Intelligent assessment system for cognitive function

By developing a digital, automated, and intelligent cognitive function assessment system, the problems of long time consumption, strong subjectivity, and uneven resource distribution of existing assessment methods have been solved, achieving fast, objective, and accurate assessment results that can adapt to the needs of multilingual and cultural backgrounds.

CN118919066APending Publication Date: 2024-11-08苏州复变医疗科技有限公司
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410817460.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-06-24
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

Existing methods for assessing cognitive impairment suffer from problems such as being time-consuming, highly subjective, unevenly distributed resources, large differences in assessment results, and difficulties in data storage and sharing, making it difficult to adapt to the assessment needs of different cultural and linguistic backgrounds.

Method used

A digital, automated, and intelligent cognitive function assessment system has been developed, including a knowledge base module, a database module, a human-computer interaction module, an algorithm module, and an application module. It supports multilingual and multicultural assessments, uses modular encapsulation and intelligent algorithms for scoring, and provides standardized testing procedures and scoring criteria.

Benefits of technology

It enables rapid, objective, and accurate cognitive function assessment, reduces the burden of manual intervention and data processing, improves the automation level of the assessment and the accuracy of the results, and adapts to the assessment needs of different cultural and linguistic backgrounds.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118919066A_ABST
    Figure CN118919066A_ABST
Patent Text Reader

Abstract

The invention discloses an intelligent assessment system for a cognitive function. The system comprises a knowledge base module, a database module, a man-machine interaction module, an algorithm module, a front-end doctor application module and a front-end patient application module. The knowledge base module is used for storing professional data related to cognitive function evaluation; the database module is used for storing the content of the customizable part of the test question; the man-machine interaction module is used for realizing automation of a cognitive evaluation process; the algorithm module is used for understanding feedback of the test user and scoring in combination with a professional standard; the front-end doctor application module is used for providing an editable cognitive function evaluation question generation module for a doctor; and the front-end patient application module is used for providing a module capable of autonomously evaluating and automatically evaluating for the patient. The invention provides a digital, automatic and intelligent cognitive function evaluation system, and the cognitive function can be evaluated more objectively through a standardized test process and a scoring standard.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of computer science and technology, and in particular to an intelligent evaluation system for cognitive functions. Background Art

[0002] Cognitive dysfunction seriously affects patients' disease recovery and quality of life, and brings heavy mental stress and financial burden to their families.

[0003] From the perspective of cognitive function assessment, the existing clinical screening methods for cognitive dysfunction are basically based on subjective judgment by doctors combined with paper-based scales. The shortcomings of this method are mainly reflected in two aspects. On the one hand, the use of professional peer-assessment scales requires specialists to spend a long time to conduct detailed inquiries with patients based on the items of the scale; on the other hand, affected by personal subjective factors, the assessment results given by different doctors may vary greatly.

[0004] Judging from the distribution of medical resources, there is currently a shortage of specialist doctors and their regional distribution is uneven. In particular, the experience of health care doctors who are capable of conducting primary screening varies, which makes missed diagnoses and misdiagnoses prone to occur.

[0005] From the perspective of the subjects participating in the assessment, the language, culture, customs, etc. of different countries or regions are different, and the concepts and educational levels of the subjects participating in the cognitive assessment are different. It is difficult to design a universal assessment method. Therefore, doctors often need to select and adjust relevant scales based on actual conditions.

[0006] Judging from the media carrier of the evaluation tool, the evaluation results of paper quality tables are not easy to save and share, and have certain limitations in data transmission and storage. Summary of the invention

[0007] In order to solve the problems in the prior art, the present invention has developed a digital, automated and intelligent evaluation system based on the cognitive function evaluation scale, which can evaluate cognitive function more objectively with standardized testing procedures and scoring standards.

