WO2023001301A1 - 基于用户能力的个性化认知训练任务推荐方法及系统 - Google Patents

基于用户能力的个性化认知训练任务推荐方法及系统 Download PDF

Info

Publication number
WO2023001301A1
WO2023001301A1 PCT/CN2022/107482 CN2022107482W WO2023001301A1 WO 2023001301 A1 WO2023001301 A1 WO 2023001301A1 CN 2022107482 W CN2022107482 W CN 2022107482W WO 2023001301 A1 WO2023001301 A1 WO 2023001301A1
Authority
WO
WIPO (PCT)
Prior art keywords
training
ability
task
user
score
Prior art date
Application number
PCT/CN2022/107482
Other languages
English (en)
French (fr)
Inventor
王晓怡
边志明
褚铮
Original Assignee
北京智精灵科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京智精灵科技有限公司 filed Critical 北京智精灵科技有限公司
Publication of WO2023001301A1 publication Critical patent/WO2023001301A1/zh
Priority to US18/420,648 priority Critical patent/US20240212825A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G16H20/70—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to mental therapies, e.g. psychological therapy or autogenous training
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients

Definitions

  • the invention relates to a method for recommending personalized cognitive training tasks based on user capabilities, and also relates to a corresponding personalized cognitive training task recommendation system, which belongs to the technical field of personalized recommendation.
  • personalized recommendation technology has attracted a large number of scholars' research. At present, this recommendation technology has been widely used in various large-scale multimedia and e-commerce websites, such as Amazon, Jingdong, Google News and Taobao.
  • Personalized recommendation techniques are mainly divided into three categories: content-based recommendation, collaborative filtering-based recommendation, and hybrid-based recommendation. Among them, the recommendation method based on collaborative filtering is the most widely used.
  • the primary technical problem to be solved by the present invention is to provide a personalized cognitive training task recommendation method based on user capabilities, which is reasonably designed, comprehensively considers multiple factors, and has high recommendation result accuracy.
  • Another technical problem to be solved by the present invention is to provide a personalized cognitive training task recommendation system based on user capabilities.
  • a method for recommending personalized cognitive training tasks based on user capabilities including the following steps:
  • the establishment of the machine initial test recommendation task training list includes the following steps:
  • S1-1 Establish a general cognitive ability model, and construct an ability weight matrix (L ⁇ K) and a training task ability weight matrix (T ⁇ K) according to the general cognitive ability model, and then compare the ability weight matrix and the training task ability The weights in the weight matrix are standardized, where,
  • x is a vector composed of individual scales or comprehensive scale items
  • ⁇ LK is a vector composed of cognitive ability
  • ⁇ X is the relationship between the scale and cognitive ability, and is the factor of the scale on cognitive ability Load matrix
  • ⁇ LK is the error of scale measurement
  • y is the vector composed of training tasks
  • ⁇ TK is the vector composed of cognitive abilities
  • ⁇ y is the relationship between training tasks and cognitive abilities, and is the cognitive ability of training tasks
  • the factor loading matrix of ; ⁇ TK is the error in the calculation of the training task
  • L represents a set of scales
  • K represents a set of capabilities
  • wmn represents the weight of scale lm ⁇ L to capability kn ⁇ K, wmn ⁇ L ⁇ K
  • T the training task set
  • K the ability set
  • rfj the weight of the training task tf ⁇ T to the ability kj ⁇ K, rfj ⁇ T ⁇ K;
  • S1-2 Refers to administering the test through the pre-set scale of the system, collecting data according to the answers to the scale, and constructing the correlation score wuk between user u and ability k,
  • G(u,i) represents the ability shared by user u and training task i
  • Wuk represents the association score between user u and ability k
  • rfj represents the weight of training task tf ⁇ T to ability kj ⁇ K
  • Set I(u) refers to the set of all task training in the user's current system. This set will continue to increase with system upgrades. For example, the 2.0 system has a total of 77 task training;
  • the establishment of the manual evaluation recommendation task training list includes the following steps:
  • I represents the training task set
  • D represents the disease set
  • wid represents the weight of training task i ⁇ I to disease d ⁇ D, wid ⁇ I ⁇ D;
  • H(u) represents the set of diseases diagnosed by the user
  • Wid represents the correlation score of training task i and disease d
  • Qui represents the correlation score of user u and disease d
  • the set I'(u) refers to the set of general task training corresponding to the disease based on literature or experience. It can be the union of multiple diseases. This set will continue to change with the deepening of disease cognition.
