidious onset, the lack of specific symptoms in the early stages, and the widespread irreversible damage to the brain by the time the clinical diagnosis is made. For example, pathological changes in the brain of AD patients may begin to accumulate 20 years before cognitive symptoms appear. In the early stages of FTD, it is often misdiagnosed as psychiatric illness or depression, and the intervention window is missed. Due to the lack of sensitive and reliable early biomarkers and screening methods, the clinical identification of high-risk individuals (such as patients with mild cognitive impairment and MCI) and the timing of intervention are insufficient. At the same time, the current assessment of cognitive function mainly relies on scales such as the Mini-Mental State Examination (MMSE), the Montreal Cognitive Assessment (MoCA), and neuropsychological tests, which have limited ability to detect impairments in high-level cognitive functions (such as executive function, insight, and purpose planning), and often focus on low- and middle-level functions such as memory and language. However, higher-order functional disorders, such as lack of self-awareness and reduced behavioral purposefulness, lack quantitative indicators in traditional assessments, so clinical identification and targeted treatment are insufficient. As a result, although some patients have no obvious memory impairment, their "wisdom" and "purpose" functions such as decision-making ability and social cognition have been significantly reduced, but they have not been paid attention to in time, which has missed the opportunity for intervention.

In the face of the above challenges, it is necessary to explore new theoretical frameworks and technical means to systematically analyze the whole chain evolution mechanism of cognitive degenerative diseases from molecular pathology to high-level cognitive functions, so as to develop new diagnostic and therapeutic strategies. The Data-Information-Knowledge-Wisdom-Purpose (DIKWP) model provides an innovative perspective for this. The DIKWP model is an artificial consciousness cognitive framework originally created by Professor Yucong Duan's team, which adds the highest level of "Purpose/Intention" to the classic DIKW (Pyramid) model (Data–Information–Knowledge–Wisdom).Dimension. The model believes that there are five layers of abstract semantic hierarchy in the cognitive system, from data to information, to knowledge, Wisdom, to Purpose, and that the transmission between each level is not simple, but through the network structure to achieve two-way feedback and iterative update. That is to say, the low-level perceptual data is extracted into information, further integrated to form knowledge, and then elevated to wisdom through experience and judgment, and finally guided by wisdom to form the purpose and purpose of action. In turn, the purpose and purpose of the higher level play a guiding and constraining role in the information processing and knowledge application of the lower level. The introduction of the "Purpose" layer is the key innovation of the model, which acts as the "North Star" of the entire cognitive chain, ensuring that cognitive activities are clearly goal-oriented. The DIKWP model establishes a unified cognitive semantic language that enables every step of complex cognitive processes to be parsed and understood. This model framework was originally applied to the field of artificial intelligence, providing a new way to solve the "black box" problem of large-scale pre-trained models and improve the interpretability of AI systems. By embedding a "purpose" layer inside the model and making the process traceable at each level, DIKWP makes the AI decision-making process transparent, so that every step of the reasoning is well documented. It is worth noting that the DIKWP model is essentially a general cognitive structure abstraction, which is equally applicable to the description and analysis of human cognitive processes. Therefore, the introduction of the DIKWP model into the study of neurodegenerative cognitive disorders is expected to transform the data from data-information-knowledge-wisdom-purpose For example, AD patients have a decrease in information processing speed and difficulty in acquiring knowledge (corresponding to information and knowledge barriers) in the early stage, a decrease in judgment and problem-solving ability in the middle stage (Wisdom layer disorder), and a late loss of life goals and autonomous behavior motivation (Purpose layer disorder). This DIKWP-based cognitive degradation profile will be more comprehensive than the traditional memory-executive function division, which can capture the subtle changes of higher-order cognitive function degeneration and provide new clues for identifying early abnormalities.

At the same time, from the perspective of Artificial Consciousness theory, it also brings innovative value to the understanding and intervention of advanced functional disorders of such diseases. Dementia is not only a loss of memory, but also involves a decline in the level of consciousness: patients often have reduced self-awareness, empathy, and lack of behavioral purpose, which can be regarded as manifestations of impaired human consciousness. If we think of the brain as a system of information processing and consciousness formation, then neurodegenerative diseases can be seen as the system that produces local or global dysfunction, resulting in a decrease in the ability to "quasi-conscious". Most of the existing studies explain the symptoms at the molecular and network levels, but lack the discussion on the phenomenological aspects of consciousness. The introduction of artificial consciousness theory can help build bridges from neural mechanisms to conscious experience. For example, drawing on consciousness frameworks such as Global Workspace Theory (GWT) or Integrated Information Theory (IIT), it is possible to re-understand why AD patients gradually lose integrated cognition of the environment and self, and why FTD patients lack awareness of their own lesions (anosognosia). The DIKWP artificial consciousness model proposed by Professor Yucong Duan itself integrates the ideas of multiple mainstream consciousness theories, and simulates the cognitive flow and self-regulation mechanism of human consciousness through a five-layer framework and a "dual circulation" structure. The application of this model to the patient's brain is expected to elucidate the disintegration process of higher-order cognitive functions (such as decision-making purpose and value judgment), and clarify the association between these higher-order dysfunctions and underlying neuropathology. This has important implications for the development of targeted interventions, such as the Purpose Drive that stimulates patient retention.

In summary, this study takes "pathological mechanisms and novel diagnosis and treatment strategies of neurodegenerative diseases with cognitive impairment" as the core proposition, and takes the DIKWP model and artificial consciousness theory as the core theoretical support, aiming to open up the research chain from molecular pathology to advanced cognition. On the one hand, the DIKWP framework system was used to characterize the degeneration trajectory of patients' cognitive function at the levels of data, information, knowledge, wisdom and purpose, and to reveal the dynamic change law of the whole cognitive chain with the evolution of the disease. On the other hand, from the perspective of artificial consciousness, we can grasp the barriers of higher-order functions (such as self-awareness, purpose-driven, etc.) that cannot be fully quantified by traditional methods, and design innovative evaluation indicators and intervention strategies for these aspects. Through the in-depth intersection of basic mechanism research and artificial intelligence technology, this project will provide new ideas and means to solve the mechanism problems of such diseases and explore new diagnosis and treatment options, which has great scientific significance and application value.

2. Research objectives and overall technical roadmap

The overall goal of the research is to systematically study the mechanism and relationship of functional deterioration at each level around the five-layer cognitive structure proposed by the DIKWP model for cognitive degenerative diseases, establish a new hierarchical diagnosis and evaluation system, and develop a consciousness-like assisted diagnosis and treatment strategy combining biology and artificial intelligence. Specifically, the objectives of this project include: (1) Revealing data, information, knowledge, wisdom, and purpose Under the five-layer framework, the degradation mechanisms and mutual influences of each level in the progression process of cognitive impairment were clarified, and the key pathological links and potential intervention targets were clarified. (2) Construct a hierarchical evaluation index system and quantitative model of cognitive function based on DIKWP to achieve fine representation and early warning of patients' cognitive status; (3) develop AI-driven intelligent assessment and diagnosis tools, using machine learning to fuse multimodal data to improve the detection rate and diagnostic accuracy of early cognitive abnormalities (especially the decline in higher-order wisdom function and consciousness level); (4) Propose and validate new cognitive intervention strategies from the perspective of consciousness, and design training programs that can activate the functions of patients' Wisdom and Purpose layers; (5) Integrate the above results to develop a consciousness-inspired cognitive rehabilitation system, which can be used to assist clinical cognitive training and home rehabilitation of patients, and ultimately improve the quality of life of patients.