[0008] The embodiment of the present invention provides an intelligent evaluation system for cognitive function, including:

[0009] Knowledge base module, database module, human-computer interaction module, algorithm module, front-end doctor application module, front-end patient application module;

[0010] The knowledge base module is used to store professional data related to cognitive function assessment;

[0011] The database module is used to store the content of the customizable part of the test questions;

[0012] The human-computer interaction module is used to receive the user's interactive operation to realize the automation of the cognitive assessment process;

[0013] The algorithm module is used to intelligently understand the test user's feedback during the cognitive function assessment process and score the relevant feedback in combination with professional standards;

[0014] The front-end doctor application module is used to provide doctors with an editable cognitive function assessment question generation module;

[0015] The front-end patient application module is used to provide patients with a module that can independently complete cognitive function assessment and can automatically perform assessment.

[0016] Optionally, the knowledge base module includes a test question stem unit and a label unit;

[0017] The test question stem unit includes test questions used in clinical cognitive function assessment, which is used to provide a basis for the construction of an intelligent assessment system for cognitive function;

[0018] The label unit includes a scale type, a test dimension, and a question type, and is used to assign corresponding scale type, test dimension, and question type labels to each test question.

[0019] Optionally, the database module includes a language library unit, a word library unit and a picture library unit;

[0020] The language library unit is configured with multiple languages ​​and local dialects, and is used to provide different language options for test questions;

[0021] The word library unit is configured with different categories of vocabulary, which is used to provide customizable vocabulary categories for test questions with adjustable content;

[0022] The picture library unit is configured with image and icon content, and is used to provide customizable images and icons for test questions with adjustable content.

[0023] Optionally, the human-computer interaction module includes a touch interaction unit, a voice interaction unit and a motion interaction unit;

[0024] The touch interaction unit is used to provide the user with corresponding touch interaction operation options according to the characteristics of the test questions, so that the user can complete the cognitive assessment independently;

[0025] The voice interaction unit is used to provide voice information to the test user and understand the test user's voice feedback on the voice information;

[0026] The action interaction unit is used to prompt the test user to perform a specific action in the action question, and to judge the accuracy of the action after the user performs the action.

[0027] Optionally, the algorithm module includes a speech recognition unit, a text matching unit, a text understanding unit, a trajectory recognition unit, a graphic recognition unit, and an action recognition unit;

[0028] The speech recognition unit is used to support speech-to-text functions in multiple languages ​​and local dialects;

[0029] The text matching unit is used to compare the test user's answer with the standard answer in the question-and-answer type and give a corresponding score;

[0030] The text understanding unit is used to understand the test user's answer from a semantic level and give a corresponding score;

[0031] The track recognition unit is used to recognize the operation of the test user in the connection question and give a corresponding score;

[0032] The graphic recognition unit is used to recognize the drawing of the test user in the graphic memory question and give a corresponding score in combination with the standard graphic;

[0033] The action recognition unit is used to recognize the actions performed by the test user in the action questions and give corresponding scores in combination with standard actions.

[0034] Optionally, the front-end doctor application module is configured on the doctor application side platform, including a basic menu unit, a content editing unit, and a tool preview unit;

[0035] The basic menu unit is used to provide a menu bar related to basic operations for users to operate;

[0036] The content editing unit is used to provide editable digital components for users to generate cognitive assessment tools;

[0037] The tool preview unit is used to provide a preview of the effect of generating a cognitive assessment tool according to the combined digital components.

[0038] Optionally, the front-end patient application module is configured on the patient application side platform, including an information input unit, a human-machine evaluation unit, and a report generation unit;

[0039] The information input unit is used to input the basic information of the test user;

[0040] The human-machine evaluation unit is used for the test user to independently perform cognitive evaluation tests;

[0041] The report generating unit is used to intelligently generate a test report after the test user completes the cognitive assessment test.

[0042] Optionally, the encapsulation process of the questions in the intelligent assessment system includes:

[0043] Assign the corresponding scale type, test dimension and question type to each question through label units;

[0044] Bind the topics with adjustable content to the corresponding database;

[0045] The corresponding human-computer interaction module and algorithm module are embedded in each question to complete the modular encapsulation of the question and generate the corresponding digital components.