  • the training tasks include catching sparrows in the sky, making a final decision, searching for branches and picking fruits, etc.;
  • the establishment of the optimal recommended task training list includes the steps: the training task is recommended through double evaluation, and the matching degree Pui and P'ui of the user u to the training task i are respectively weighted and then summed to calculate the recommendation score S of each training task ,
  • a represents the weight corresponding to the initial test by the machine
  • b represents the weight corresponding to the manual evaluation
  • each training task is ranked according to the score
  • the order of the score is the ranking of the recommended training tasks.
  • association score wuk between the user u and the ability k is constructed through the following steps:
  • the user completes the pre-set n comprehensive evaluation sub-units of the machine preliminary test in turn, and generates original scores Xn respectively;
  • the value range of i is 1 ⁇ n;
  • Wuk is the comprehensive evaluation subunit score after standardized conversion, that is, the correlation score between user u and ability k. , the greater the value of Wuk;
  • Xi is the original score in the comprehensive evaluation subunit; is the average of the original scores of the comprehensive assessment subunits of the healthy population matched with the age, gender, occupation, and educational level of the subjects; the standard deviation of the unit raw score; and ⁇ i are also called comprehensive norm parameters of healthy users;
  • wmn represents the weight of scale lm ⁇ L to ability kn ⁇ K.
  • a personalized cognitive training task recommendation system based on user capabilities is provided, including a machine preliminary test module, a manual evaluation module and a comprehensive recommendation module,
  • the machine preliminary test module is used to establish a machine preliminary test recommended task training list, and the machine preliminary test recommended task training list is established through the following steps:
  • S1-1 Establish a general cognitive ability model, and construct an ability weight matrix (L ⁇ K) and a training task ability weight matrix (T ⁇ K) according to the general cognitive ability model, and then compare the ability weight matrix and the training task ability The weights in the weight matrix are standardized, where,
  • x is a vector composed of individual scales or comprehensive scale items
  • ⁇ LK is a vector composed of cognitive ability
  • ⁇ X is the relationship between the scale and cognitive ability, and is the factor of the scale on cognitive ability Load matrix
  • ⁇ LK is the error of scale measurement
  • y is the vector composed of training tasks
  • ⁇ TK is the vector composed of cognitive abilities
  • ⁇ y is the relationship between training tasks and cognitive abilities, and is the cognitive ability of training tasks
  • the factor loading matrix of ; ⁇ TK is the error in the calculation of the training task
  • L represents a set of scales
  • K represents a set of capabilities
  • wmn represents the weight of scale lm ⁇ L to capability kn ⁇ K, wmn ⁇ L ⁇ K
  • T the training task set
  • K the ability set
  • rfj the weight of the training task tf ⁇ T to the ability kj ⁇ K, rfj ⁇ T ⁇ K;
  • S1-2 Refers to administering the test through a pre-set scale, collecting data according to the answers to the scale, and constructing the correlation score wuk between user u and ability k,
  • G(u,i) represents the ability shared by user u and training task i
  • Wuk represents the association score between user u and ability k
  • rfj represents the weight of training task tf ⁇ T to ability kj ⁇ K
  • S1-4 Establish a training list of recommended tasks for the initial test of the machine: sort all the training tasks in the set I(u) according to the matching degree of the user.
  • the set I(u) refers to the set of training for all tasks in the user's current system
  • the manual evaluation module is used to establish a manual evaluation recommended task training list, and the manual evaluation recommended task training list is established through the following steps:
  • S2-1 Establish a disease training model, determine the degree of association wid between training task i and disease d, construct a disease training task weight matrix (I ⁇ D) and perform standardized processing, where,
  • I represents the training task set
  • D represents the disease set
  • wid represents the weight of training task i ⁇ I to disease d ⁇ D, wid ⁇ I ⁇ D;
  • H(u) represents the set of diseases diagnosed by the user
  • Wid represents the correlation score of training task i and disease d
  • Qui represents the correlation score of user u and disease d
  • S2-4 Establish a manual evaluation and recommendation task training list, and perform weighting, deduplication, and sorting on all training tasks in the set I'(u) according to the user matching degree;
  • the comprehensive recommendation module is used to set up the optimal recommended task training list, and establishes the optimal recommended task training list through the following steps:
  • the matching degrees Pui and P’ui of user u to training task i are weighted and summed respectively, and the recommendation score S of each training task is calculated.