Overall technical route: This project adopts the route of "basic mechanism research-cognitive modeling-technology development-system integration", which is promoted layer by layer from bottom to top, and at the same time, the DIKWP theory runs through the whole process, integrating multidisciplinary methods. The technical route is generally divided into five links, which are as follows:

• Multimodal Neurobiological Mechanisms Study →

• Cognitive layer degradation modeling based on DIKWP →

• DIKWP Structured Cognitive Assessment Tool Development →

• Class awareness-based intervention program design →

• Intelligent cognitive rehabilitation system integration.

First, through multimodal neurological and molecular biology studies, abnormal changes in neurons, synapses, glia, metabolism and other aspects of diseases are obtained, so as to provide a biological basis for cognitive models; Secondly, the above findings were integrated into the cognitive framework of DIKWP to establish a model of disease cognitive function deterioration, and to clarify the mapping relationship between functional impairment and biological mechanism at each level. Then, guided by the model, hierarchical evaluation indicators and methods were designed, and AI-driven evaluation tools were developed to achieve comprehensive quantitative assessment and intelligent diagnosis of patients' cognitive status. Then, based on the DIKWP model, the intervention ideas were proposed, and the training methods and class awareness aids that could stimulate the ability of the patients' Wisdom/Purpose layer were designed. Finally, the evaluation and intervention module is integrated into an intelligent rehabilitation system that combines software and hardware, and verified and optimized in real scenarios. The above technical roadmap covers the complete process from mechanism exploration to application transformation, ensuring that the research results can be closed-loop verification, gradual iteration, and finally application. The details of each phase and key technical solutions are described in detail below.

3. Research content and technical route

3.1 Neuromolecular mechanisms and brain microenvironment

The first part of this study focuses on the neurobiological basis of cognitive impairment degenerative diseases, aiming to elucidate the neuromolecular mechanisms and changes in the brain microenvironment underlying the deterioration of various layers of cognitive function. We will conduct in-depth research in three key areas: synaptic function, glial cells, and brain-body metabolism

• Synaptic Injury and Altered Network Connectivity: Synapses are the basic units of neural information transmission, and their structural and functional integrity are critical to cognitive processes. Numerous studies have shown that synaptic loss and dysfunction are highly correlated with cognitive decline and are the most important anatomical basis in the pathogenesis of dementia such as AD. In this project, we will use brain histopathology, animal models, and neuroimaging to detect changes in synaptic density and plasticity in the early and advanced stages of the disease, with a particular focus on synaptic plasticity decline in cognitively relevant brain regions such as the hippocampus and prefrontal cortex. At the same time, super-resolution imaging, electrophysiological recording and other techniques were used to study the abnormalities of presynaptic and postsynaptic elements (such as neurotransmitter release of presynaptic neurons, postsynaptic dendritic spine density and receptor expression), and the effects of excitatory and inhibitory synaptic dysbalance on neural network dynamics were investigated. We hypothesize that the loss of synaptic connections first affects the processing efficiency of the data layer and the information layer in the DIKWP model, because sensory input and primary information processing depend on intact synaptic networks; As synaptic lesions expand, knowledge acquisition and integration (knowledge layer) and further impact Wisdom decision-making ability ( Wisdom layer). Through synaptic pathology studies, we will identify synaptic structural/functional indicators that are closely related to dysfunction in various layers of cognition and screen for possible targets for intervention (e.g., synaptic plasticity-related proteins).

• Glia and neuroinflammation: Microglia and astrocytes, among others, form the "helper" networks in the brain, but in neurodegenerative diseases, they often go from being supporters to destroyers. Microglia overactivation and astrocyte dysplasia are often seen in the brain of AD and FTD patients, accompanied by a chronic inflammatory environment. These glial changes are strongly associated with synaptic injury: activated microglia can engulf synapses through the complement pathway, and astrocyte dysfunction affects neuronal metabolic supply and ion homeostasis. In this study, immunohistochemistry and single-cell sequencing were used to analyze the changes in glial cell morphology and gene expression profiles in the brains of patients and model animals. The phagocytic activity of microglia, the secretion of inflammatory factors, and the nutritional support function of astrocytes in the course of the disease were investigated. We will elucidate the link between glial-mediated chronic inflammation and cognitive impairment at the Wisdom/Purpose level: for example, does excessive clearance of synapses by microglia lead to disconnection of higher cognitive circuits? Does the reduced support of astroglia for presynaptic terminals impair the ability to process complex information? Through longitudinal monitoring of the brain microenvironment, we hope to identify several neuro-immune regulatory pathways as intervention entry points to delay cognitive degeneration. Novel glial-related biomarkers, such as inflammatory factors or glial-derived proteins in plasma/cerebrospinal fluid, will also be explored in the study to inform early diagnosis.

• Mechanisms of Brain-Body Metabolic Imbalances: There is growing evidence that neurodegenerative cognitive impairment is strongly associated with systemic metabolic health. For example, patients with AD often have systemic metabolic problems such as insulin resistance, dyslipidemia, and decreased cerebral glucose utilization. The energy crisis in the brain can directly impair the function of neurons and accelerate the accumulation of pathological products, thus forming a vicious circle. In this study, brain imaging (such as FDG-PET to detect brain glucose metabolism), peripheral metabolic index monitoring (glucose tolerance, blood lipids, inflammatory mediators) and metabolomics analysis will be used to comprehensively evaluate the brain-body metabolic status of patients. We focus on how impaired neuronal energy metabolism (decreased mitochondrial function and ATP production) affects the efficiency of information processing and knowledge acquisition in cognitive processes, as well as the chronic damage to the brain caused by peripheral metabolic disorders (such as diabetes mellitus, hyperlipidemia) through vascular and inflammatory pathways. Based on these data, we will construct a metabolic-cognitive correlation map of disease and identify metabolic factors that are significantly associated with the rate of cognitive decline. For example, the presence or absence of changes in the levels of certain metabolites predicts the shift in cognition from mild impairment to dementia. For the discovery of important metabolic pathways, we will also conduct intervention experiments (such as diet control, exercise training, or metabolic drug interventions) in animal models to verify whether improving systemic metabolism can help maintain cognitive function. This part of the study will provide a mechanism for non-pharmacological intervention strategies (e.g., exercise therapy, dietary modification) for cognitive impairment.

