Alzheimer's disease early screening method and system based on eye tracking

Abstract

The application belongs to the technical field of medical information and neuroscientific technology, and particularly relates to an early screening method and system for Alzheimer's disease based on eye movement tracking, comprising the following steps: collecting human eye movement data of a subject during the execution of a cognitive task; constructing a personalized eye movement dynamics model based on the human eye movement data, extracting an eye movement behavior feature vector; utilizing a neural-eye movement isomorphic mapping relationship; identifying a neurodegenerative attenuation feature related to Alzheimer's disease; and detecting whether an epilepsy-like discharge feature exists in the virtual neural function signal. Through the construction of the neural-eye movement isomorphic mapping relationship, the virtual neural function signal of the deep brain structure can be reconstructed by using the eye movement data, the screening cost is effectively reduced, the evaluation accuracy of the function state of the hippocampus and amygdala is improved, and through the comorbidity correction of the epilepsy-like discharge feature, the accuracy and reliability of the early screening of Alzheimer's disease are significantly improved.

Description

Technical Field

[0001] This invention belongs to the fields of medical information technology and neuroscience technology, specifically relating to an eye-tracking-based method and system for early screening of Alzheimer's disease. Background Technology

[0002] Alzheimer's disease (AD) is a progressive neurodegenerative disease. Early screening is crucial for slowing disease progression and intervention. Currently, commonly used screening methods include neuropsychological scales, cerebrospinal fluid biomarker testing, and positron emission tomography (PET). However, scale testing is highly subjective and easily affected by the education level and cooperation of the subjects. PET and cerebrospinal fluid testing are invasive, costly, and pose radiation risks, making them difficult to popularize in primary hospitals or large populations. In addition, some AD patients often have comorbidities such as epileptiform discharges, which traditional screening methods often overlook, resulting in insufficient diagnostic specificity and failing to meet the clinical needs for early and accurate screening.

[0003] In recent years, non-invasive screening technologies based on physiological signals have gradually attracted attention. Among them, eye-tracking technology has been attempted for cognitive assessment due to its convenience. It usually adopts a multimodal fusion scheme that combines eye trackers with electroencephalography (EEG) equipment to try to infer brain function through surface electrical signals. However, surface EEG signals have low spatial resolution and cannot directly reflect the activity status of deep brain structures such as the hippocampus and amygdala. At the same time, existing schemes are mostly simple data superposition at the hardware level and lack a deep mapping mechanism from eye movement behavior to deep brain neural activity. This makes it impossible to reconstruct the functional signals of deep brain regions without wearing complex EEG equipment, which limits its widespread application in routine physical examinations and preoperative surgical assessments.

[0004] Problems with existing technology: First, it relies on multiple devices and hardware, increasing the burden on test subjects and the cost of equipment maintenance. Second, it cannot effectively reconstruct the functional state of deep brain structures (such as the hippocampus and amygdala) through a single modality, resulting in insufficient spatial positioning accuracy. Third, for the complex situation of AD comorbid epilepsy, it lacks a correction mechanism based on neurophysiological characteristics, which can easily lead to misdiagnosis or missed diagnosis, making it difficult to meet the clinical needs of high-precision early screening. Summary of the Invention

[0005] The purpose of this invention is to provide an eye-tracking-based method and system for early screening of Alzheimer's disease. By constructing a neural-eye-tracking isomorphic mapping relationship, virtual neural function signals of deep brain structures can be reconstructed using eye-tracking data, effectively reducing screening costs, improving the accuracy of assessing the functional status of the hippocampus and amygdala, and significantly improving the accuracy and reliability of early screening of Alzheimer's disease through comorbid correction of epileptiform discharge characteristics.