[0046] The present invention takes "backstage-doctor-user" as the basic framework, and develops a set of intelligent assessment systems based on the cognitive function assessment scale, which has the features of customized questions, self-generated tools, and fully automatic testing. The system can implement standardized test processes and scoring standards, and can more objectively evaluate cognitive functions, while achieving rapid testing and reducing the burden of manual intervention and data processing. In addition, the system encapsulates test questions in a modular form of "question stem-label-material-interaction-algorithm", which is convenient for doctors to select, adjust and combine test questions according to actual needs, so as to adapt to local conditions and different people; the human-computer interaction module invented by the system improves the automation level of cognitive assessment tools, and the intelligent algorithm module enhances the accuracy, objectivity and reliability of cognitive assessment results. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A basic framework diagram of an intelligent evaluation system for cognitive functions provided by an embodiment of the present invention;

[0048] Figure 2 An example diagram of a system front-end doctor application side interface design provided by an embodiment of the present invention;

[0049] Figure 3 An example diagram of a question-and-answer question integrating voice interaction and virtual human technology on the patient application side of a system front end provided by an embodiment of the present invention;

[0050] Figure 4 This is an example diagram of a graphic interactive question on the patient application side of the system front end provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0052] Example

[0053] The embodiment of the present invention provides an intelligent evaluation system for cognitive function, including: a knowledge base module, a database module, a human-computer interaction module, an algorithm module, a front-end doctor application module, and a front-end patient application module. Figure 1 The intelligent evaluation system in the embodiment of the present invention takes "backstage-doctor-user" as the basic framework. Figure 1 It shows the contents included in the system framework.

[0054] The knowledge base module is created based on existing clinical guidelines related to cognitive assessment and is used to store professional data related to cognitive function assessment.

[0055] In this embodiment, the professional data includes scale type, test dimension and question type, which are all sorted out according to professional guidelines. Among them, the scale type is used for clinical cognitive function assessment, including but not limited to MoCA-B, MMSE, ADL20, PDQ-5, SCD-Q, AVLT, TMTA-B, BNT and AFT.

[0056] The examination dimensions cover the dimensions of various abilities of the test subjects that need to be examined during the cognitive assessment process, including but not limited to executive ability, memory ability, fluency, orientation ability, calculation ability, abstract ability, visual perception ability, naming ability, attention, language ability, etc.

[0057] Question types include but are not limited to multiple-choice questions, essay questions, connection questions, word memory questions, name listing questions, time and space answer questions, calculation combination questions, graphic memory questions, picture answer questions, action questions, etc.

[0058] Furthermore, the above-mentioned knowledge base module includes a test question stem unit and a label unit.

[0059] The test question stem unit includes test questions used in the clinical evaluation of cognitive function, which is used to provide a basis for the construction of an intelligent evaluation system for cognitive function. Since the test questions contained in the professional cognitive assessment scale currently used in clinical practice are relatively fixed, and for professional considerations, the modification of the main content of the test questions needs to be professionally demonstrated, this embodiment introduces a series of professional test questions in the test question stem unit as a knowledge base reserve.

[0060] The label unit includes scale type, test dimension and question type, and is used to assign corresponding scale type, test dimension and question type labels to each test question. The label in this embodiment is both the attribute of each question and provides an index for the subsequent doctor selection question module.

[0061] Furthermore, the database module is used to store the content of the customizable part of the test questions, including a language library unit, a word library unit and a picture library unit.

[0062] Language, culture, age, education, comprehension and other factors may affect the acceptance of standardized scale tests by the group participating in cognitive assessment. Therefore, in actual clinical practice, doctors often adjust the content appropriately to ensure the test effect while making it easier for the other party to accept. The embodiment of the present invention provides a special database module to provide the necessary materials for doctors to adjust or customize the content of test questions.

[0063] The language library unit includes a text part and a voice part, which are configured with multiple languages ​​and local dialects, and are used to provide different language options for test questions. The multiple languages ​​include Chinese, English, etc., and the local dialects include Cantonese, Sichuanese, etc. The text part mainly contains the text content such as the test question stem and notes, and the voice part mainly contains the voice content broadcast by the system.