Through the above multimodal mechanism research, we try to elucidate the biological causes of the degradation of cognitive function at different levels of "molecular-cellular-network-holistic". Expected Results: Identify at least 2-3 key pathological mechanisms or targets (e.g., specific synaptic proteins, glial pathways, or metabolic factors) and explain their causal relationship with cognitive impairment. These mechanistic discoveries will not only enrich the understanding of disease occurrence and progression, but also provide scientific support for subsequent cognitive modeling and intervention design. The data obtained from mechanistic studies (including pathological indicators, molecular markers, etc.) will be used as input to the cognitive model of DIKWP, so that the model has a solid biological foundation.

3.2 DIKWP model-driven cognitive layer degradation modeling

After grasping the important neurobiological changes of the disease, the second part of the study will construct a theoretical model of cognitive deterioration driven by the DIKWP model to explain and simulate the full-chain decline of cognitive function with the progression of pathology. Based on the DIKWP artificial consciousness framework proposed by Professor Yucong Duan, the model maps the biological abnormalities found in the previous section to various abstract levels of the cognitive system, so as to reveal the characteristics and internal connections of diseases at different cognitive levels.

Specifically, we will first define the meaning of the five levels of DIKWP in the context of disease cognition:

• Data: Refers to the primitive sensory input received by the brain and the neural signals generated. In a healthy state, the data layer includes the normal acquisition of sensory information such as vision, hearing, and touch, as well as neuronal firing patterns. In disease states, the data layer may involve decreased sensory function (e.g., audiovisual and visual impairment) as well as abnormalities in neuronal firing (e.g., epileptic firing or diminished resting-state functional connectivity of neural networks). The compromise of the data layer will directly affect the quality of information extraction.

• Information layer: Refers to the representation of raw data after it has been interpreted, filtered, and encoded. In cognitive processes, it is equivalent to attention selection, retrieval of meaningful signals, and temporal storage of working memory. Disorders may lead to reduced function of the information layer, such as difficulty concentrating, decreased working memory capacity, and slower processing of information from complex environments. These changes are often the source of early subjective cognitive difficulties.

• Knowledge: refers to the semantic network, factual knowledge and skill reserve formed through learning and memory. Under normal circumstances, individuals are constantly integrating new information into existing knowledge structures. The typical effects of disorders such as AD on the knowledge layer are memory loss, difficulty learning new things, and loss of semantic knowledge (e.g., language disorder). Patients with FTD may also have difficulties in semantic comprehension at the knowledge level. The degradation of the knowledge layer directly corresponds to the symptoms of forgetfulness and the difficulty in applying knowledge seen in clinical practice.

• Wisdom: refers to the ability to understand, judge, synthesize, and innovate on the basis of mastered knowledge, including higher-order cognitive functions such as abstract thinking, decision-making and planning, problem solving, and situational response. The Wisdom layer is an individual's ability to apply knowledge to solve real-world problems and make informed decisions. In the process of dementia, the function of the wisdom layer gradually disintegrates, which is manifested as a decrease in judgment, a deterioration in the ability to solve daily problems, and difficulty in coping with new situations. In frontotemporal dementia in particular, loss of executive function and social cognition is an important manifestation of Wisdom layer disorder. Deficits at this level often affect the patient's ability to live independently earlier than memory problems.

• Purpose Layer (Purpose/Intention): This is the highest layer that is particularly emphasized in the DIKWP model. It represents the motivation, goal, and will of the individual, i.e., the "ultimate purpose of the service of the cognitive process". The Purpose layer runs through and guides the activities of all levels below, which corresponds to the vision and direction of the entire cognitive system. Under normal circumstances, a clear purpose keeps our cognitive behavior centered around achieving a goal. In neurodegenerative diseases, the function of the Purpose layer is often overlooked but critical. For example, patients with advanced AD have apathy, loss of interest and initiative in any activity, which is typical of Purpose dysfunction; Patients with FTD may present with disturbances in purpose-directed behaviors, such as impulsive behaviors or repetitive stereotyped behaviors, reflecting impaired purpose-making mechanisms. The disintegration of the purpose layer means that the patient gradually loses the ability to self-drive behavior and the understanding of the purpose of his behavior, which eventually leads to the inability to take care of himself.

Once the above layers are defined, we will map the biological mechanisms to the cognitive level changes through theoretical analysis and modeling. This includes collating the important neuropathological events identified in Section 3.1 and classifying and mapping them to potentially affected links in each layer of the DIKWP. For example, reduced hippocampal synaptic density primarily affects knowledge layer (memory storage) ability, while frontal neural network disconnection and dopaminergic dysfunction may explain the Wisdom layer (executive function and motivation) disorders. We will use systems dynamics modeling or causal mapping to characterize the causal chain from molecular pathology to cognitive deterioration. In this process, the two-way feedback characteristics of the DIKWP model are introduced: not only how the pathology accumulates from the bottom up leads to cognitive decline, but also how the decrease in cognitive activity in turn accelerates physiological deterioration (e.g., the loss of the Purpose layer leads to passive inactivity, which in turn aggravates the decline of metabolism and brain function, and the vicious circle of "use in and in waste").

In order to quantify the model, we will combine the actual data to estimate the modeling parameters. For example, the cognitive assessment scores (including memory, executive function scores, etc.) and the corresponding brain imaging and biomarker data were used to establish a correlation model between functional and pathological indicators at each layer in longitudinal follow-up. Structural equation model (SEM) or Bayesian network can be used to fit the pathway model of cognitive decline by using DIKWP layers as latent variables and pathological mechanism indicators as observation variables. This makes it possible to estimate the relative progress of degradation at each level and its contribution to the overall cognition. We will also try to construct a simplified cognitive system model (e.g., agent-based simulation) through computer simulation, assign it DIKWP structure, and introduce AD-like pathological parameters (such as reducing connection weights, increasing random noise, etc.) to observe the changes in system behavior to verify the rationality of the model.

Through this module, we hope to obtain a comprehensive model that can explain the phenomenon of "cognitive dimensionality reduction" (that is, the loss of higher-order abilities such as Wisdom / Purpose, which gradually degenerates into only fragmented knowledge and even difficulty in processing basic information). This model will shed light on the differences in cognitive deterioration pathways that may exist between different patients: for example, some patients develop memory deficits first (knowledge level degenerates first), while others exhibit decreased judgment (). The Wisdom layer is damaged first), which may be related to differences in their pathological type or brain reserves. Models will also help predict disease progression: by observing changes in lower-level functions, it is possible to infer when there will be significant disintegration of high-level Wisdom and Purpose, so as to guide early clinical intervention. More importantly, the model will point out key turning points and weaknesses at all levels, providing a theoretical basis for us to design targeted assessments and interventions.

