Cognitive assessment system

Abstract

A cognitive assessment system includes a computing device, a server coupled to the computing device, and an eye tracking device coupled to the computing device. The computing device includes a cognitive assessment program configured to perform at least one cognitive assessment task to assess cognitive function of a user. The server includes a database. The database stores historical performance data of the user, a historical performance model of the user, task performance data of a healthy population, and task performance data of a patient population. The eye tracking device is configured to capture eye movement information of the user.

Description

Technical Field

[0001] This invention relates to a cognitive assessment system, and more particularly to a cognitive assessment system using an eye-tracking device. Background Technology

[0002] Cognitive abilities tend to decline with age. Statistics show that one in twelve people over the age of 65 suffers from dementia, and this number continues to rise. The rapidly increasing proportion of people with cognitive impairments places a heavy burden on society in terms of care. Therefore, effectively detecting cognitive decline and maintaining the cognitive health of older individuals is a key focus of numerous research efforts.

[0003] Cognitive assessment refers to evaluating an individual's cognitive abilities through various methods and tools such as interviews, questionnaires, and tests. Traditional cognitive function assessments are mostly conducted in the form of interviews and paper-and-pencil questionnaires, such as the Montreal Cognitive Assessment (MoCA) and the Mini-Mental State Examination (MMSE). The advantage of these tests is that they provide results quickly. However, paper-and-pencil tests are not suitable for repeated testing on the same individual within a short period. This limitation makes traditional paper-and-pencil tests unsuitable for long-term, continuous assessment of changes in an individual's cognitive abilities. To address these limitations, an increasing number of computer-based assessment tests have emerged. The advantage of computer-based assessment tests is that their execution is not limited by time or space. Users can conduct cognitive function assessments using their own computer devices, such as tablets, smartphones, and personal computers. Assessments can also be conducted at home or in a laboratory setting. Furthermore, computer-based assessment tests allow for greater flexibility in the types of questions asked, resulting in more accurate results.

[0004] With recent advancements in various sensing technologies, sensors can easily measure heart rate, respiratory rate, skin conductance, and even brain waves and eye movements. Among these, eye movements are the most suitable and widely discussed indicator for assessing cognitive function. This is because they not only reflect an individual's attentional direction but also provide dynamic data with high temporal and spatial resolution. Compared to traditional assessments, eye movements can more effectively measure an individual's behavioral decision-making process and provide a more accurate assessment of cognitive abilities.

[0005] Many eye-tracking-based cognitive assessment systems exist today. However, these systems are still under development and face numerous limitations. In particular, the full potential of eye-tracking assessment has not yet been realized. Most eye-tracking-based cognitive assessment systems employ a single task, making it difficult to distinguish changes in different cognitive functions over time. Furthermore, existing eye-tracking-based cognitive assessments largely detect cognitive abnormalities by comparing a user's single test results with norms from patient or healthy groups, lacking the ability to monitor and predict individual cognitive changes over time. Summary of the Invention

[0006] An embodiment provides a cognitive assessment system, including a computing device, a server, and an eye-tracking device. The computing device includes a cognitive assessment program for performing at least one cognitive assessment task. The at least one cognitive assessment task is used to assess a user's cognitive function. The server is coupled to the computing device and includes a database. The database stores the user's historical performance data, the user's historical performance model, performance data of a healthy group, and performance data of a patient group. The eye-tracking device is coupled to the computing device for capturing the user's eye movement information. The user performs at least one cognitive assessment task executed by the cognitive assessment program executed by the computing device. The computing device generates the user's performance data based on the at least one cognitive assessment task. The server receives the user's performance data from the computing device. The server compares the user's performance data and the user's historical performance data with the performance data of the healthy group and the patient group to generate a comparison result. The server generates a risk indicator based on the comparison result and generates a cognitive function assessment report based on the risk indicator and the comparison result. The user's performance data includes one eye movement information. Attached Figure Description

[0007] Figure 1 This is a schematic diagram of the cognitive assessment system in the embodiment.

[0008] Figure 2A and Figure 2B This is a flowchart of the cognitive assessment procedure in the embodiment.

[0009] Figure 3A and Figure 3B This is a schematic diagram illustrating the process of a simple image memorization task and the corresponding simple memorization task in the embodiment.

[0010] Figure 4A and Figure 4B This is a schematic diagram illustrating the process of repeating visual search operations and corresponding examples in the embodiments.

[0011] Figure 5 This is a schematic diagram of the reverse scanning operation in the embodiment.

[0012] Figure 6 and Figure 7 This is a schematic diagram of the spatial prediction operation and the target's movement route in the embodiment.

[0013] Figure 8A and Figure 8B This is a schematic diagram of the relational memory operation in the embodiment.

[0014] Figure 9 This is a schematic diagram of a visual search operation in another embodiment.

[0015] The annotations in the attached figures are explained as follows:

[0016] 100: Cognitive Assessment System

[0017] 110: Computing device

[0018] 120: Eye-tracking device

[0019] 130: Server

[0020] 112,132: Processor

[0021] 114,134: Memory devices

[0022] 116: Input / output device

[0023] 118: Monitor

[0024] 122: Image Sensor

[0025] 200: Method

[0026] S102~S130: Steps Detailed Implementation

[0027] Figure 1 This is a schematic diagram of a cognitive assessment system 100 according to an embodiment of the present invention. The cognitive assessment system 100 includes a server 130, an eye-tracking device 120, and a computing device 110 coupled to the server 130 and the eye-tracking device 120.