[0064] The word library unit is configured with different categories of vocabulary, which is used to provide customizable vocabulary categories for test questions with adjustable content. Specifically, the vocabulary in the word library unit includes but is not limited to a fruit word library, a digital word library, an animal word library, and an object word library. Since people have different degrees of familiarity with the things represented by the words, and in order to increase the diversity of the test, these word libraries can be used to adjust the relevant content in certain question types. For example, the word memory questions that test memory ability in MoCA-B / MMSE / AVLT involve different types of words, the fruit name-listing questions that test fluency in MoCA-B / AFT involve various fruits, the time and space questions that test orientation ability in MoCA-B / MMSE involve words that represent time and place, such as seasons, years, months, days, and regions, and the calculation combination questions that test calculation ability in MoCA-B / MMSE involve numbers. Specific examples can be found in the table below:

[0065]

[0066] The picture library unit is configured with image and icon content, and is used to provide customizable images and icons for test questions with adjustable content. For example, in the picture recognition questions in MoCA-B and BNT, the tester needs to point out what is in the displayed image; the visual perception test in MoCA-B requires the tester to circle the corresponding objects in an image containing multiple objects and say the names of the objects; the connection questions in MoCA-B and TMTA-B require the tester to connect the icons as required. Tasks such as these that combine pictures account for a certain proportion in cognitive assessments, and doctors or researchers may select or adjust such pictures. The image and icon content stored in the picture library unit of the present invention facilitates such operations. See the table below for specific examples:

[0067]

[0068] The database module in this embodiment covers various types of words, pictures, icons, etc., especially involving multiple languages ​​and local dialects, providing rich possibilities for doctors to customize the content of evaluation questions.

[0069] Furthermore, the human-computer interaction module is used to receive user interaction operations to realize automation of the cognitive assessment process. This module is designed and developed according to the requirements of the question type.

[0070] Specifically, the human-computer interaction module includes a touch interaction unit, a voice interaction unit and a motion interaction unit, which can adapt to test questions of different styles and promote the automation of the cognitive assessment system.

[0071] The touch interaction unit is used to provide the user with corresponding touch interaction operation options according to the characteristics of the test questions, so that the user can complete the cognitive assessment independently.

[0072] Touch interaction operations are operations performed by users by touching the screen, such as clicking, sliding, zooming, and the like. The touch interaction gestures in the embodiments of the present invention include clicking, sliding, and pinching. (1) Click touch uses a single finger to lightly touch the screen to select or start an application. The click touch in the present invention mainly supports basic operations such as "start", "finish", "submit", "exit", and "play", as well as supports clicking options in multiple-choice questions, clicking icons in connection questions, and simple reaction tests. (2) Slide touch uses a finger to slide on the screen to scroll or switch interfaces. The slide touch in the present invention mainly supports outlining the curve of a specific object in the question of circling an object in a picture, and is used for drawing in graphic memory. (3) Pinch touch uses two fingers to pinch or separate to zoom in and out of an image.

[0073] The voice interaction unit is used to provide voice information to the test user and understand the test user's voice feedback on the voice information.

[0074] Voice interaction, that is, users use voice and natural language to achieve communication and operation between people and systems. The voice interaction in the present invention involves pure voice interaction, voice interaction combined with images, voice interaction integrated with virtual human technology, etc. Pure voice interaction supports questions and answers, word memory questions, name listing questions, time and space answer questions, calculation combination questions, etc. in the cognitive assessment process. These question types require voice communication between the system and the test user. The communication process includes system inquiry, user response, system understanding, system questioning, user response, system understanding and other processes, among which system understanding, user response, and system questioning may be repeated for multiple rounds. Voice interaction combined with images supports image question and answer questions in the cognitive assessment process. The test user needs to say the items contained in the image according to the image, or describe the content in the image in his own words. Voice interaction integrated with virtual human technology enhances the user's sense of reality and immersion during the autonomous testing process.