Expected Results: This part will develop a theoretical framework and a preliminary mathematical model of the "DIKWP Cognitive Degeneration Model", and elucidate: (1) the mapping of the correspondence between biological pathology and cognitive function levels; (2) the initial sequence of cognitive degradation and the interaction mechanism of each cognitive layer; and (3) potentially modifiable links (e.g., the compensatory effect of strengthening a certain level of function on overall cognition). This lays the groundwork for the subsequent development of assessment tools and interventions that can be optimised for specific levels of functional impairment in an evidence-based manner.

3.3 Development of intelligent evaluation and early screening diagnostic tools

With the cognitive degeneration model, we will move on to the third part of the study: the development of intelligent cognitive impairment assessment and early screening and diagnosis tools based on the DIKWP hierarchical structure. The core idea is to combine traditional cognitive assessment methods with artificial intelligence technology, and use machine learning to extract subtle clues of "early consciousness abnormality" or "wisdom function decline" from multi-source data, so as to achieve sharper and more comprehensive detection than existing methods. Early and accurate assessment is essential for timely intervention and slowing the disease progression.

The specific research content and technical route include:

• Design of DIKWP hierarchical evaluation indicator system: Based on the model established in Section 3.2, we will design the evaluation tasks and indicators covering each layer of DIKWP. Traditional scales focus on memory (knowledge layer) and orientation/language (information layer), while we will supplement the tests of the Wisdom and Purpose levels. For example, at the Wisdom layer, we can design test scenarios to examine the subject's comprehensive reasoning, planning, and problem-solving skills (e.g., having the patient complete a daily task plan that spans multiple steps, assessing their logic and co-ordination skills); At the Purpose layer, motivational questionnaires or behavioral observations can be used to quantify the intensity of the patient's willingness to actively plan activities and the frequency of goal-directed behaviors. In addition, the functions of the data layer and the information layer also need to be objectively evaluated, such as sensory impairment tests, simple reaction time tasks, attention span tests, etc. We will integrate these tests to form a DIKWP Cognitive Function Assessment Scale. The scale will be presented in modules, each corresponding to a level of DIKWP, so that the results of the assessment can depict the level at which the patient has major deficits. For example, a patient with a fair memory score but a low Wisdom score suggests that higher-order executive function impairment may precede memory impairment.

• Multimodal data collection and digital evaluation: In order to improve the objectivity and granularity of evaluation, we will introduce a variety of new digital technologies. First of all, the cognitive assessment software is used to implement the test on a tablet or computer, and the accurate data such as reaction time and accuracy are automatically recorded to reduce the error of manual scoring. Second, wearable devices are utilized to monitor behavioral and physiological signals related to cognition, such as gait and daily activity patterns (representing purpose-driven initiative), sleep quality (related to brain health), and changes in speech intonation (which may reflect language organization and emotional willingness). Studies have shown that non-invasive signals such as voice and gait contain cognitive health information that can be used for inexpensive and efficient screening. We will collect data on short free speech and gait walking in subjects, analyze the content coherence, complexity and emotional expression of speech through artificial intelligence, analyze the speed, stride length and stability of gait, and extract features reflecting cognitive status. For example, patients with early AD may have increased speech pauses and poor vocabulary, which can be detected by NLP algorithms. People with mild cognitive impairment have increased gait fluctuations when walking, which can also be captured by wearable devices. In addition, we consider the response of neurophysiological signals such as electroencephalography (EEG) or functional near-infrared imaging (fNIRS) in simple tasks to assess brain information processing efficiency and network connectivity. The acquisition of multimodal data will provide rich input for machine learning models and help improve the accuracy of evaluation.

• Machine Learning Model Extracts Early Marker Features: After obtaining the above multidimensional data, we will build a machine learning/deep learning model to mine the key feature combinations that best distinguish early patients from normal aging individuals. Since the DIKWP model provides theoretical guidance for our feature selection, we will classify features into the model according to the hierarchy (e.g., perception/reaction time as the data layer feature, memory score as the knowledge layer feature, complex task performance as the Wisdom layer feature, etc.). Initially, we will use supervised learning methods to train classification models (such as support vector machines, gradient boosting trees, deep neural networks, etc.) to determine whether subjects have mild cognitive dysfunction or early dementia tendencies. The model training will use the annotated datasets collected in this project (classification such as normal cognition, MCI, early AD, etc.), and be tuned through cross-validation. At the same time, we will also try unsupervised**/self-supervised methods to automatically learn latent representations from the data to spot patterns that are imperceptible to the human eye. For example, an autoencoder is used to fuse multimodal data, extract a low-dimensional representation, and then analyze the differences in the distribution of this representation among different populations. White-box AI diagnosis is also one of the features of this project, and we plan to use the structure of DIKWP to explain the model: for example, through hierarchical regression analysis, the importance of the characteristics of each DIKWP layer to the model output, so that doctors can understand the basis of model decision-making. This approach is similar to** the white-box approach adopted by Prof. Yucong Duan's team in AI model evaluation, which makes the AI cognition and decision-making process transparent. Eventually, we will form an intelligent early screening algorithm, whose inputs include scale scores, numerical behavioral parameters, physiological indicators, etc., and output them as an assessment report of the respondent's cognitive health status. For example, the report will give a quantitative score for each DIKWP level (similar to a "data/information/knowledge/wisdom/purpose" 5D radar chart), the degree of deviation relative to normal people of the same age, and an overall risk score. In this way, even the slightest high-level functional abnormality, such as the Wisdom layer or the Purpose layer score is lower than the normal range, our system can alert you early. This full-link evaluation system has been verified in the evaluation of model awareness level in the field of AI, and it has been proved that it can comprehensively analyze the cognitive ability structure of the tested object.

• Early Screening and Clinical Validation: The evaluation tools developed need to be validated for validity and utility in real populations. We plan to work with hospitals and communities to conduct clinical trials on a certain scale. In memory clinics or geriatric units, we will evaluate older adults who come to the clinic complaining of memory/cognitive problems using the system, and a specialist will make a diagnosis based on current clinical criteria (gold standard, such as clinical diagnosis of MCI or mild dementia). By comparison, we can calculate the sensitivity and specificity of the system for MCI/early dementia and compare the performance with traditional scale screening (e.g., MoCA). Our goal is to significantly improve the detection rate of early anomalies and reduce missed and misdiagnosed. For example, the sensitivity of the system to MCI is expected to increase by at least 20% compared to MoCA, and the specificity will remain at a high level. In addition, we will test the system in cognitively normal older adults to assess their predictive ability by looking at the relationship between their assessment scores and cognitive outcomes at follow-up. If certain indicators (e.g., Wisdom tier score) are found to significantly predict the risk of cognitive decline over the next 2 years during cohort follow-up, this proves that our assessment system is forward-looking. Eventually, we will refine the human-machine interface and reporting format of the assessment tool to make it suitable for daily use by clinicians and primary care providers.