[0028] The computing device 110 includes a processor 112, a memory device 114, an input / output device 116, and a display 118. The input / output device 116 can be a keyboard, microphone, camera, etc., or an external device such as a mouse or game controller. The display 118 can be embedded in the computing device 110 or be an external device. The memory device 114, which stores cognitive assessment programs and related media materials, can be random access memory (RAM), flash memory, hard disk, or any combination thereof. The processor 112 executes the programs in the memory device 114 according to commands received from the user or server 130. The processor 112 can also calculate the user's performance data in the cognitive assessment task based on gaze-related information data, transmit the performance data from the computing device 110 to the server 130, and respond to instructions from the server 130. The memory device 114 can also store eye-tracking programs. In such an embodiment, the processor 112 can also be used to receive image data from the eye-tracking device 120 and calculate the user's gaze-related information data based on the received image data and the eye-tracking program stored in the memory device 114. The gaze-related information data includes gaze position, pupil size, and eye movement classification (such as fixation, saccades, blinks, etc.). The computing device 110 can be a smartphone, tablet computer, laptop computer, extended reality (e.g., AR, VR, and MR) device, and / or smart glasses.

[0029] Eye-tracking device 120 includes an image sensor for capturing images of a user's face or eyes. The image sensor may be a webcam, an infrared camera, or any other sensor capable of capturing images. Eye-tracking device 120 may be an external device or embedded in computing device 110. In one embodiment, eye-tracking device 120 also includes a separate processor that receives images captured by eye-tracking device 120 to calculate information data related to eye gaze. The separate processor may be implemented as a small low-power application-specific integrated circuit, a digital signal processor, or a field-programmable gate array. In such an embodiment, eye-tracking device 120 sends gaze-related information data to computing device 110. In other embodiments, images captured by eye-tracking device 120 are sent to and processed by computing device 110. In some embodiments, eye-tracking device 120 may also include a user-facing illumination device to illuminate the user's face and / or eyes to facilitate image capture. The eye-tracking device 120 can be a long-range eye tracker or a wearable eye tracker, such as an eye tracker embedded in a wearable device (e.g., an augmented reality device, smart glasses), a glasses-type eye tracker, or an eye tracker embedded in the user's own glasses. The eye-tracking device 120 can be connected to the computing device 110 via a wired network, a wireless network (e.g., a Wi-Fi hotspot, Bluetooth), and / or a Universal Serial Bus.

[0030] In embodiments where the eye-tracking device 120 is a wearable eye tracker, the eye-tracking device 120 may further include an image sensor 122 for capturing images of the user's field of vision. The image sensor 122 may be a standalone image sensor or an image sensor embedded in a computing device. The gaze position is represented by user-centric coordinates (e.g., a coordinate system defined by the image sensor or the embedded display of the wearable device), where the origin moves with head movement. To more accurately analyze the user's gaze-related information data, the eye-tracking device 120 needs to record the user's head movements and convert the information from user-centric coordinates to world coordinates. To record head movements, in some embodiments, the cognitive assessment system 100 may also include a set of machine-readable markers, which may be digital signals presented on the display of the computing device or physical objects placed in the environment. Such machine-readable markers may be geometric shapes, text, symbols, specific colors, color gradients, or light signals with specific intensities and / or wavelengths. The image sensor 122 of the eye-tracking device 120 can capture images of the user's view including the machine-readable markers. The computing device 110 or eye-tracking device 120 can calculate head movements based on displacements marked in consecutive image frames captured by the image sensor 122. In another embodiment, the system may further include a gyroscope and / or accelerometer for recording the user's head movements. Once the head movements are recorded, the computing device 110 or eye-tracking device 120 can convert the coordinates of gaze-related information data from user-centric coordinates to world coordinates based on the head movements.

[0031] Server 130 may be a cloud server connected to computing device 110 via a wired or wireless network. Server 130 includes processor 132 and memory device 134. Memory device 134 stores historical data of individuals (i.e., users) being evaluated, cognitive assessment data of healthy groups, and cognitive assessment data of patient groups.

[0032] Data collected from users, healthy groups, and patient groups includes absolute and relative (i.e., the difference between current performance and performance in the first assessment) task performance in cognitive evaluations. Using machine learning, this data can be used to form models to estimate the probability that a user is categorized into a specific group (e.g., healthy or patient) and the individual's risk of developing a disease. The models are built based on an individual's task performance in a single cognitive assessment and relative task performance patterns across multiple cognitive assessments.

[0033] Figure 2A and 2B This is a flowchart of method 200 of the cognitive assessment procedure in this embodiment. Method 200 includes the following steps:

[0034] S102: The user inputs background information of the arithmetic unit 110;

[0035] S104: User calibration of eye-tracking device 120;

[0036] S106: The computing device 110 performs basic tasks for the user and obtains the user's basic task performance data;

[0037] S108: The computing device 110 transmits the user's basic job performance data to the server 130;

[0038] S110: Server 130 compares the basic job performance data with the stored data and generates the first comparison result;

[0039] S112: Server 130 calculates the first risk indicator based on the first comparison result;

[0040] S114: Server 130 determines whether the first risk indicator is greater than the first threshold; if yes, proceed to step S116; if no, proceed to step S130.