[0006] The specific technical solution adopted by this invention is as follows: An eye-tracking-based method for early screening of Alzheimer's disease includes the following steps: Collect eye movement data of subjects during the performance of cognitive tasks; A personalized eye movement dynamics model is constructed based on the aforementioned human eye movement data, and eye movement behavior feature vectors are extracted. By utilizing the neural-eye movement isomorphic mapping relationship, the eye movement behavior feature vector is mapped to the target brain structure space, and the virtual neural function signal of the target brain structure is reconstructed. Identify Alzheimer's disease-related neurodegenerative attenuation features in the virtual neural function signals; The presence of epileptiform discharge features in the virtual neural function signal is detected, and the Alzheimer's disease screening results are corrected based on the coexistence relationship between the epileptiform discharge features and the neurodegenerative attenuation features.

[0007] According to another aspect of the present invention, the mapping to the target brain structure space includes: The eye-tracking behavior feature vector is input into a pre-trained depth mapping network, which outputs virtual time-series signals corresponding to the spatial locations of the amygdala and hippocampus.

[0008] According to another aspect of the present invention, the correction of the Alzheimer's disease screening result includes: If the epileptiform discharge features originate from the hippocampus region in the target brain structure space and are positively correlated with the neurodegenerative attenuation features, then it is determined to be an epileptic state comorbid with Alzheimer's disease. If the epileptiform discharge features are widely distributed and not correlated with the neurodegenerative attenuation features, then it is determined to be primary epileptiform interference.

[0009] According to another aspect of the present invention, the method further includes the following steps: The associated weight data in the neural-eye movement isomorphic mapping relationship is iteratively updated based on subsequent diagnostic results to achieve adaptive evolution of the mapping model.

[0010] According to another aspect of the present invention, an eye-tracking-based early screening system for Alzheimer's disease is also provided, comprising: The eye-tracking data acquisition module is used to collect the subject's eye movement data; An eye-tracking modeling unit, connected to the eye-tracking data acquisition module, is used to construct a personalized eye-tracking dynamics model for the subject based on the human eye movement data; The brain structure mapping and reconstruction unit is connected to the eye movement modeling unit and is used to map the personalized eye movement dynamics model to the target brain structure space based on a preset neural-eye movement isomorphic mapping relationship, and reconstruct the virtual neural function signal of the target brain structure. The screening and assessment unit, connected to the brain structure mapping and reconstruction unit, is used to analyze the pathological features in the virtual neural function signals and generate Alzheimer's disease screening results. The target brain structure space includes at least the amygdala and hippocampus, and the pathological features include neurodegenerative attenuation features and epileptiform discharge features.

[0011] According to another aspect of the present invention, the neural-eye movement isomorphic mapping relationship is established through training a machine learning model, wherein the training samples of the machine learning model include synchronously collected human eye movement data and corresponding brain imaging functional data.

[0012] According to another aspect of the present invention, the virtual neural function signal includes the neural oscillation power spectrum characteristics and neural connectivity strength characteristics of the target brain structure.

[0013] According to another aspect of the present invention, a visualization output module is also included, which is used to display the virtual neural function signals of the target brain structure space superimposed on a standard brain atlas in the form of a three-dimensional heat map.

[0014] According to another aspect of the present invention, an eye-tracking-based early screening device for Alzheimer's disease is also included, characterized in that it comprises: A processor used to execute computer programs to implement an eye-tracking-based early screening method for Alzheimer's disease; A memory, connected to the processor, is used to store the computer program and the neural-eye movement isomorphic mapping database; An input interface, connected to the processor, is used to receive human eye movement data transmitted from an external eye-tracking device; The neural-eye movement isomorphic mapping database stores the correlation weight data between eye movement behavior features and brain structural and functional states.

[0015] According to another aspect of the present invention, a secure communication module is also included, which is used to encrypt and transmit the screening results to the hospital information system and record data access logs to meet medical data compliance requirements.

[0016] According to another aspect of the present invention, an electronic device is also provided, the electronic device including a memory and a processor; the memory is used to store a program; the processor executes the program to implement the method described in any one of the foregoing.

[0017] According to another aspect of the present invention, a computer-readable storage medium is also provided, the storage medium storing a computer program that, when executed by a processor, implements the method described in any one of the preceding embodiments.