Expected Results: A prototype of an intelligent cognitive assessment system based on the DIKWP model was produced, including: (1) a set of DIKWP hierarchical cognitive assessment scales and task packages with reliability and validity; (2) The early screening AI model fused with multimodal data has reached a practical level in terms of area under the ROC curve (AUC) and other performance indicators; (3) the test in the clinical setting proved that the tool can significantly improve the sensitivity and accuracy of early diagnosis compared with traditional methods; and (4) the assessment report format and indicator interpretation manual, making it a generalizable application. This achievement will fill the gap in the current cognitive assessment in high-level functional testing, provide a white-box and digital decision support tool for clinical practice, and provide a quantitative basis for follow-up intervention.

3.4 Design of consciousness-assisted therapy systems

In the fourth part of the project, we will develop a consciousness-like assisted therapy system for disease intervention. Guided by the DIKWP model, the system helps delay cognitive decline and improve patients' ability to take care of themselves by activating the patient's residual Wisdom layer and Purpose layer functions, combined with artificial intelligence technology. Different from traditional cognitive training, which focuses on memory exercises, our strategy emphasizes the reconstruction of higher-order cognition and active consciousness, which is a new model of "quasi-consciousness" rehabilitation.

The main research contents and innovations are as follows:

• Design of cognitive training tasks (Wisdom layer activation): According to the Wisdom layer function of DIKWP, we will design a series of training tasks to exercise the comprehensive cognitive ability of patients. These tasks go beyond simple memorization and involve more reasoning, planning, and problem solving. For example, you can design a scenario simulation task: give the patient a daily life situation (e.g., planning a shopping trip, preparing a meal) and ask them to list the steps to deal with the changes and problems that arise along the way. This training is designed to stimulate the patient's brain to execute the network, exercising logical reasoning and adaptability. Strategy games and puzzles (puzzles, Sudoku, etc.) can improve the ability to think abstractly and pay attention. Studies have shown that regular cognitive training can improve memory and executive function in people with mild to moderate dementia. We will present these trainings in a gamified format to make them more fun and engaging. Each task will be combined with a quantitative assessment to monitor the patient's performance progress, allowing for individualized adjustment of the training difficulty.

• Motivational Guidance and Purpose Layer Activation: In view of the common lack of motivation and depression and apathy of patients, we will introduce elements of motivational stimulation and meaning therapy to help patients rebuild their sense of purpose in life. On the one hand, the system will set up a goal management module: for example, guide patients to set daily/weekly small goals (such as tidying up the room, talking to friends), and give reminders and feedback, so that patients can gradually develop the habit of active planning. By breaking down "big goals" into "small tasks", patients can regain a sense of control and accomplishment. On the other hand, we draw on life review therapy in psychology, allowing patients to recall life stories and refine positive emotions through multimedia interaction, and find the meaning of continuous efforts from past experiences. These processes help activate the remaining spark in the patient's sense of self and rekindle interest in social interactions and activities. In addition, we plan to use social robots or virtual assistants to interact with patients and provide them with emotional support and companionship. Studies have shown that virtual peers with personality traits can reduce loneliness and apathy in people with dementia, encouraging them to communicate and be more active. Our system will implement an AI assistant with simple conversation and emotion recognition capabilities, which will talk to patients on a regular basis to understand their emotional state and encourage participation in training. This human-computer interaction attempts to simulate social interaction and has a positive effect on maintaining the patient's mental activity.

• Virtual Reality and Multi-Sensory Stimulation: In order to enhance the training effect and fun, we will introduce virtual reality (VR) and augmented reality (AR) technologies to build immersive cognitive training scenarios. For example, develop a VR home living environment in which patients are required to complete a series of daily tasks (finding objects, cooking, etc.) to exercise their multi-step execution and spatial memory skills. The immersion of VR can stimulate the multi-sensory pathways of the patient's brain and enhance the learning effect, and some studies have confirmed that VR intervention can improve the cognitive and motor function of the elderly with mild cognitive impairment. In addition, sensory stimulation therapies such as music and light stimulation (such as playing familiar songs with image recall) are combined to activate brain networks at multiple levels. These multi-sensory inputs enrich the stimuli at the data and information layers in the DIKWP framework, helping to lay the foundation for higher-level cognitive training.

• Human-Computer Interaction and Biofeedback: Our quasi-consciousness rehabilitation system will focus on the friendliness and intelligent adjustment of human-computer interaction. On the one hand, through the large screen, tablet or VR equipment to provide an intuitive interface, patients can simply operate or voice commands to interact with the system; On the other hand, we collect various data from the training process, such as the patient's facial expressions and voice (reflecting emotional engagement), heart rate, and physiological indicators such as electrodermal energy (reflecting the level of concentration and relaxation). Using this feedback, the system can adjust the pace and content of the training in real time, for example by automatically reducing the difficulty of the task when fatigue is detected, or giving positive encouragement. Our goal is to create an adaptive learning system that adapts the protocol to the patient's condition to improve training effectiveness and adherence.

• Integration and prototype implementation of rehabilitation system: The above sub-modules (cognitive training games, motivation guidance units, VR scenes, human-computer interfaces, etc.) will be integrated on a unified platform. We will develop a software prototype (which can run on a tablet or an all-in-one rehabilitation terminal) with the necessary hardware (e.g. VR glasses, wearable monitoring devices). The prototype system (tentatively named "CogConscious") will have features such as user management, training plan development, data logging and analysis, and more. To make it easier for healthcare professionals to use, we will design a management interface that allows you to browse patient training data, adjust intervention priorities, and link with assessment tools. For example, if the assessment results show that a patient has a low Wisdom tier score, the management recommends focusing on the corresponding training module. The entire system focuses on security and privacy protection, and conforms to medical AI system specifications.

• Preliminary efficacy evaluation: We plan to carry out a small-sample efficacy observation study in the later stage of the project, select a certain number of patients with mild to moderate cognitive impairment, and use this system for 6-12 months of intervention training, and the weekly training frequency and duration are determined according to individual tolerance. The control group was given conventional cognitive interventions (e.g., paper-and-pencil training or traditional rehabilitation instruction). We will compare changes in cognitive assessment scores, daily functioning scales, and behavioural symptoms (e.g. depression, apathy scores) between the two groups before and after the intervention. It is assumed that our class awareness intervention can achieve better results, such as a significant reduction in deterioration on the executive function test, a smaller decline in scores on the Life Independence Questionnaire, and even an improvement in some positive indicators. With special attention to metrics related to the Wisdom and Purpose levels, such as the number of patients' voluntary activities, willingness to actively communicate, etc., it is expected that the training group will outperform the control. This will validate the viability of our innovative intervention strategy.

Expected Result: Delivery of a prototype of a quasi-consciousness cognitive training and rehabilitation system and demonstration of its practical value in a small-scale trial. Specifically, it includes: (1) a set of training task sets and motivational stimulation programs for higher-order functions of dementia, which are scientific and operable as evaluated by experts; (2) System software prototype and hardware solution, verify usability through user testing; (3) preliminary clinical trial data showing the effect of cognitive training in maintaining or improving patients' cognitive function (e.g., the decline rate of memory and executive function in the 6-month training group was reduced by X% compared with that in the control group, and the quality of life score was improved); (4) At least 1-2 relevant research papers reporting the training program and test results. If the results are significant, we will lay the foundation for further large-scale validation and product rollout.