[0041] S116: Server 130 generates a first abnormal indicator based on the first risk indicator;

[0042] S118: The computing device 110 performs advanced tasks for the user and obtains the user's advanced task performance data;

[0043] S120: The computing device 110 transmits the user's advanced task performance data to the server 130;

[0044] S122: Server 130 compares the advanced job performance data with the stored data to generate a second comparison result.

[0045] S124, The server calculates the second risk indicator based on the second comparison result;

[0046] S126: Server 130 determines whether the second risk indicator is greater than the second threshold; if yes, proceed to step S128; if no, proceed to step S130.

[0047] S128: Server 130 generates a second abnormal indicator based on the second risk indicator;

[0048] S130: Server 130 generates a cognitive assessment report.

[0049] In step S102, the user can input background information through the input / output device 116. The background information may include name, gender, age, educational background, native language, etc. In another embodiment, the background information may also include fingerprints and / or the user's facial identity. The computing device 110 can automatically load stored user information when recognizing the user's fingerprints and / or facial identity. The background information may be stored in the computing device 110 or in the server 130.

[0050] After receiving background information, in step S104, the user completes the calibration of the eye-tracking device 120 according to the instructions of the computing device 110. The calibration process includes three steps. In the first step, the computing device 110 instructs the user to confirm the position of the eye-tracking device 120. The user can move the eye-tracking device 120 to the correct position according to graphic or voice instructions. This step can be performed automatically after the eye-tracking device 120 is initially powered on. If the eye-tracking device 120 is correctly positioned, the computing device 110 will continue to execute the second step. In the second step, the user views one or more gaze points. The gaze point can be a virtual object presented on a screen or paper. Visual and / or audio feedback is provided after detecting the user's gaze point. In another embodiment, the second step can also be performed in the form of an interactive game. In the third step, the computing device 110 presents the calibration results for the user to evaluate the performance of the eye-tracking device 120. The user can determine whether the quality of the received gaze-related information data is sufficient. The calibration result can be the average score of all gaze points viewed during the calibration process or the score of each gaze point viewed during the calibration process. The aforementioned score can be a value of accuracy and / or reliability calculated based on an eye-tracking algorithm (e.g., the numerical distance between the received gaze position and a specific gaze point). The calibration results can be displayed in the form of visual and / or audio information. If the calibration results are not accurate enough, the second step can be repeated to recalibrate the eye-tracking device 120. In one embodiment, if the eye-tracking device 120 has been previously calibrated, the computing device 110 can use the previous calibration data based on the user's background information. After the eye-tracking device 120 is calibrated, in step S106, the computing device 110 automatically performs basic tasks. The following paragraphs describe the tasks employed in method 200. Depending on the application, these tasks can be categorized as basic tasks or advanced tasks.

[0051] Figure 3A and 3BThis is a schematic diagram of the process and corresponding steps of a simple image memorization task. The memorization task is divided into two stages: (1) the memorization stage; and (2) the recall stage. Before the task is executed, the computing device can preload the task's configuration data and media material. The configuration data includes three types of time information regarding image presentation: (1) the first time period refers to the time each image is displayed on the display; (2) the second time period refers to the interval between displaying two consecutively presented images; and (3) the third time period refers to the time between the memorization stage and the recall stage. The configuration data also includes: (1) a first number, indicating the predetermined number of attempts in the memorization stage; (2) a second number, indicating the predetermined number of attempts in the recall stage; and (3) the order of attempts performed in the task. In a simple image memorization task, each attempt refers to the display of a specific pair of two images. The two images can be the same or different from each other. The media material includes a first gallery and a second gallery, used for the memorization stage and the recall stage, respectively.

[0052] During the memory phase, the processing device 110 randomly selects an image from a first image library. Each selected image constitutes an attempt. The selected image is presented to the user and maintained for a first period of time (e.g., 5 seconds). Afterward, a first blank screen is displayed and maintained for a second period of time (e.g., 1 second). After the first blank screen ends, the processing device 110 can confirm whether the number of attempts has been reached. If the first number has been reached, the processing device 110 can end the memory phase and the second blank screen is subsequently displayed and maintained for a third period of time (e.g., 1 second). Otherwise, the processing device 110 can continue to execute a new attempt.

[0053] Following the second blank screen, the processing device 110 executes the recall phase. The recall phase is similar to the memory phase, except that during each image presentation (i.e., attempt), two different images are displayed side-by-side. One is the image presented to the user during the memory phase (i.e., the repeated image), and the other is an image randomly selected from a second image library that was not displayed during the memory phase (i.e., the new image). The processing device 110 then instructs the user to look at the new image. Similar to the memory phase, at the end of each first blank screen, the processing device 110 determines whether the number of attempts in the recall phase has reached the second count. If so, the processing device 110 can end the task and calculate the task performance. The task performance data is calculated based on a combination of task information data and user gaze-related information data. The task information data includes timestamps of the images appearing and disappearing during presentation, as well as the image's position during the memory and recall phases.