[0018] According to another aspect of the present invention, a computer program product is also provided, including a computer program that, when executed by a processor, implements the method described in any one of the preceding embodiments.

[0019] The technical effects achieved by this invention are as follows: This invention constructs a neural-eye movement isomorphic mapping relationship, enabling the reconstruction of virtual neural function signals of deep brain structures using eye movement data. Utilizing the deep coupling mechanism between eye movement behavior and brain neural activity, and through a personalized eye movement dynamics model and deep mapping network, surface eye movement features are transformed into virtual signals in the medial temporal lobe region. This not only reduces the complexity and cost of screening equipment, making large-scale early screening possible, but also effectively overcomes the shortcomings of low spatial resolution of surface EEG signals and their inability to directly reflect deep brain region activity. Through precise assessment of the functional status of the amygdala and hippocampus, it significantly improves the sensitivity and specificity of early Alzheimer's disease identification, providing a novel, non-invasive, efficient, and easily deployable screening method for clinical use, and has significant clinical application value.

[0020] This invention improves screening accuracy, particularly for the complex situation of Alzheimer's disease comorbid epilepsy. By detecting epileptiform discharge characteristics in virtual neural function signals and their coexistence with neurodegenerative attenuation characteristics, the system can automatically correct screening results, effectively distinguishing between primary epilepsy interference and AD comorbidity, avoiding misdiagnosis or missed diagnosis due to neglecting neurophysiological abnormalities. Based on the identification method of virtual signal spatial distribution, it can provide diagnostic clues without long-term EEG monitoring, greatly shortening the diagnostic path. In addition, the model adaptively evolves based on the confirmed diagnosis results, continuously optimizing the mapping weights as clinical data accumulates, ensuring accuracy for long-term use. Attached Figure Description

[0021] Figure 1 This is a flowchart of the eye-tracking-based early screening method for Alzheimer's disease according to the present invention; Figure 2 This is a schematic diagram of the structure of the eye-tracking-based early screening system for Alzheimer's disease according to the present invention. Detailed Implementation

[0022] To make the objectives and advantages of this invention clearer, the invention will be specifically described below with reference to embodiments. It should be understood that the following text is merely used to describe one or more specific embodiments of the invention and does not strictly limit the scope of protection specifically claimed by the invention.

[0023] It should be noted that the terms "first," "second," etc., used in this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0024] According to embodiments of the present invention, a method embodiment for early screening of Alzheimer's disease based on eye tracking is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0025] like Figure 1 As shown, the eye-tracking-based early screening method for Alzheimer's disease includes the following steps: S1. Collect eye movement data of the subjects during the cognitive task; S2. Construct a personalized eye-tracking dynamics model based on human eye movement data and extract eye-tracking behavior feature vectors; S3. Using the neural-eye movement isomorphic mapping relationship, the eye movement behavior feature vector is mapped to the target brain structure space to reconstruct the virtual neural function signal of the target brain structure; S4. Identify Alzheimer's disease-related neurodegenerative attenuation features in virtual neural function signals; S5. Detect whether epileptiform discharge features exist in virtual neural function signals, and correct the Alzheimer's disease screening results based on the coexistence relationship between epileptiform discharge features and neurodegenerative attenuation features.

[0026] The acquisition of human eye movement data in step S1 is achieved through a non-contact infrared eye tracker. The subject sits in front of the display screen, and the cognitive tasks include visual memory tasks, emotional face recognition tasks, or attention maintenance tasks. The raw data formats acquired include fixation point coordinate sequences, pupil diameter change sequences, and blink timestamps.

[0027] The personalized eye-tracking dynamics model in step S2 is constructed by analyzing the smoothness of the subject's eye movement trajectory, peak saccadic velocity, and fixation stability during the task. The extracted eye-tracking behavior feature vector is a multidimensional array containing temporal features (such as average fixation duration) and frequency domain features (such as nystagmus frequency).