[0075] The action interaction unit is used to prompt the test user to perform a specific action in the action question, and to judge the accuracy of the action after the user performs the action.

[0076] The action interaction in the present invention mainly supports action questions, such as the listening and acting questions in the MMSE that test language ability. The user needs to perform corresponding actions according to the system instructions, and the system needs to be able to recognize the action and proceed to the next process.

[0077] Furthermore, the algorithm module is used to intelligently understand the test user's feedback during the cognitive function assessment process and score the relevant feedback in combination with professional standards.

[0078] The algorithm module includes a speech recognition unit, a text matching unit, a text understanding unit, a trajectory recognition unit, a graphic recognition unit, and an action recognition unit. The algorithm module is designed and implemented according to the requirements of the question type.

[0079] Wherein, the speech recognition unit is used to support speech-to-text functions in multiple languages ​​and local dialects.

[0080] The speech recognition unit is equipped with a speech recognition algorithm, which is used to support speech interaction and intelligent evaluation. The present invention supports language recognition of multiple languages ​​and some local dialects. For example, speech recognition of common languages ​​such as Chinese and English uses packaged common algorithm modules; speech recognition algorithm modules for small languages ​​and local dialects need to be implemented through end-to-end training methods such as data collection, model architecture design, large model pre-training, local data fine-tuning, and model deployment.

[0081] In particular, the speech recognition algorithm in this embodiment includes a speech keyword detection algorithm based on a small sample. The goal of this algorithm is to quickly identify and match the reserved keywords in the database from the other party's speech content during the interaction with the user. Even when the user uses a small language or local dialect, the model can iterate quickly.

[0082] Specifically, the present invention trains a neural network model with strong representation ability through a small sample contrast learning algorithm, so that the model can cluster the phonetic representations of the same word in space and separate the phonetic representations of different words in space in different language contexts. The algorithm includes the following steps:

[0083] Step 1, data collection: prepare a vocabulary list. In addition to the vocabulary involved in the database of the present invention, a wider range of vocabulary can also be added. Invite a group with a balanced distribution of gender, age, native language, etc. on the crowdsourcing platform. Each person repeatedly records n audio segments for each vocabulary. n can be set to a relatively small value. Thus, an audio database is obtained.

[0084] Step 2, data enhancement: for the data in the audio database, each piece of data is enhanced. The present invention uses the SpecAugment method at the spectrum level to perform data enhancement, but the enhancement method can also introduce noise addition, rate disturbance, etc. at the time domain level.

[0085] Step 3, model architecture: The present invention obtains spectral features of the original audio, performs data enhancement based on the spectral features, and then realizes vector representation through a multi-layer convolutional neural network (CNN).

[0086] Step 4: Training method: The present invention uses a contrastive learning training method to construct For each audio data x, a positive sample is randomly selected from the audio of the same source language and the same word, denoted as x + , select N-1 negative samples from other audios, denoted as This trains the model.

[0087]

[0088] Step 5, model reasoning: In the application stage, use the trained convolutional neural network to extract audio clip features and calculate the similarity with the audio features in the database. If the similarity is higher than the threshold, it means that the user's answer can be matched.

[0089] The text matching unit is used to compare the test user's answer with the standard answer in the question-and-answer type, and give a corresponding score. The text matching unit is configured with a text matching algorithm, which is suitable for scoring test questions with standard answers, such as question-and-answer questions, word memory questions, name-listing questions, time-space answer questions, calculation combination questions, stroop tests, etc. The algorithm needs to compare the test user's answer with the standard answer to score. The text matching algorithm designed and developed in the present invention includes two types: fuzzy matching and precise matching. Fuzzy matching requires semantic understanding of the test user's reply. As long as it is semantically close to the standard answer or meets the requirements of the question stem, it can be regarded as consistent. The present invention trains a large language model for fuzzy matching. Precise matching requires the longest string matching of the test user's reply content with the standard answer, and the score is determined according to the calculated matching degree. The present invention uses the longest common subsequence matching algorithm based on dynamic programming.