[0054] Task performance can be determined using the following task performance data during the recall phase: (1) the proportion of time the user spends fixating on the repeated image and the new image respectively; (2) the difference in the proportion of time the user spends fixating on the repeated image and the new image; (3) the number of fixations on the repeated image and the new image respectively; (4) the number of saccades between the repeated image and the new image; (5) the number of saccades and eye movements in the repeated image; (6) the number of saccades and eye movements in the new image; (7) the time from the start of image presentation to the first fixation on the repeated image; and (8) the time from the start of image presentation to the first fixation on the new image. Further detailed assessment of cognitive function can be performed by analyzing the above task performance data according to the time interval between the presentation of the same image in the memory and recall phases. The temporal analysis may include trend analysis of different task performance data, and / or setting a threshold (e.g., 120 seconds), grouping each attempt based on whether the time interval between the presentation of the image in the memory and recall phases exceeds the threshold, and comparing between groups.

[0055] For example, for a given image, the time between the image presented in the memory phase and the same image presented in the recall phase might range from 10 seconds to 180 seconds. As time increases, the proportion of time a user spends looking at the new image decreases, and the difference in the proportion of time a user spends looking at repeated images and new images decreases. This phenomenon may indicate that it becomes more difficult for users to retain images in their memory over time. People with impaired memory tend to perform worse over longer time intervals. In addition to the aforementioned fixation-related parameters, user performance can also be determined by other parameters, such as saccades and pupil size.

[0056] Figure 4A and 4B This is a schematic diagram illustrating the process of a repetitive visual search task and corresponding examples. Before performing a visual search task, the computing device 110 can preload the task's configuration data and media materials. The configuration data includes the time spent on each step of the task, the number of searches in a single attempt (e.g., two searches per attempt), the number of objects displayed in each attempt, and the total number of attempts.

[0057] The first search target and search array for each attempt are randomly generated by the system. Objects in the search array can be models, numbers, or graphics. Upon entering the attempt phase, a placeholder is displayed in the search array during the first time interval, e.g., 1 second. The placeholder indicates the location of objects (including target and non-target objects). After the first time interval, the target object is displayed on the screen (the placeholder remains). Then, after a second time interval, e.g., 1 second, the placeholder can be removed. The first time interval can be longer than the second time interval. After the second time interval, the user is instructed to search for and gaze at the target object as quickly as possible. The results are sent to the computing device 110 regardless of whether the user selects the correct or incorrect target object. After each search, the computing device 110 can confirm whether the predetermined number of searches performed in the current attempt has been reached. If not, the attempt is repeated, and the computing device 110 can select the next search target based on the gaze-related information data collected in this search. For example, if the search array includes ten target objects, and the user has only correctly viewed three target objects, the next search target object can be selected from the three correctly viewed targets.

[0058] When the number of attempts has reached the predetermined total number of attempts, the computing device 110 can calculate the job performance data based on gaze-related information data and job information data (e.g., the location of each object and the search timestamp).

[0059] The following are examples of performance data calculated in repetitive visual search tasks to estimate various cognitive abilities: (1) time required to fixate on a non-target object; (2) number of fixations before fixing on a non-target object; (3) number of fixations on a non-target object; (4) fixation time on a non-target object; (5) time required to fixate on a target object; (6) number of fixations before fixing on a target object; (7) number of fixations on a target object; (8) fixation time on a target object; and (9) number of re-fixations on a target indicator object. Performance data can be used to estimate a user's executive function. Differences between successive searches can be used to estimate a user's memory function. To increase the sensitivity of performance data, the objects can be varied to make repetitive visual search tasks more difficult (e.g., more objects and / or higher similarity between search objects). Furthermore, the concept of conjunction search can be used in object selection. If the objects used include semantic relationships, repetitive visual search tasks can also be used to estimate language-related abilities. Finally, repetitive visual search tasks can be implemented in a way that has higher ecological validity. That is, the search array can be more than just a set of abstract objects. Repetitive visual search tasks can require users to search for an object within a scene, such as finding clothes in a closet or a specific book in a room. The search can use two-dimensional or three-dimensional images, and can even be performed in extended reality.

[0060] In step S108, the computing device 110 transmits the user's basic performance data to the server 130. In step S110, the server 130 compares the basic performance data with the basic performance data of groups in the database to generate a first comparison result. In step S112, the server 130 calculates a first risk indicator based on the first comparison result. In one embodiment, to calculate the first risk indicator, the server 130 first selects a group in the database corresponding to the user's background information (e.g., age, gender, educational background) and compares the user's basic performance data with a set of data. This set of data includes the user's historical performance data, performance data of healthy groups and / or sick groups (e.g., subjective cognitive decline, mild cognitive impairment, Alzheimer's disease, and other diseases involving cognitive impairment). Furthermore, performance includes the overall distribution of absolute and relative performance, as well as the trend of an individual's relative performance across multiple cognitive assessments, which can be used to construct a historical performance model. The first comparison result may include the user's ranking within a specific group and the probability that the user is classified into that specific group.

[0061] Historical performance models can be used to estimate changes in cognitive function over a period of time. These models are generated based on relative performance data across multiple cognitive assessments. User data, data from healthy populations, and / or patient data are each used to form a historical data model. The probability of a user being classified as either healthy or patient can be calculated based on the goodness-of-fit of the user's relative performance data with the historical performance data models for healthy populations and / or patient populations. This goodness-of-fit can be achieved using probability, root mean square error, or any other statistical method that can describe the difference between the observed performance data and the historical performance models. In other embodiments, performance data may also include trends in an individual's absolute performance in cognitive assessments, and historical performance models may also be generated based on these trends in absolute performance.