[0028] The neural-eye movement isomorphic mapping relationship in step S3 is a pre-trained transformation function. The target brain structure space refers to the three-dimensional anatomical space of the brain, with a focus on the medial temporal lobe. The virtual neural function signal is not a physical electrical signal obtained directly by electrodes, but a digital signal sequence that reflects the functional state of the brain structure and is inferred by a computational model based on eye movement behavior characteristics.

[0029] The neurodegenerative attenuation characteristics in step S4 are manifested as a decrease in energy in specific frequency bands of the virtual signal or a decrease in neural connection efficiency.

[0030] The correction process in step S5 is a logical judgment process in which the system compares the spatiotemporal overlap between the detected epileptiform features and AD pathological features.

[0031] Based on the steps described above, eye movements are jointly regulated by the brainstem, thalamus, and subcortical structures (including the amygdala and hippocampus). Neurodegenerative changes caused by Alzheimer's disease disrupt these neural circuits, leaving specific dynamic imprints on eye movement behavior. These imprints are then used to infer brain state through mathematical mapping via the coupling mechanism between the brain and eye. Simultaneously, hippocampal sclerosis in AD patients easily induces epileptiform discharges. These discharge signals affect the eye movement control center through neural networks, leaving implicit features in eye movement data. These features can be identified after mapping and reconstruction. Therefore, deep brain structure function can be assessed without wearing an EEG cap, lowering the screening threshold and reducing discomfort for subjects. Furthermore, it effectively distinguishes between simple AD and AD comorbid epilepsy, avoiding misdiagnosis of AD due to epilepsy interference.

[0032] Furthermore, traditional techniques believe that eye movements can only reflect cortical function. Through higher-order dynamic modeling, eye movement data contains information sufficient to reconstruct the functional state of the limbic system (amygdala, hippocampus), enabling a single-modal device to achieve the diagnostic efficacy of a multimodal device and eliminating the technical errors caused by multi-device synchronization.

[0033] As an optional embodiment, mapping to the target brain structure space includes: inputting eye movement behavior feature vectors into a pre-trained deep mapping network and outputting virtual time-series signals corresponding to the spatial locations of the amygdala and hippocampus.

[0034] In this embodiment, the pre-trained deep mapping network adopts an encoder-decoder architecture. The encoder part is used to compress eye-tracking behavior feature vectors and extract high-order latent features; the decoder part is used to expand the latent features into time-series signals corresponding to the brain structure space.

[0035] The spatial locations of the amygdala and hippocampus are coordinate regions defined based on standard brain atlases (such as the MNI space). The output virtual time-series signal simulates the morphology of local field potentials (LFPs) in terms of data structure, including amplitude, phase, and frequency information.

[0036] The mapping process can be represented as: in, Indicates time Virtual neural function signals, Indicates time The feature vector of eye movement behavior. This represents the nonlinear transformation function of a deep mapping network. This represents the network weight parameters.

[0037] Furthermore, deep neural networks possess nonlinear fitting capabilities, enabling them to learn complex, nonlinear relationships between eye movement features and brain neural activity. Through training, deep neural networks learn how to translate minute changes in eye movements into the intensity of brain region activity.

[0038] Furthermore, the use of deep mapping networks improves the accuracy and robustness of the reconstructed signal, enabling it to adapt to individual differences among different subjects. The output time-series signal allows for subsequent frequency domain and time-series analysis, enriching the diagnostic evidence.

[0039] Deep mapping networks not only achieve signal reconstruction, but also have a certain denoising function. During the mapping process, the network automatically filters eye movement noise (such as involuntary blink artifacts) that is unrelated to brain function, so that the signal-to-noise ratio of the output virtual signal is better than that of the directly acquired surface EEG signal.

[0040] As an optional implementation, correcting Alzheimer's disease screening results includes: If the epileptiform discharge features originate from the hippocampus region in the target brain structure and are positively correlated with neurodegenerative attenuation features, then it is determined to be an epileptic state comorbid with Alzheimer's disease. If the epileptiform discharge features are widely distributed and not correlated with neurodegenerative attenuation features, it is determined to be primary epileptiform interference.