4. Feasibility analysis

This project is led by Prof. Yucong Duan, whose team has deep research accumulation and unique advantages in the intersection of artificial intelligence and cognitive science, which provides a reliable guarantee for the smooth implementation of this research.

First of all, the theoretical foundation is strong and the core technology is original. Professor Yucong Duan is an internationally renowned expert in the field of cognitive computing and artificial consciousness, and is currently an academician of the International Academy of Advanced Technology and Engineering, and the chairman of the World Association of Artificial Consciousness. He took the lead in proposing the DIKWP artificial consciousness model, which extended the Purpose layer on top of the classic DIKW framework to construct a network cognitive structure with two-way feedback. This theoretical system is an academic milestone and is considered to be one of the key paths to achieve strong artificial intelligence and artificial consciousness. What is even more commendable is that Professor Duan's team has put the DIKWP model into concrete implementation and obtained a series of independent intellectual property rights: up to now, the team has been authorized 114 invention patents (including 15 PCT international patents), covering many cutting-edge directions such as large model training, artificial consciousness construction, and cognitive operating system. These patented technologies have laid a solid technical reserve. In particular, in the area of explainable AI (white-box AI), the team used the DIKWP model to successfully design a method that makes the AI decision-making process transparent and controllable. For example, they propose to embed DIKWP into a large language model, decompose its reasoning into five links: data, information, knowledge, wisdom, and purpose, each step has a mathematical definition, and the model output can be supervised, so as to realize the semantic operating system prototype of the large model. This shows that the team has the ability to translate complex cognitive theory into engineering implementation, which is of great benefit to the development of interpretive diagnostic AI and transparent decision support system for this project.

Secondly, the previous research results are rich and the multidisciplinary experience is rich. In recent years, Yucong Duan's team has produced a series of internationally leading results in artificial awareness and cognitive assessment. For example, the DIKWP white-box evaluation system led by it has become an innovative benchmark for the "awareness level" of evaluation models in the AI field: the "Large Language Model Awareness Level "Knowledge" Quotient "White-box DIKWP Evaluation Report" released in 2025 builds a full-link evaluation system based on the DIKWP model, comprehensively quantifying the cognitive ability of mainstream large models from data, information, knowledge, wisdom, and purpose. The report shows that this system can deeply analyze the performance of modules such as perception, reasoning, problem solving, and purpose of the model, setting a new standard for AI cognitive evaluation. This result demonstrates the team's creativity in the cognitive assessment of complex systems and the ability to organize international collaborations (with the participation of more than 90 institutions around the world). This full-chain assessment concept will be directly borrowed from the development of human cognitive assessment in this project. In addition, in the medical field, the team has begun to explore the application of the DIKWP model: for example, the DIKWP artificial consciousness dialogue system has been developed in the intelligent medical consultation scenario, which can deeply understand the patient's description and give doctor-like consultation guidance (this study was published in a well-known academic forum and gained attention). In addition, the "Purpose-driven DIKWP Physiological and Artificial Consciousness Prototype" constructed by the graduate students of the team under the guidance of Professor Duan won the Best Poster Award at the 2023 China Digital Service Conference. The prototype integrates the cognitive cycle of data-information-knowledge-wisdom and the purpose driving mechanism of physiological state, and preliminarily demonstrates the feasibility of artificial consciousness system in simulating biological cognition. These cross-disciplinary attempts show that the team has accumulated some experience not only in the field of pure AI, but also in the combination of AI + medical engineering, which lays the foundation for interdisciplinary cooperation in this project.

Third, the technical and experimental conditions are complete, and multidisciplinary cooperation supports. With the support of the School of Computer Science of Hainan University and cooperative medical institutions, this project will make full use of the existing experimental conditions. The team has a supporting artificial intelligence experiment platform, including high-performance computing servers and GPU clusters, which can meet the needs of deep learning model training and massive data processing. The cognitive laboratory is equipped with VR equipment, EEG/NIR imagers, wearable sensor kits, etc., which can carry out human-computer interaction experiments and physiological data collection and analysis. In addition, the team has established cooperative relationships with a number of hospitals and research institutions, such as the Department of Geriatrics of Hainan Provincial People's Hospital and the Cognitive Neuroscience Laboratory of Ocean University of China, which will provide support in patient recruitment, sample testing, and image acquisition. The project team members have diverse disciplinary backgrounds, including artificial intelligence experts and software engineers, as well as neurobiologists, clinicians and rehabilitation therapists, to achieve a true cross-disciplinary team. Among them, the medical partners have the qualifications and experience to carry out clinical research, which can ensure the quality of clinical samples and trials. This project plans to apply for ethical approval and obtain informed consent from patients, and conduct research under the premise of ensuring ethics and privacy.

Finally, the project team has rich experience in project management and achievement transformation. Prof. Yucong Duan has undertaken a number of national research projects and has mature experience in the organization and risk management of large-scale projects. In the past five years, the team has published dozens of high-level papers, authorized more than 100 patents, and actively promoted the transformation of production, education and research, and cooperated with science and technology enterprises to apply some patented technologies to practice. For example, the team is working with a high-tech company to build a joint laboratory for artificial awareness to explore the implementation of the DIKWP model in business intelligence software. This means that if the results of this project achieve the expected progress, there will be ample opportunities to accelerate the transformation through school-enterprise cooperation, incubation and entrepreneurship, etc., and bring practical benefits to the society.

In summary, the team responsible for the project is fully prepared in terms of theory, technology, resources and cooperation. The original DIKWP theory provides strong academic support for the project, the evaluation and prototype system accumulated in the early stage prove the feasibility of the scheme, and the abundant multidisciplinary resources ensure the implementation conditions of the research. It can be expected that under the guarantee of the above favorable conditions, the project will be able to complete the scheduled research tasks with high quality and achieve the expected goals.

5. Milestones and milestones

The proposed research cycle of this project is 5 years, and the work focus and assessment nodes are divided by year. The milestones and milestones for each year are as follows:

• Year 1: Basic mechanism research and model conception stage. It focuses on experimental research on neuromolecular mechanisms and brain microenvironment, and accumulates data on disease mechanisms. At the same time, the framework of DIKWP cognitive degradation model was preliminarily established. Milestone 1: Elucidate at least 2 key biological mechanisms closely related to cognitive deterioration (e.g., the discovery that specific synaptic protein abnormalities are associated with memory decline, or the activation of an inflammatory pathway is associated with executive function decline), and form research reports or publish papers. A preliminary portrait of the DIKWP hierarchy of cognitive impairment was drawn, including a hypothetical relationship diagram of the functional changes of each layer. At the end of the first year, a workshop was held to evaluate the findings of the mechanism and to revise/enrich the cognitive model accordingly.