[0062] Server 130 also includes a discrimination model based on the first comparison result to calculate a first risk indicator. The discrimination model can distinguish users with a certain degree of cognitive impairment from healthy individuals. Embodiments can estimate the risk indicator based on the numerical distance between a user's performance data and a threshold defined by the discrimination model. Specifically, the discrimination model can estimate a user's performance as a position in geometric space. Each parameter in the first comparison result (e.g., the user's ranking in each group, and the difference between the user's data and the historical performance model for each group) can be a dimension of that space, and the dimensionality of the space can be further reduced by merging parameters through relationships between these parameters. Depending on the dimension of the space, the threshold can be a point, line, or plane in the space. The numerical distance between the user's performance data and the threshold can be calculated as the Euclidean distance. A cumulative distribution function can be used to estimate the risk indicator. In the function, the x-axis is the Euclidean distance, and the y-axis is the probability of classifying the user into a given group. The estimated risk indicator is the probability corresponding to a given numerical distance. In other embodiments, server 130 can generate a discrimination model based on underlying performance data.

[0063] In step S114, server 130 determines whether the user's first risk indicator is greater than a first threshold. If the first risk indicator is greater than the first threshold, server 130 generates a first anomaly indication, and computing device 110 executes advanced operations. Otherwise, server 130 generates a cognitive assessment report for the user. If the user has previously performed a cognitive assessment and a second anomaly indicator has been identified, then regardless of whether the first and second risk indicators are greater than the aforementioned thresholds, cognitive assessment system 100 will execute step S118.

[0064] In step S118, the computing device 110 responds to a command from the server 130 and executes an advanced task. The advanced task can further examine whether the user's cognitive impairment is multifaceted. For example, the advanced task may include: (1) a reverse visual skipping task; (2) a spatial prediction task; (3) a novel repetitive visual search task; (4) a reading task; and (5) an associative memory task.

[0065] Figure 5 This is a schematic diagram of a reverse gaze skipping task. The task configuration data includes fixation duration, central target duration, peripheral target duration, feedback marker duration, total number of attempts, and peripheral target location. The task configuration data is pre-loaded into the computing device 110 before the task begins. After the task begins, the fixation point is displayed at the center of the screen and lasts for a first predetermined time. The fixation point can be a disc, cross, or other geometric shape. The fixation point is then replaced by the central target and lasts for a second predetermined duration (e.g., 500-1500 milliseconds). The central target can be a disc, bullseye, or other geometric shape different from the fixation point. Then, the peripheral target is displayed at a peripheral location and maintained for a third predetermined time (e.g., 1000 milliseconds). The peripheral location can be any position away from the center of the screen within a certain viewing angle. For example, the peripheral location could be 6° or 12° from the center of the screen on the horizontal axis. Once the peripheral target is displayed, the user is instructed to look at a designated location as quickly as possible; this designated location is a mirror image of the peripheral location. After the peripheral target display time ends, the cognitive assessment system 100 provides feedback to the user. Feedback markers can be visual indicators that notify the user whether they are correctly looking at a designated location. After the feedback marker is displayed, all images disappear from the screen. The processing unit 110 can check whether the total number of attempts has been reached. If so, the operation ends. Otherwise, the above process can be repeated.

[0066] The computing device 110 can analyze task information data and user gaze-related information data. The task information data includes timestamp data of the above steps (e.g., timestamp of the display of the central target and timestamp of the display of the peripheral targets). The computing device 110 then evaluates the user's task performance data by calculating the correspondence between the target position and the user gaze-related information data. The data used to evaluate the task performance data includes: (1) the number and / or proportion of times the user correctly gazes at the designated position on the screen after the peripheral target is displayed; (2) the number and / or proportion of times the user gazes at the peripheral target first and then moves their gaze to the designated position after the peripheral target is displayed; (3) the number and / or proportion of times the user gazes only at the peripheral target without moving their gaze to the designated position before the feedback mark is displayed; (4) the reaction time from the start of the display of the peripheral target to the user's first gaze at the designated position; (5) the reaction time between the time the peripheral target is displayed and the time when the user moves their gaze to the designated position after seeing the peripheral target; and (6) the numerical distance between the gaze position and the designated position. The above task performance data can also be evaluated by trend analysis of task performance data changing with the number of attempts.

[0067] Figure 6 and Figure 7 This is a schematic diagram of the spatial prediction task and target movement path in this embodiment. The task configuration data includes a geometric array, the total number of experiments, the target movement time, the target movement path, and the duration of each target position displayed within the movement path. The target movement path may include the movement mode, movement length, and task difficulty. In the spatial prediction task of this embodiment, a series of geometric shapes can be displayed on the screen. For example... Figure 6 The diagram shows a 5×8 circular array.

[0068] At the start of each attempt, the computing device 110 can randomly select a movement path from multiple target movement paths and can choose a hollow circle in the array as the starting position (such as the geometry of a marker). This is to notify the user of the target's starting position so that the attempt is about to begin. The marker can be displayed for a certain period of time (e.g., 1000 ms), after which the hollow circle of the marker is replaced by the target (a solid circle). The solid circle can be displayed for a certain period of time (e.g., 1000 milliseconds). Then, following the default movement path, the target is displayed at different positions in the array, each display lasting for a different duration (e.g., 800 ms–1200 ms). The cognitive assessment system 100 prompts the user to react to the movement of the target using the input / output device 116 or other means. The attempt ends when the target disappears from the last position of the movement path. Figure 7 Some exemplary movement paths on a 7×10 array used for spatial prediction operations are shown.