[0041] Furthermore, the identification of epileptiform discharge features is achieved by detecting spikes or sharp waves in the virtual time-series signal.

[0042] The origin of the hippocampus region is determined by back projection in virtual space using a source localization algorithm.

[0043] Positive correlation means that as the degree of neurodegenerative decline deepens, the frequency or amplitude of epileptiform discharges also tends to increase.

[0044] Widespread distribution refers to the diffuse distribution of discharge characteristics in the virtual brain space, without a clear focal point. The judgment logic can be expressed as: The pathological changes in Alzheimer's disease often begin in the medial temporal lobe (hippocampus), leading to abnormal neuronal excitability and a predisposition to focal epileptic discharges. Primary epilepsy, on the other hand, often involves a wider range of neural networks. This pathophysiological difference can be used to differentiate the diagnosis by examining the spatial distribution and correlation of signals.

[0045] Furthermore, it can significantly improve the specificity of diagnosis. For patients with comorbidities, the results suggest that clinicians should consider antiepileptic treatment concurrently and optimize the treatment plan. For primary epilepsy interference, it avoids misdiagnosing epilepsy patients as Alzheimer's disease (AD).

[0046] Furthermore, the spatial distribution characteristics of virtual signals can be used to identify epilepsy types, which traditionally requires long-term video EEG monitoring. This embodiment, however, can provide such identification information during the screening stage, significantly shortening the diagnostic process.

[0047] As an optional embodiment, the eye-tracking-based early screening method for Alzheimer's disease further includes the following steps: The associated weight data in the neuro-eye movement isomorphic mapping relationship is iteratively updated based on subsequent diagnostic results to achieve adaptive evolution of the mapping model.

[0048] According to the above steps, once a subject is diagnosed by the clinical gold standard (such as PET scan or cerebrospinal fluid test), the diagnosis label will be sent back to the system.

[0049] The system uses the confirmed case label as a monitoring signal to calculate the loss function between the screening results and the confirmed case results.

[0050] The backpropagation algorithm is used to fine-tune the association weights in the neural-eye-tracking isomorphic mapping. The update formula is illustrated below: in, For learning rate, This is the gradient of the loss function.

[0051] As clinical data accumulates, the mapping model can continuously correct biases and adapt to different population characteristics or device drift. This is an online learning or incremental learning mechanism.

[0052] Furthermore, it can ensure the long-term accuracy and adaptability of the system. As the hospital uses the system for longer, its screening accuracy will become higher and higher, forming a data closed loop.

[0053] Furthermore, the system is equipped with self-learning capabilities and can adapt to regional or ethnic differences among different hospital populations, thus solving the problem of poor performance of general models in specific hospitals.

[0054] As an optional embodiment, an eye-tracking-based early screening system for Alzheimer's disease, used to perform the aforementioned method, is characterized by comprising: The eye-tracking data acquisition module is used to collect the subject's eye movement data; The eye-tracking modeling unit, connected to the eye-tracking data acquisition module, is used to construct a personalized eye-tracking dynamics model for the subject based on human eye movement data. The brain structure mapping and reconstruction unit is connected to the eye movement modeling unit. It is used to map a personalized eye movement dynamics model to the target brain structure space based on a preset neural-eye movement isomorphic mapping relationship, and reconstruct the virtual neural function signal of the target brain structure. The screening and assessment unit, connected to the brain structure mapping and reconstruction unit, is used to analyze pathological features in virtual neural function signals and generate Alzheimer's disease screening results. The target brain structure space includes at least the amygdala and hippocampus, and the pathological features include neurodegenerative attenuation features and epileptiform discharge features.

[0055] This system can be deployed on servers or workstations within the hospital.

[0056] Based on the above, the eye-tracking data acquisition module can be a standalone desktop eye tracker or an embedded sensor integrated into the screening terminal.

[0057] The eye-tracking modeling unit, brain structure mapping and reconstruction unit, and screening and assessment unit can be software functional modules that run on a central processing unit.