• Year 2: Cognitive degeneration model development and validation stage. According to the data of the first year, the DIKWP model was improved, and the quantification and preliminary validation of the model were completed. Milestone 2: Construct a mathematical model of DIKWP cognitive degeneration, which can be used to simulate the changes in functional scores of each cognitive layer at different stages of the disease. Data from a cohort of follow-up patients were used to validate the explanatory power of the model (e.g., the predicted time of decline of the Wisdom layer by the model was consistent with the actual clinical stage). Write model theory papers or software work descriptions, and apply for copyright. At the end of the second year, the model was judged to be logical and reasonable and the parameters were based on expert argumentation, which could be used to guide the development of the evaluation system.

• Year 3: Intelligent assessment tool development and experimentation phase. Based on the model, the hierarchical evaluation system was designed, and the software development and algorithm training of the evaluation tool were completed. Milestone 3: Developed a prototype DIKWP cognitive assessment system to achieve scale testing, multimodal data collection and risk scoring functions. Complete data collection of no less than 100 subjects, including groups such as cognitively normal, MCI, and early dementia. The early detection AI model was trained, and its performance indicators reached 0.85 for MCI recognition, > sensitivity > 85%, and specificity > 80% (the specific indicators can be adjusted according to the experimental data). Write a technical report or apply for a patent for an evaluation tool. At the end of the third year, a mid-term evaluation meeting was held, and through comparative experiments against clinical standards, it was proved that the evaluation system was better than the traditional scale, and had the value of continuous optimization and promotion.

• Year 4: Quasi-awareness intervention program development and system integration stage. Design and implement a cognitive training system for class consciousness to integrate assessment and training. Milestone 4: Completed the prototype integration of the "quasi-consciousness assisted therapy system", including the development and debugging of functional modules such as cognitive training games, motivation guidance modules, VR scenes, human-machine interfaces and data synchronization. The system was piloted in a laboratory setting, and feedback was collected from at least 30 patients, with zero adverse events. Optimize the user-friendliness and difficulty of the training protocol based on feedback. At the end of the fourth year, a small sample efficacy observation study was initiated to compare the changes in cognitive scores before and after 6 months of training. Expected results: The training group significantly improved or slowed down the decline in executive function or life ability assessment compared with the control group. Organize the details of the intervention plan and apply for relevant invention patents (such as "cognitive training method based on artificial consciousness").

• Year 5: System Improvement and Clinical Demonstration Stage. In the final year, the evaluation and training system is integrated and refined to validate and prepare for application on a larger scale. Milestone 5: Completion of clinical pilot validation – no less than 2 partner hospitals/rehabilitation centers were selected, a total of 50-100 target patients were recruited, the system was used in the actual clinical setting for evaluation and partial intervention, and follow-up results were collected. Statistical analysis showed that the evaluation indicators of this system were significantly correlated with clinical outcomes, and the functional maintenance of the intervention group was better than that of the historical control. Draft clinical guidelines were formed, including the application process of DIKWP cognitive assessment and intervention. At the same time, he completed the project summary and published no less than 2 papers in top academic journals, covering new discoveries of mechanisms, evaluation of system accuracy and intervention effects. Finally, submit the general project report, including product prototypes, data sets and models, patent software lists, etc. At the end of the fifth year, the project results passed the expert acceptance, and it is recommended to enter the next step of large-scale promotion and application.

The above phases ensure that the project progresses step by step, with clear goals and measurable results every year. From the perspective of scientific output, it is expected to publish more than 5 SCI papers (including 2 papers with high impact factor), apply for 3-5 national invention patents, and cultivate several interdisciplinary talents. From the perspective of application transformation, at the end of the project, a complete prototype of the "DIKWP Assessment and Rehabilitation System for Cognitive Impairment" will be delivered, laying the foundation for subsequent productization. Through these milestones, the project will solidly deliver on its research objectives.

6. Assessment indicators and results form

According to the requirements of the guidelines and the characteristics of the project itself, we have formulated the following main assessment indicators and results to evaluate the implementation effect of the project:

• Mechanistic research indicators: elucidate at least 2 or more innovative disease mechanisms or intervention targets, including evidence of their molecular/cellular mechanisms. Specifically, it can be manifested as the publication of more than 2 high-level papers (impact factor>5). These mechanistic findings need to be recognized by peer review and their association with cognitive function described in detail in the project report. The effectiveness of the target can be assessed for reference: whether the intervention of the target can improve cognitive function or alleviate pathology in animal models.

• **Biomarker metrics: Identify several sensitive and specific markers for early diagnosis that can be used to distinguish healthy aging from MCI/**early dementia. The requested marker had an AUC > 0.85 in the project dataset, and the sensitivity and specificity were both >80%. These markers can be molecules (proteins, metabolites) in body fluids, neuroimaging parameters, or numerical behavioral features. The results are in the form of corresponding patents or papers, as well as technical specifications of the testing method.

• DIKWP Cognitive Scoring System: Establish a set of cognitive function assessment scales or scoring systems at the DIKWP level. The reliability and validity of the scale were good: the Cronbach α coefficient of reliability was >0.8, and the correlation coefficient of the criterion correlation with the standard scale was >0.7. The system is tested with no less than 100 samples, and the range of norm and critical values is given. The results are in the form of a scale manual (including instructions for use and explanation guides) and copyright registration. This scoring system will be one of the landmark achievements of the project.

• Performance of intelligent evaluation tools: The AI evaluation tools developed need to reach the practical level. Quantitative indicators include: the detection accuracy of mild cognitive impairment is ≥85%, which is at least 15% higher than that of traditional methods (such as MoCA); The accuracy of discerning functional impairment at different cognitive levels was > 90% by professional evaluation. Tools should also be interpretable, providing hierarchical reporting and descriptions of key indicators. The results are in the form of software prototypes, user guides, and registrations (e.g., filed as Class II medical device software).

• Detection rate of abnormal Wisdom function: Our approach should significantly improve the detection rate of abnormalities in higher-order Wisdom function compared to existing clinical assessments. For example, in patients with early frontotemporal dementia with impaired executive function but normal memory, who may have been missed in the past, our system should be able to identify >80% of such cases. Through clinical trial statistics, the consistency and difference between conventional diagnosis and diagnosis of this system were compared, which proved that this system has advantages in identifying very early or atypical cases. This is used as a measure of the effectiveness of the system.

• Intervention Effect of Artificial Consciousness Assistance System: The effect of quasi-consciousness rehabilitation system was verified in a small clinical trial. The assessment indicators included: the average annual decline rate of cognitive scores in the intervention group decreased by more than 20% compared with the control group, and the ADL score of specific functions (such as ADL scores of activities of daily living) was maintained or improved from baseline, while the control group decreased significantly. At the same time, there was a trend of improvement in the scores of patients' behavioral symptoms, such as apathy scale and depression scale. Although the sample size is limited, it should at least show a positive direction of effect. If possible, further statistical tests showed that the difference was significant (P<0.05). In addition, patient compliance and safety measures should be assessed to ensure no serious adverse events and > 60% of the average daily training participation time. The results were in the form of intervention program sets and clinical observation reports.