  • a represents the weight corresponding to the initial test by the machine
  • b represents the weight corresponding to the manual evaluation.
  • Each training task is arranged according to the score, and the ranking order of the scores is the ranking of the recommended training tasks.
  • the present invention improves the accuracy of the cognitive training recommendation method through an optimization algorithm, it can perform personalized recommended cognitive training tasks based on the user's ability, and achieve mutual interaction of treatment plans through the combination of machine algorithms and manual evaluation. Supplement, balance machine and human issues, reduce decision errors.
  • the invention is based on continuous enrichment of cognitive task training, and has strong compatibility and expandability.
  • FIG. 1 is a flowchart of a method for recommending personalized cognitive training tasks based on user capabilities in an embodiment of the present invention
  • Fig. 2 is an example diagram of building a cognitive ability according to an embodiment of the present invention
  • Fig. 3 is another example diagram of constructing cognitive ability according to an embodiment of the present invention.
  • FIG. 4 is a schematic structural diagram of a personalized cognitive training task recommendation system based on user capabilities in an embodiment of the present invention.
  • the establishment of the machine initial test recommendation task training list includes the following steps:
  • S1-1 Establish a general cognitive ability model, and construct an ability weight matrix (L ⁇ K) and a training task ability weight matrix (T ⁇ K) according to the general cognitive ability model, and then compare the ability weight matrix and the training task ability The weights in the weight matrix are standardized, where,
  • x is a vector composed of individual scales or comprehensive scale items
  • ⁇ LK is a vector composed of cognitive ability
  • ⁇ X is the relationship between the scale and cognitive ability, and is the factor of the scale on cognitive ability Load matrix
  • ⁇ LK is the error of scale measurement
  • y is the vector composed of training tasks
  • ⁇ TK is the vector composed of cognitive abilities
  • ⁇ y is the relationship between training tasks and cognitive abilities, and is the cognitive ability of training tasks
  • the factor loading matrix of ; ⁇ TK is the error in the calculation of the training task
  • L represents a set of scales
  • K represents a set of capabilities
  • wmn represents the weight of scale lm ⁇ L to capability kn ⁇ K, wmn ⁇ L ⁇ K
  • T the training task set
  • K the ability set
  • rfj the weight of the training task tf ⁇ T to the ability kj ⁇ K, rfj ⁇ T ⁇ K;
  • S1-2 Refers to administering the test through the pre-set scale of the system, collecting data according to the answers to the scale, and constructing the correlation score wuk between user u and ability k,
  • G(u,i) represents the ability shared by user u and training task i
  • Wuk represents the association score between user u and ability k
  • rfj represents the weight of training task tf ⁇ T to ability kj ⁇ K
  • S1-4 Establish the training list of recommended tasks for the machine preliminary test: sort all the training tasks in the set I(u) according to the user matching degree,
  • Set I(u) refers to the training set for all tasks in the user's current system. This set will continue to increase with system upgrades. For example, the 2.0 system has a total of 77 task trainings.
  • the establishment of a manual evaluation recommendation task training list includes the following steps:
  • S2-1 Establish a disease training model: determine the degree of association wid between the training task i and the disease d, construct the disease training task weight matrix (I ⁇ D) and perform standardized processing, where,
  • I represents the training task set
  • D represents the disease set
  • wid represents the weight of training task i ⁇ I to disease d ⁇ D, wid ⁇ I ⁇ D;
  • H(u) represents the set of diseases diagnosed by the user
  • Wid represents the correlation score between training task i and disease d
  • Qui represents the correlation score between user u and disease d.
  • the set I'(u) refers to the set of general task training corresponding to the disease based on literature or experience. It can be the union of multiple diseases. This set will continue to change with the deepening of disease cognition.
  • the training tasks include catching sparrows in the sky, making a final decision, and picking branches and fruits.