[0090] The text understanding unit is used to understand the test user's answer from a semantic level and give a corresponding score. The text understanding unit is configured with a text understanding algorithm, which is suitable for scoring relatively open questions and answers, such as inquiries about the test user's daily status. The present invention uses a language large model similar to the fuzzy matching algorithm for text understanding.

[0091] In particular, the text understanding algorithm in this embodiment includes a text understanding method based on a large language model (LLM).

[0092] In natural language understanding tasks, there are many techniques for fine-tuning using pre-trained language models (PLMs), and the effect is significantly improved compared to learning from scratch. However, the continuous increase in the size of pre-trained language models (PLMs), for example, from millions to billions or tens of billions of parameters, is a qualitative change. In the process of fine-tuning them, hardware requirements, data requirements and actual costs are also constantly rising. In addition, the abundance of downstream tasks also makes the design of the pre-training and fine-tuning stages very complicated, so we hope to explore a more compact, lightweight, less expensive and more universal fine-tuning method. Prompt (In-context-learning) technology is applied to the present invention.

[0093] The core idea of ​​Prompt technology is to enable the model to complete the task more accurately by designing specific prompts or guiding sentences. Simply put, this technology takes a description of a task and a small number of examples as input, and then generates output using a pre-trained language model. In the cognitive function test, this embodiment designs different prompts for different question types. The following is an example of a word-listing question:

[0094]

[0095] The designed prompt is used as the input of the large language model (LLM), and the model gives the final score. The model input and output are: assistant = LLM (system + user).

[0096] The track recognition unit is used to recognize the test user's operation in the connection question and give a corresponding score; the track recognition unit is configured with a track recognition algorithm, which is suitable for scoring the connection question, and compares the connection result completed by the test user when doing the question with the correct connection plan, so as to give a corresponding score.

[0097] The graphic recognition unit is used to recognize the drawing of the test user in the graphic memory question and give a corresponding score in combination with the standard graphic; the graphic recognition unit is configured with a graphic recognition algorithm, which is suitable for scoring graphic memory questions, such as C-BVMT, C-DSST and other questions that require the test user to draw.

[0098] In particular, the trajectory recognition algorithm in this embodiment includes a handwritten graph representation algorithm, the goal of which is to learn the vector representation of handwritten graphs for judging, comparing and scoring the graphs made by the test user. The model can also adapt quickly to new graphs.

[0099] Specifically, the present invention uses a representation model based on a bottleneck structure, which is divided into an encoder part and a decoder part. The representation model is trained through a self-encoding structure and a self-supervised learning method. The algorithm specifically includes the following steps:

[0100] Step 1: Data preparation: Collect handwritten graphic data of n categories and m samples in each category to form an image database.

[0101] Step 2, model structure: The bottleneck structure is composed of convolutional modules, depthwise separable convolutional modules and fully connected modules. The advantage of this structure is that it can filter out a large amount of redundant information in the sample and retain effective features that focus on handwritten graphics.

[0102] Step 3, training method: (1) Using the bottleneck structure, first compress the sample size of [3,800,800] to [1,25,25] (or [1,100], depending on whether the last layer is convolution or full connection) through the encoder module, and then restore the [1,25,25] feature to a sample size of [3,800,800] through the decoder module. (2) After the model designed based on the bottleneck structure is trained, the encoder part is used as a pre-training model. (3) Add a 2-layer fully connected MLP structure after the encoder part to form a classification network, and perform secondary training to judge the classification network of n categories, effectively improving the model's ability to increase the distance between classes and reduce the distance within classes. (4) The encoder module after secondary training is used as a feature extractor for handwritten graphics. (5) If necessary, the above convolution, separable convolution, and full connection can be replaced by transformer structures.

[0103] Step 3, model reasoning: The handwritten graphics and the standard graphics library are batched and sent to the feature extractor to obtain features f and f1, f2, etc., which are then calculated by the cosine similarity module and the similarities between f1 and f2 are given respectively. A threshold (such as 0.9, 0.95, etc.) is used to determine whether the handwritten graphics meet the requirements.