[0058] The target brain structure space is stored in the form of a three-dimensional coordinate grid within the system.

[0059] Virtual neural function signals are transmitted between units within the system in the form of digital signal streams.

[0060] The principle behind this is that modular design decouples the functions of different parts of the system, making maintenance and upgrades easier. For example, when the mapping algorithm is optimized, only the brain structure mapping and reconstruction unit needs to be updated, without replacing the hardware.

[0061] Furthermore, a complete hardware and software integrated architecture ensures the feasibility of the method. A systematic processing flow guarantees the efficiency and standardization of the screening process.

[0062] This system architecture enables distributed deployment. Eye-tracking data acquisition can be performed in wards or outpatient terminals, while complex mapping reconstruction and evaluation can be completed on the hospital's central server, thus optimizing the allocation of medical resources.

[0063] As an optional embodiment, the neural-eye movement isomorphic mapping relationship is established through training a machine learning model. The training samples of the machine learning model include synchronously collected eye movement data of the population and corresponding brain imaging functional data.

[0064] In this embodiment, the process of establishing the mapping relationship is an offline training process.

[0065] By collecting a sample of volunteers, the samples were simultaneously subjected to high-precision eye-tracking and functional brain imaging (such as functional magnetic resonance imaging fMRI or magnetoencephalography MEG).

[0066] Eye-tracking data was used as input, and activation signals from the amygdala and hippocampus in brain imaging data were used as labels (Ground Truth).

[0067] Train a machine learning model to minimize the difference between the reconstructed signal and the real imaging signal.

[0068] Brain imaging functional data are standardized to eliminate individual differences in anatomical structure.

[0069] Furthermore, by using multimodal data as teacher signals, the model is taught how to infer brain activity from eye movements. Once training is complete, only eye movement data is needed for actual screening.

[0070] Furthermore, through the training methods described above, the system actually learns the common neural-eye movement coupling patterns of the population, enabling the system to have a certain generalization ability and be applied to new subjects who have not participated in the training.

[0071] As an optional embodiment, the virtual neural function signal includes the neural oscillation power spectrum characteristics and neural connectivity strength characteristics of the target brain structure.

[0072] In this embodiment, the system performs Fast Fourier Transform (FFT) or Wavelet Transform on the reconstructed virtual time series signal to obtain the neural oscillation power spectrum.

[0073] The neural connectivity strength characteristics are obtained by calculating the coherence or phase synchronization of virtual signals between the amygdala and the hippocampus.

[0074] The electroencephalogram (EEG) characteristics of AD patients typically show increased slow-wave power, decreased fast-wave power, and weakened functional connectivity between brain regions. These characteristics are encoded in virtual signals.

[0075] Furthermore, multi-dimensional assessment indicators not only consider signal strength but also the brain regions' ability to work together, comprehensively reflecting the state of brain function.

[0076] Furthermore, even without direct EEG acquisition, the reconstructed power spectrum features can still sensitively capture subtle neural oscillation abnormalities in the early stages of Alzheimer's disease (AD), with sensitivity superior to traditional scale scores in specific frequency bands.

[0077] As an optional embodiment, the eye-tracking-based early screening system for Alzheimer's disease also includes a visualization output module for displaying virtual neural function signals of the target brain structure space as a three-dimensional heat map superimposed on a standard brain atlas.

[0078] In this embodiment, the visualization output module is connected to the doctor's display terminal.

[0079] Three-dimensional heatmaps use color intensity to represent the strength or degree of abnormality of virtual neural function signals (e.g., red indicates high risk of abnormality, and blue indicates normal).

[0080] A standard brain atlas can be an MNI152 template or an individual patient's MRI structural image.

[0081] Doctors can visually observe the signal status of the amygdala and hippocampus by rotating and zooming the view.

[0082] Furthermore, abstract data is transformed into intuitive images that align with doctors' reading habits and assist in clinical decision-making.