[0069] When the attempt ends, the computing device 110 can check whether the total number of attempts has been reached. If so, the job is completed. Otherwise, the above process will be repeated. Job information data and user gaze-related information data can be used to calculate job performance data. Job information data includes: (1) timestamps of the target being displayed at each position; (2) the position of the target shown. The user's predicted behavior can be obtained from the correspondence between the user's gaze-related information data and the job information data. In the embodiment, the main predicted behavior is the predicted gaze jump, which can be defined as a gaze jump performed before the target moves to the next position, and the endpoint of the gaze jump falls on another position in the array. If the endpoint of the predicted gaze jump coincides with the next target position, then the predicted gaze jump is a correct predicted gaze jump; otherwise, it is an incorrect predicted gaze jump.

[0070] The following is performance data available from predicted jumps to assess a user's cognitive function. Performance data includes: (1) the number of correct and incorrect predicted jumps; and (2) the ratio of correct to incorrect predicted jumps. An embodiment can calculate performance data for each location on the movement route by averaging all attempts. Furthermore, data for each location on the movement route may not be equally important. At least in the first two locations, the user may not yet know the movement route the target follows. Therefore, in some cases, the user's performance data can be determined by the jumps detected in the last few locations on the movement route. Finally, an embodiment can also present the aforementioned performance data as a function of the locations on the movement route.

[0071] Figure 8A and 8B This is a schematic diagram of the relational memory task in the embodiment. In the memory task, each experiment includes: (1) a search phase; (2) a recall phase. The task configuration data includes: (1) the presentation time of the gaze point; (2) the position information of the grid; (3) the type, number, and position of the target object and reference object in the grid for each attempt; (4) the set duration of each attempt; (5) the duration of the search phase and the recall phase; (6) the number of attempts; (7) the time between the search phase and the recall phase; (8) the feedback time; and (9) the task difficulty for each attempt.

[0072] In a memory task, the target object may be geometrically similar to a reference object, but differs in at least one characteristic such as color, orientation, or shape. At the start of each attempt, the target object is displayed in the center of the screen for 2 seconds. Then, a search array is displayed on the screen for 6 seconds. The search array displays the target object and multiple reference objects. The position, number, and similarity of the reference objects are determined by task configuration data. The user needs to find and fixate on the target object in the search array. The processing device 110 can determine whether the user has successfully fixated on the target object. Next, a fixation point is displayed on the screen for 3 seconds or longer (e.g., 6 seconds), such as... Figure 8A As shown. The point of focus can be a geometric shape, such as a disk or a cross. The geometry of the point of focus can differ from the geometry of the target object and the reference object.

[0073] Next, the experiment enters the recall phase. The recall array will be displayed for 6 seconds. The recall phase displays the same grid as the memory phase, but only the reference objects are shown, and the positions of the reference objects are offset. This changes the absolute positions of the reference objects in the grid, while maintaining their relative positions. The user needs to find the correct relative position of the target object in the grid (e.g., in...). Figure 8A and 8B (Hollow and dashed shapes in the image). If the user successfully gazes at the relative position of the target object, the processing device 110 can confirm that the user's gaze is correct. After the recall phase, the system 100 displays feedback information to inform the user of their performance. The feedback information can be visual or audio information. Similar to the above task, at the end of each attempt, the processing device 110 can check whether the predetermined total number of attempts has been reached. If so, the task is completed, and the user's performance data is evaluated by analyzing the correspondence between task information data and gaze-related information data. Otherwise, the attempt can be repeated. The duration of the attempt phase is merely an example, and the invention is not limited thereto.

[0074] Figure 8B An exemplary search array for the search phase and some memory arrays corresponding to the memory phase of the search array are shown. Figure 8B At the top, the black dots indicate the location of the target object that the user should focus on during the search phase. Figure 8B At the bottom, the hollow circles represent the relative positions of the target objects the user should focus on during the recall phase. These arrays are merely examples. In other embodiments, the target and reference objects may also be raster patterns, such as Gabor patches. The invention is not limited thereto.

[0075] The following is the performance data used for task performance evaluation. Performance data includes: (1) the number and / or percentage of correct responses during the search phase; (2) the reaction time to fixate on the target object during the search phase; (3) the number of fixations required to obtain a correct response during the search phase; (4) the proportion of time the user spends fixating on the target object, reference objects, and other locations in the grid during the search phase; (5) the number of fixations on the target object, reference objects, and other locations in the grid during the search phase; (6) the number and / or percentage of correct responses during the recall phase; (7) the reaction time to find the relative position of the target object during the recall phase; (8) the number of fixations required to obtain a correct response during the recall phase; (9) the number and / or proportion of fixations on the relative position of the target object, reference objects, and other locations during the recall phase; and (10) the length of time the user spends fixating on the relative position of the target object, reference objects, and other locations during the recall phase. All of these performance data can be further grouped according to task difficulty, and comparisons can be made between different groups of performance data.

[0076] The difficulty of each attempt is related to the similarity of features between the target and reference objects. These features include orientation, shape, and color. For example, the target and reference objects could be circular images with raster patterns. The raster patterns on the images can have different orientations. For example, the similarity between a target object at 0° and a reference object at 10° is higher than the similarity between a target object at 0° and a reference object at 15°. Furthermore, the difficulty of the task is positively correlated with the time between the search and recall phases.