[0104] The action recognition unit is used to recognize the actions performed by the test user in the action questions and give corresponding scores in combination with the standard actions. The action recognition unit is configured with an action recognition algorithm, which is suitable for scoring action questions. The present invention trains an action recognition model based on the Transformer architecture to identify, compare and score the actions of the test user.

[0105] The intelligent algorithm module in this embodiment involves a variety of intelligent algorithms such as speech recognition, text matching, text understanding, trajectory recognition, action recognition, etc., especially multiple languages ​​and local dialects. On the one hand, it supports the ability to understand the replies of users in different regions, and on the other hand, it uses algorithms to evaluate their replies, making the evaluation results more accurate, objective and reliable.

[0106] Furthermore, the front-end doctor application module is used to provide doctors with an editable cognitive function assessment question generation module.

[0107] The front-end doctor application module is configured on the doctor application side platform. Through this module, doctors can select and combine different cognitive assessment questions according to label indexes as needed, and can adjust the text, voice, image, icon, etc. as needed for specific content.

[0108] Specifically, the front-end doctor application module includes a basic menu unit, a content editing unit, and a tool preview unit.

[0109] The basic menu unit is used to provide menu bars related to basic operations, such as software settings, tag indexes, and database selection related menu bars for users to operate; the content editing unit is used to provide editable digital components for users to generate cognitive assessment tools; the tool preview unit is used to provide a preview of the effect of generating cognitive assessment tools based on the combined digital components. Figure 2 A sample diagram of the system front-end doctor application interface design is given, through which doctors can perform relevant operations.

[0110] Furthermore, the front-end patient application module is configured on the patient application side platform to provide patients with a module that can independently complete cognitive function assessment and can automatically perform assessment.

[0111] After the doctor generates a set of assessment questions based on demand, the system can convert them into a set of intelligent and automated cognitive assessment tools, which generally include information input units, human-computer assessment units, report generation units, etc., allowing patients to use the cognitive assessment tools generated by the doctor's decision completely independently and obtain assessment results.

[0112] Wherein, the information input unit is used to input the basic information of the test user;

[0113] The human-machine evaluation unit is used for the test user to independently perform cognitive evaluation tests;

[0114] The report generating unit is used to intelligently generate a test report after the test user completes the cognitive assessment test.

[0115] See also Figure 3 and Figure 4 , Figure 3 An example diagram of a question-and-answer session integrating voice interaction and virtual human technology on the patient application side of the system front end is given. Figure 4 An example diagram of a graphical interactive question on the patient application side of the system front end is given.

[0116] Further, based on the above embodiment, the embodiment of the present invention encapsulates the specific question into a modular package of "question stem-label-material-interaction-algorithm", and the specific steps are as follows:

[0117] Step 1: assign corresponding scale type, test dimension and question type to each question through label unit. In this embodiment, the knowledge base module stores a large number of cognitive assessment test questions, and the original question stem is used as a template, which is associated with three types of labels in the background database for subsequent indexing.

[0118] Step 2: Bind the adjustable content questions to the corresponding database. The intelligent cognitive function assessment system of the present invention provides doctors with customized options without violating medical professional standards and affecting the assessment results. The database module contains various warehouses such as language, words, and pictures, which can be associated with the corresponding required questions in the background database for subsequent adjustment.

[0119] Step 3: Embed the corresponding human-computer interaction module and algorithm module for each question to complete the modular encapsulation of the question and generate the corresponding digital components.

[0120] In the backend system, digital processing of images, texts, voices, and videos is done for each question, and corresponding human-computer interaction design and integration are done so that the entire application process can be automated.

[0121] At the same time, in the background system, scoring criteria are formulated for each question and integrated into the algorithm module to compare the test user's feedback with the scoring criteria, thereby providing an intelligent evaluation.

[0122] The packaging format of the test questions in this embodiment makes it easy for doctors to select, adjust and combine test questions according to actual needs, so as to adapt to local conditions and individual needs.