[0083] Furthermore, it improves the interpretability of screening results, allowing doctors to see the problems intuitively, increasing their trust in the system's results. Visualization also enables non-neurologists (such as hepatobiliary and pancreatic surgeons) to quickly understand the patient's cognitive risk status, facilitating interdisciplinary collaboration and preoperative risk assessment.

[0084] As an optional embodiment, the eye-tracking-based Alzheimer's disease early screening system further includes an eye-tracking-based Alzheimer's disease early screening device, comprising: A processor used to execute computer programs to implement an eye-tracking-based early screening method for Alzheimer's disease; The memory, connected to the processor, is used to store computer programs and a neural-eye movement isomorphic mapping database; The input interface, connected to the processor, is used to receive human eye movement data transmitted from an external eye-tracking device; The neural-eye movement isomorphic mapping database stores the correlation weight data between eye movement behavior features and brain structural and functional states.

[0085] The device in this embodiment can be a standalone industrial computer or an embedded box.

[0086] The processor uses a multi-core CPU or a GPU with neural network acceleration to meet the real-time computing requirements of deep mapping networks.

[0087] The memory includes cache and persistent storage, which are used to store temporary computational data and long-term model data, respectively.

[0088] The input interface supports USB, Bluetooth, or network protocols, and is compatible with eye-tracking devices from multiple brands.

[0089] The associated weight data is stored in encrypted form to prevent model parameter leakage.

[0090] Furthermore, hardware infrastructure ensures the algorithm runs efficiently and stably. A dedicated database improves data retrieval efficiency, and the integrated hardware and software design facilitates deployment across various hospital departments, offering strong compatibility and protecting the hospital's existing investment in eye-tracking equipment.

[0091] As an optional embodiment, the eye-tracking-based early screening system for Alzheimer's disease also includes a secure communication module for encrypting and transmitting screening results to the hospital information system and recording data access logs to meet medical data compliance requirements.

[0092] In this embodiment, the secure communication module uses an encryption protocol that conforms to medical industry standards (such as HL7 overTLS).

[0093] The screening results are anonymized before transmission, removing direct identification of the subjects and retaining only the necessary medical ID.

[0094] The data access log records who accessed which screening data, when, and which data point, ensuring traceability.

[0095] The system interfaces with the hospital information system (HIS) or electronic medical record system (EMR) to automatically archive screening reports into the patient's medical record.

[0096] Furthermore, medical data involves patient privacy and must comply with laws and regulations (such as the Personal Information Protection Law and the Regulations on Medical Data Security Management). Encryption and logging are standard security measures used to protect patient privacy and meet the compliance requirements of hospital information system construction. Automated archiving reduces the workload of doctors' manual data entry and avoids transcription errors.

[0097] Furthermore, by integrating with enterprise information systems, screening data can serve as part of patients' long-term health records, providing a structured, high-quality data source for subsequent retrospective research.

[0098] It should be noted that the above modules can be implemented by software or hardware. For the latter, they can be implemented in the following ways, but are not limited to: all the above modules are located in the same processor; or, the above modules are located in different processors in any combination.

[0099] According to another aspect of the present invention, an electronic device is also provided, the electronic device including a memory and a processor; the memory is used to store a program; the processor executes the program to implement the method of any of the foregoing.

[0100] According to another aspect of the present invention, a computer-readable storage medium is also provided, the storage medium storing a computer program that, when executed by a processor, implements the method of any of the foregoing.

[0101] According to another aspect of the present invention, a computer program product is also provided, including a computer program that, when executed by a processor, implements the method described in any of the foregoing.

[0102] The above description is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention. Structures, devices, and operating methods not specifically described or explained in this invention are implemented according to conventional methods in the art unless otherwise specified or limited.