[0077] Figure 9 This is a schematic diagram of a visual search operation according to another embodiment. The operation configuration data includes: (1) the target setting time; (2) the duration of the search phase and the test phase; (3) the duration of the feedback; (4) the time between the search phase and the test phase; (5) the number of attempts in the search phase and the test phase; (6) a list describing the targets in the search phase and the test phase; (7) a list of scenarios for each attempt in the test phase; (8) an array with vertical and horizontal lines for displaying the visual objects in each phase; (9) the type of visual objects displayed in the array; and (10) the position of the visual objects in the array. The visual objects used in the operation can be geometric figures or pictures, or geometric figures with different degrees of rotation.

[0078] When the search phase begins, a target object (e.g., a triangle) is displayed on the screen to inform the user what to search for. The search array is then displayed on the screen for 2 seconds. The user needs to search for the target object during this time. After the search time ends, the target objects in the array are marked with a different color or in bold as feedback. After the attempts are completed, the computing device 110 can check whether the predetermined total number of attempts has been reached. If not, the above attempts can be repeated. If the predetermined total number of attempts has been reached, the attempts can proceed to the testing phase. The testing phase is similar to the search phase, except that in the testing phase, according to the scenario list, in some attempts, the computing device 110 can randomly swap the target object with one of the non-target objects in one of the search arrays (e.g., one of three objects with the same shape as the target object but with different degrees of rotation, such as...). Figure 9 (as shown), and does not provide feedback to users.

[0079] After the task is completed, the computing device 110 can evaluate the user's performance by analyzing task information data and user gaze-related information data. Task information data includes timestamps for each step and the position of objects in each step. Task performance data includes: (1) the ratio of time the user gazes at the target object and non-target objects; (2) the number of times the user gazes at the target and non-target objects during the test phase. Task performance data can also be grouped based on the following conditions: (1) the test phase context determined by the context list; (2) the time interval between searching for a specific object during the search phase and searching for the same object during the test phase.

[0080] After completing the advanced task in step S118, in step S120, the computing device 110 transmits the user's advanced task performance data to the server 130. In step S122, the server 130 compares the user's advanced task performance data with other data to generate a second comparison result. In step S124, the server 130 calculates a second risk indicator based on the second comparison result. In one embodiment, to calculate the second risk indicator, the server 130 first selects a group in the database corresponding to the user's background information (e.g., age, gender, educational background) and compares the user's advanced task performance data with a set of data. This set of data includes the user's historical task performance data, task performance data of healthy groups and / or sick groups (e.g., subjective cognitive decline, mild cognitive impairment, Alzheimer's disease, and other diseases involving cognitive impairment).

[0081] Furthermore, performance includes the distribution of absolute and relative performance, as well as trends and patterns in an individual's relative performance across multiple cognitive assessments. Historical performance models can be generated for at least one set of data (i.e., data from users, healthy groups, and / or patient groups) based on the relative performance in cognitive assessments. In other embodiments, performance also includes absolute performance across multiple cognitive assessments, and historical performance can be generated based on absolute performance data. A second comparison result can include a user's ranking within a specific group and the probability that the user is classified into that specific group. The probability of a user being classified into a specific group can be calculated based on the fit between the user's relative performance across multiple cognitive assessments and the historical performance models for healthy and / or patient groups. This fit can be achieved using probability, root mean square error, or any other statistical method that can describe the difference between the performance data and the historical performance models.

[0082] To calculate the second risk indicator, server 130 may further calculate a local risk value for each job according to the method described in step S112, and then calculate the second risk indicator based on all local risk values. The second risk indicator may be the average of all local risk values ​​weighted by a specific constant. This constant may be defined by sensitivity, specificity, area under the receiver operating characteristic curve, and / or accuracy for each job. These constants are used to distinguish between patient and healthy groups. In other embodiments, server 130 may also incorporate the first comparison result into the calculation of the local and second risk indicators. In other embodiments, to calculate the second risk indicator, server 130 may form a discrimination model constructed from both basic job performance data and advanced job performance data to distinguish between patient and healthy groups. The job performance data includes job performance data from a single test and job performance data from several subsequent tests. In this case, the second risk indicator may be the Euclidean distance between the user's overall job performance and a threshold defining the discrimination model in Euclidean space.

[0083] More specifically, the discrimination model can estimate a user's performance as a position in geometric space. Each performance data point can construct a one-dimensional space. In some cases, statistical methods (i.e., factor analysis and principal component analysis) can be used to extract the main factors influencing the performance data. The dimensionality of the space can be further reduced to the number of main factors. Depending on the number of dimensions of the space, the threshold can be a point, line, or plane in the space. The numerical distance between a user's performance data and the threshold can be defined as the Euclidean distance. The cumulative distribution function can be used to estimate the risk indicator. In this function, the x-axis is the Euclidean distance, and the y-axis is the probability of classifying the user into a specific group. The estimated risk indicator corresponds to the probability at a specific numerical distance.

[0084] After obtaining the second risk indicator in step S124, server 130 determines in step S126 whether the second risk indicator is greater than the second threshold. If the second risk indicator is greater than the second threshold, then in step S128, server 130 can generate a second abnormal indicator.

[0085] In step S130, server 130 may generate a cognitive assessment report for the user. The cognitive assessment report may include: (1) scores, charts and / or symbols indicating the user’s performance relative to a healthy group whose background data is consistent with their performance; (2) risk indicators indicating the probability that the user will develop a certain cognitive impairment; and (3) suggestions for the user’s daily activities to reduce the risk of developing cognitive disorders.