• Papers and patent achievements: During the implementation of the project, it is expected to publish more than 5 academic papers (SCI, artificial intelligence, brain science and other fields), of which at least 1-2 representative achievements will be submitted to top journals or conferences (such as Nature, Neurology, AAAI, etc.). Apply for 3-5 national invention patents, including core innovation points such as evaluation algorithms and training systems, and strive to achieve authorization. The above outputs will be used as indicators to test the innovation and academic value of the project.

• Database and platform construction: Establish a DIKWP multimodal database for cognitive impairment, collect and sort out the multimodal data (including clinical scales, sensor data, images, biomarkers, etc.) and corresponding labels generated by this project. The amount of data is not less than 200 subjects, and standardized storage and sharing interfaces are realized. The database will provide a valuable resource for subsequent research and algorithm development. If possible, some datasets can be opened to the public to increase their impact. At the same time, a prototype of a cloud-based platform was developed to modularize the evaluation and training capabilities for later deployment and expansion.

In summary, the assessment indicators cover the scientific indicators such as disease mechanism and marker sensitivity required by the guidelines, and also extend to the cognitive scoring system, intelligent diagnosis and intervention effect characteristic of this project. The form of results includes not only academic outputs such as papers and patents, but also emphasizing the actual systems and data products that can be used. We will strictly monitor the progress of the project according to the above indicators to ensure that the final results are of high quality and meet the standards, and truly achieve a comprehensive breakthrough from mechanism to application.

7. Application and promotion paths

The expected results of this project have important clinical application prospects and industrial transformation value, and we will actively plan the application and promotion path, accelerate the results from the laboratory to the clinic and the market, and benefit the majority of patients.

First of all, in terms of clinical translation, we will rely on cooperative hospitals and medical centers to carry out clinical demonstration and application of project results. In the later stage of the project, it is planned to cooperate with representative medical institutions inside and outside the province (such as geriatric departments and memory clinics of large tertiary hospitals) to establish a DIKWP assessment and rehabilitation demonstration center for cognitive impairment. Deploy our assessment systems and rehabilitation equipment in demonstration centers, used by professional doctors and therapists, to detect and train cognitively impaired patients who come to the clinic or rehabilitate. This will test the performance of our system in a real-world medical environment on the one hand, and increase clinical awareness through a demonstration effect on the other. We will develop clinical manuals and training courses to train healthcare professionals to ensure they master the operation of the system and the interpretation of data. The effects of the demonstration application (e.g., improving the early diagnosis rate, prolonging the maintenance time of patient function, etc.) will be counted through case follow-up. If the verification results are excellent, we will recommend the DIKWP assessment system to the national health authorities and academic organizations to include the DIKWP assessment system in the diagnosis and treatment guidelines or expert consensus for cognitive impairment, and promote it to become a standardized assessment tool.

Secondly, in terms of productization and industrial promotion, we plan to package the project technology into a digital medical product. Specifically, it includes: "Cognitive Impairment Hierarchical Assessment System" software and "Quasi-Consciousness Cognitive Training Platform" hardware. We will improve the security and stability testing of the software in accordance with the requirements of medical device regulations, and strive to apply for medical device registration after the completion of the project, so that it can obtain legal sales qualifications. At the same time, contact powerful medical AI companies or rehabilitation equipment manufacturers to discuss cooperation and transformation models. A possible model is to invest in the technology to co-build a start-up, or to license the patent to a large company for production and promotion. For example, the evaluation software can interface with the existing electronic medical record system or health management app in China, and embed it into the existing medical IT platform to achieve large-scale deployment. The rehabilitation training platform can cooperate with professional elderly care and rehabilitation equipment manufacturers, who will produce integrated equipment with VR and biofeedback functions according to our design, and roll it out in nursing homes, community rehabilitation stations, etc.

Thirdly, we attach great importance to the connection with the Wisdom pension and public health platform. China is facing the challenge of aging, and the state has issued policies such as the Action Plan for the Prevention and Treatment of Alzheimer's. Our results are aligned with this national need and can be used as an important part of Wisdom's retirement efforts. In the roll-out phase, we will work with aged care service providers, community health centers, etc., to roll out the assessment and training system to communities and families. For example, a geriatric cognitive screening program in collaboration with community health services, using our assessment tools to regularly check the cognitive function of the elderly, will help to detect problems early and refer them for timely intervention. Deploy training systems in nursing homes or day care centres to provide daily cognitive activities for seniors with mild cognitive impairment. In addition, we can build a remote assessment and rehabilitation platform and develop a mobile application, so that some elderly people who are unable to go out can also use a simplified version of assessment and training at home (of course, there are professionals to guide them remotely). Combined with the Internet of Things technology, the data can be uploaded to the cloud for real-time monitoring by doctors. This model can reduce the cost of promotion and expand the reach.

In terms of publicity and education, we will raise the awareness and acceptance of DIKWP cognitive diagnosis and treatment through various channels. This includes sharing project results at academic conferences and continuing classes, and conducting training lectures for clinicians; For the general public and patients' families, the advantages of this new assessment and rehabilitation method are introduced through popular science articles, media reports, etc. Raise the public's awareness of early screening and active cognitive training, and create a good social environment for promotion and application.

Finally, considering the sustainable development of the results, we plan to continue to promote the construction of an interdisciplinary collaboration platform after the end of the project. Strive to apply for the establishment of the "Cognitive Impairment Artificial Consciousness Research Center" or provincial and ministerial key laboratories, based on the results of this project, gather experts in related fields, further improve the technology and develop new applications. For example, the DIKWP assessment system will be extended to the cognitive assessment of other brain diseases (Parkinson's, depression, etc.), or the combination of quasi-consciousness rehabilitation and robotic care will be developed to develop intelligent companion robots. These follow-up studies and applications will become more feasible through the successful demonstration of the results of the project.

In short, the promotion path of the results of this project is clear, and the influence will be amplified step by step through the steps of clinical demonstration, productization cooperation, and community integration. It is expected that within 3-5 years after the completion of the project, our DIKWP assessment tool is expected to be applied in many memory clinics in China, and the quasi-consciousness rehabilitation system will become a characteristic project of some elderly care institutions, gradually forming a new format of "artificial consciousness + cognitive health". In the longer term, if the promotion is successful, it will significantly improve the early diagnosis rate of cognitive impairment diseases and the quality of life of patients in China, reduce the burden of care, and at the same time give birth to a new industrial growth point of artificial intelligence in the field of medical rehabilitation, with significant social and economic benefits.

The above application and promotion plan ensures that the project results can be truly transformed from theoretical research to practical technology, serve the huge group of patients with cognitive impairment in China, and realize the social value and innovation leading role of major national interdisciplinary research projects.