Claims

1. An eye-tracking-based method for early screening of Alzheimer's disease, characterized in that, Includes the following steps: Collect eye movement data of subjects during the performance of cognitive tasks; A personalized eye movement dynamics model is constructed based on the aforementioned human eye movement data, and eye movement behavior feature vectors are extracted. By utilizing the neural-eye movement isomorphic mapping relationship, the eye movement behavior feature vector is mapped to the target brain structure space, and the virtual neural function signal of the target brain structure is reconstructed. Identify Alzheimer's disease-related neurodegenerative attenuation features in the virtual neural function signals; The presence of epileptiform discharge features in the virtual neural function signal is detected, and the Alzheimer's disease screening results are corrected based on the coexistence relationship between the epileptiform discharge features and the neurodegenerative attenuation features.

2. The eye-tracking-based early screening method for Alzheimer's disease according to claim 1, characterized in that, The mapping to the target brain structure space includes: The eye-tracking behavior feature vector is input into a pre-trained depth mapping network, which outputs virtual time-series signals corresponding to the spatial locations of the amygdala and hippocampus.

3. The eye-tracking-based early screening method for Alzheimer's disease according to claim 1, characterized in that, The correction of the Alzheimer's disease screening results includes: If the epileptiform discharge features originate from the hippocampus region in the target brain structure space and are positively correlated with the neurodegenerative attenuation features, then it is determined to be an epileptic state comorbid with Alzheimer's disease. If the epileptiform discharge features are widely distributed and not correlated with the neurodegenerative attenuation features, then it is determined to be primary epileptiform interference.

4. The eye-tracking-based early screening method for Alzheimer's disease according to claim 1, characterized in that, It also includes the following steps: The associated weight data in the neural-eye movement isomorphic mapping relationship is iteratively updated based on subsequent diagnostic results to achieve adaptive evolution of the mapping model.

5. An eye-tracking-based early screening system for Alzheimer's disease, used to perform the method as described in any one of claims 1-4, characterized in that, include: The eye-tracking data acquisition module is used to collect the subject's eye movement data; An eye-tracking modeling unit, connected to the eye-tracking data acquisition module, is used to construct a personalized eye-tracking dynamics model for the subject based on the human eye movement data; The brain structure mapping and reconstruction unit is connected to the eye movement modeling unit and is used to map the personalized eye movement dynamics model to the target brain structure space based on a preset neural-eye movement isomorphic mapping relationship, and reconstruct the virtual neural function signal of the target brain structure. The screening and assessment unit, connected to the brain structure mapping and reconstruction unit, is used to analyze the pathological features in the virtual neural function signals and generate Alzheimer's disease screening results. The target brain structure space includes at least the amygdala and hippocampus, and the pathological features include neurodegenerative attenuation features and epileptiform discharge features.

6. The eye-tracking-based early screening system for Alzheimer's disease according to claim 5, characterized in that: The neural-eye movement isomorphic mapping relationship is established through training a machine learning model. The training samples of the machine learning model include synchronously collected eye movement data of the population and corresponding brain imaging functional data.

7. The eye-tracking-based early screening system for Alzheimer's disease according to claim 5, characterized in that: The virtual neural function signal includes the neural oscillation power spectrum characteristics and neural connection strength characteristics of the target brain structure.

8. The eye-tracking-based early screening system for Alzheimer's disease according to claim 5, characterized in that: It also includes a visualization output module, which is used to display the virtual neural function signals of the target brain structure space as a three-dimensional heat map superimposed on a standard brain atlas.

9. The eye-tracking-based early screening system for Alzheimer's disease according to claim 5 further includes an eye-tracking-based early screening device for Alzheimer's disease, characterized in that, include: A processor used to execute computer programs to implement an eye-tracking-based early screening method for Alzheimer's disease; A memory, connected to the processor, is used to store the computer program and the neural-eye movement isomorphic mapping database; An input interface, connected to the processor, is used to receive human eye movement data transmitted from an external eye-tracking device; The neural-eye movement isomorphic mapping database stores the correlation weight data between eye movement behavior features and brain structural and functional states.

10. The eye-tracking-based early screening system for Alzheimer's disease according to claim 9, characterized in that, It also includes a secure communication module for encrypting and transmitting the screening results to the hospital information system and recording data access logs to meet medical data compliance requirements.

Images