[0086] Server 130 can also determine when a user should retake a cognitive assessment task based on the user's performance data and the distribution of the first and second anomalies. If the cognitive assessment system 100 generates the first and / or second anomalies, the user should retake the cognitive assessment task within a short period (e.g., one week). Otherwise, the user may retake the cognitive assessment task after a longer period (e.g., six months). Server 130 can further connect to the user's personal calendar (e.g., Google Calendar) and schedule cognitive assessments.

[0087] In another embodiment, the computing device 110 may alert the user based on a first abnormal indicator and / or a second abnormal indicator. If the first abnormal indicator and / or the second abnormal indicator frequently appear in several consecutive assessments, the computing device 110 may determine that the user's cognitive function is declining and will remind the user to obtain medical information.

[0088] In another embodiment, the server 130 of the cognitive assessment system 100 may be connected to a medical platform and / or a medical institution. Cognitive assessment reports and alerts may be sent to the medical platform and / or the medical institution for evaluation by medical professionals.

[0089] Basic and advanced tasks can also be modified according to the application. For example, if impairment of memory function is critical to a specific application, basic tasks may include tasks that assess memory function. In other cases, if a key factor in the application is determining whether a user has multifaceted cognitive impairment, basic tasks may include sub-tasks that assess memory function and other sub-tasks that assess executive function. Furthermore, if a specific cognitive impairment, such as executive function, is identified from basic tasks, the cognitive assessment system 100 can select advanced tasks that assess executive function based on this.

[0090] The above description is only a preferred embodiment of the present invention. All equivalent changes and modifications made in accordance with the claims of the present invention should be included within the scope of the present invention.

Claims

1. A cognitive assessment system, characterized in that, Include: A computing device includes a cognitive assessment program for performing at least one cognitive assessment task, the at least one cognitive assessment task being used to assess a user's cognitive function; A server, coupled to the computing device, includes a database storing historical performance data of the user, a historical performance model of the user, performance data of a healthy group, performance data of a patient group, a historical performance model of the patient group, and a historical performance model of the healthy group; wherein the user's historical performance model is generated based on the user's historical performance data; and An eye movement tracking device, coupled to the computing device, is used to capture the user's eye movement information; in: The user performs the at least one cognitive assessment task through the cognitive assessment program executed by the computing device; The computing device generates performance data of the user based on the at least one cognitive assessment task; The server receives the user's job performance data from the computing device; The server compares the user's performance data and historical performance data with the performance data of the healthy group and the patient group to generate a comparison result. The comparison result includes the user's performance ranking and the probability that the user is classified into which group of the database. The probability that the user is classified into which group of the database is determined based on the fit between the user's historical performance data and the historical performance model of each group in the database. The server generates a risk indicator based on the comparison results, and generates a cognitive function assessment report based on the risk indicator and the comparison results; and The user's performance data includes visual information.

2. The cognitive assessment system as described in claim 1, characterized in that, The at least one cognitive assessment task includes a basic task and an advanced task.

3. The cognitive assessment system as described in claim 2, characterized in that, The basic tasks are used to assess multiple cognitive functions.

4. The cognitive assessment system as described in claim 2, characterized in that, The advanced task includes a plurality of tasks that assess a specific cognitive function, and the computing device selects to execute at least one of the plurality of tasks that assess a specific cognitive function based on the assessment results of the basic task.

5. The cognitive assessment system as described in claim 1, characterized in that, The server also includes a discrimination model for distinguishing between the patient group and the healthy group based on the comparison results.

6. The cognitive assessment system of claim 5, wherein the risk indicator is a threshold of the discrimination model and a numerical distance between the user's performance data.

7. The cognitive assessment system as described in claim 6, characterized in that, The patient group is a group of individuals with cognitive impairment.

8. The cognitive assessment system as described in claim 1, characterized in that, The cognitive function assessment report includes: Scores, charts, and / or symbols indicate a user's performance relative to a healthy group whose background data aligns with their performance. Risk indicators represent the probability that a user will develop a certain cognitive impairment; and Provide users with suggestions for daily activities to reduce the risk of developing cognitive disorders.

9. The cognitive assessment system as described in claim 8, characterized in that, If the user's risk index is greater than a threshold, the server generates an anomaly flag.

10. The cognitive assessment system as described in claim 9, characterized in that, The server assists the user in scheduling the next cognitive assessment based on the anomaly marker.

11. The cognitive assessment system as described in claim 9, characterized in that, If the frequency of occurrence of the abnormal marker is greater than a threshold, the server issues a warning message based on the abnormal marker to warn the user of a decline in cognitive function.

12. The cognitive assessment system as described in claim 11, characterized in that, If the server determines that the user has cognitive decline, the cognitive function assessment report includes relevant medical information.

13. The cognitive assessment system as described in claim 12, characterized in that, The server in question is a cloud server.

14. The cognitive assessment system as described in claim 13, characterized in that, The server is connected to a medical platform or a medical institution.

15. The cognitive assessment system as described in claim 14, characterized in that, The server transmits the cognitive function assessment report and / or the warning information to the online medical platform or the medical institution.

16. The cognitive assessment system as described in claim 1, characterized in that, The eye movement information includes gaze position, pupil size, and blink information.

Images