[0123] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. An intelligent evaluation system for cognitive function, characterized in that: include: Knowledge base module, database module, human-computer interaction module, algorithm module, front-end doctor application module, front-end patient application module; The knowledge base module is used to store professional data related to cognitive function assessment; The database module is used to store the content of the customizable part of the test questions; The human-computer interaction module is used to receive the user's interactive operation to realize the automation of the cognitive assessment process; The algorithm module is used to understand the test user's feedback during the cognitive function assessment process and score the relevant feedback in combination with professional standards; The front-end doctor application module is used to provide doctors with an editable cognitive function assessment question generation module; The front-end patient application module is used to provide patients with a module that can independently complete cognitive function assessment and can automatically perform assessment.

2. The system according to claim 1, characterized in that The knowledge base module includes a test question stem unit and a label unit; The test question stem unit includes test questions used in clinical cognitive function assessment, which is used to provide a basis for the construction of an intelligent assessment system for cognitive function; The label unit includes a scale type, a test dimension, and a question type, and is used to assign corresponding scale type, test dimension, and question type labels to each test question.

3. The system according to claim 1, characterized in that The database module includes a language library unit, a word library unit and a picture library unit; The language library unit is configured with multiple languages ​​and local dialects, and is used to provide different language options for test questions; The word library unit is configured with different categories of vocabulary, which is used to provide customizable vocabulary categories for test questions with adjustable content; The picture library unit is configured with image and icon content, and is used to provide customizable images and icons for test questions with adjustable content.

4. The system according to claim 1, characterized in that The human-computer interaction module includes a touch interaction unit, a voice interaction unit and a motion interaction unit; The touch interaction unit is used to provide the test user with corresponding touch interaction operation options according to the characteristics of the test questions, so that the test user can complete the cognitive assessment independently; The voice interaction unit is used to provide voice information to the test user and understand the test user's voice feedback on the voice information; The action interaction unit is used to prompt the test user to perform a specific action in the action question, and to judge the accuracy of the action after the test user performs the action.

5. The system according to claim 1, characterized in that The algorithm module includes a speech recognition unit, a text matching unit, a text understanding unit, a trajectory recognition unit, a graphic recognition unit, and an action recognition unit; The speech recognition unit is used to support speech-to-text functions in multiple languages ​​and local dialects; The text matching unit is used to compare the test user's answer with the standard answer in the question-and-answer type and give a corresponding score; The text understanding unit is used to understand the test user's answer from a semantic level and give a corresponding score; The track recognition unit is used to recognize the operation of the test user in the connection question and give a corresponding score; The graphic recognition unit is used to recognize the drawing of the test user in the graphic memory question and give a corresponding score in combination with the standard graphic; The action recognition unit is used to recognize the actions performed by the test user in the action questions and give corresponding scores in combination with standard actions.

6. The system according to claim 1, characterized in that The front-end doctor application module is configured on the doctor application side platform, including a basic menu unit, a content editing unit, and a tool preview unit; The basic menu unit is used to provide a menu bar related to basic operations for users to operate; The content editing unit is used to provide editable digital components for users to generate cognitive assessment tools; The tool preview unit is used to provide a preview of the effect of generating a cognitive assessment tool according to the combined digital components.

7. The system according to claim 1, characterized in that The front-end patient application module is configured on the patient application side platform, including an information input unit, a human-machine evaluation unit, and a report generation unit; The information input unit is used to input the basic information of the test user; The human-machine evaluation unit is used for the test user to independently perform cognitive evaluation tests; The report generating unit is used to intelligently generate a test report after the test user completes the cognitive assessment test.

8. The system according to claim 1, characterized in that The encapsulation process of the questions in the intelligent assessment system includes: Assign the corresponding scale type, test dimension and question type to each question through label units; Bind the topics with adjustable content to the corresponding database; The corresponding human-computer interaction module and algorithm module are embedded in each question to complete the modular encapsulation of the question and generate the corresponding digital components.

Citation Information

Cited By

  • Rehabilitation training evaluation system for dealing with schizophrenia patients

    CN120199504A