  • the establishment of the optimal recommended task training list includes the steps: the training task is recommended through double evaluation, and the matching degree Pui and P'ui of the user u to the training task i are respectively weighted and then summed to calculate the recommendation score S of each training task ,
  • a represents the weight corresponding to the initial test of the machine
  • b represents the weight corresponding to the manual evaluation
  • each training task is arranged according to the score
  • the sequence rule is the optimal recommendation.
  • association score wuk of the user u and ability k is constructed through the following steps:
  • the user completes the pre-set n comprehensive evaluation sub-units of the machine preliminary test in turn, and generates original scores Xn respectively;
  • the value range of i is 1 ⁇ n;
  • Wuk is the comprehensive evaluation subunit score after standardized conversion, that is, the correlation score between user u and ability k. , the greater the value of Wuk;
  • Xi is the original score in the comprehensive evaluation subunit; is the average of the original scores of the comprehensive assessment subunits of the healthy population matched with the age, gender, occupation, and educational level of the subjects; the standard deviation of the unit raw score; and ⁇ i are also called comprehensive norm parameters of healthy users;
  • wmn represents the weight of scale lm ⁇ L to ability kn ⁇ K.
  • the method for recommending personalized cognitive training tasks based on user capabilities includes at least the following steps:
  • step S1 establishes the training list of recommended tasks for the machine preliminary test, including the following steps
  • S1-1 Establish a general cognitive ability model, build an ability weight matrix based on the research results of intelligence theory and brain function network, train the task ability weight matrix, and then standardize the weights in the ability weight matrix and the training task ability weight matrix.
  • x is a vector composed of individual scales or comprehensive scale items
  • ⁇ LK is a vector composed of cognitive ability
  • ⁇ X is the relationship between the scale and cognitive ability, and is the factor of the scale on cognitive ability Load matrix
  • ⁇ LK is the error on the scale measurement.
  • y is a vector composed of training tasks
  • ⁇ TK is a vector composed of cognitive abilities
  • ⁇ y is the relationship between training tasks and cognitive abilities, and is the factor load matrix of training tasks on cognitive abilities
  • ⁇ TK is the calculation of training tasks error.
  • x BNT is x
  • BNT is the abbreviation of naming sub-test in MoCA test
  • ⁇ X 0.81
  • ⁇ BNT-Gc is ⁇ LK
  • BNT is naming sub-test in MoCA test
  • Gc is the abbreviation of ability to understand knowledge
  • ⁇ BNT-Gc is ⁇ LK
  • the value is 0, the expression of naming subtests and ability to understand knowledge in MoCA test is:
  • the abbreviation of ⁇ 30001-Gv is ⁇ TK, the value is 0, the expression of whack-a-mole and ability visual processing is:
  • the ability weight matrix (L ⁇ K) is constructed, where L represents the scale set; K represents the ability set; wmn represents the weight of the scale lm ⁇ L to the ability kn ⁇ K, wmn ⁇ L ⁇ K.
  • the scale set L is the Naming sub-test, the number sequence sub-test, and the sentence repetition sub-test in the MoCA test, with a total of 3 elements.
  • Ability set K is knowledge comprehension, visual processing, fluid reasoning, short-term memory, and processing speed, a total of 5 elements.
  • T represents the training task set
  • K represents the capability set
  • rfj represents the weight of the training task tf ⁇ T to the capability kj ⁇ K, rfj ⁇ T ⁇ K.
  • the scale set T is a whack-a-mole, a moving point click, a total of 2 elements.
  • Ability set K is knowledge comprehension, visual processing, fluid reasoning, short-term memory, and processing speed, a total of 5 elements.
  • S1-2 Calculate the user's own ability status through the preliminary test of the machine.
  • the preliminary test of the machine refers to the measurement through the scale set in advance by the system, and the data is collected according to the answers to the scale.
  • the measurement results will be intelligently analyzed and analyzed. Compare the norms to determine the user's own ability status, and construct the correlation score wuk between user u and ability k.
  • the tests conducted through the pre-set scales of the system include MoCA, MMSE, ADL, PHQ-9, GAD-7, life satisfaction index and other clinically widely used scales and task-based assessment tools.
  • association score wuk of the user u and ability k is constructed through the following steps:
  • the user completes the pre-set comprehensive evaluation sub-units 1, 2, 3, ... n evaluations in turn, and generates original scores X01, X02, X03, ... Xn.
  • the recommendation system provided by the present invention will automatically extract the original scores of subjects in n comprehensive evaluation subunits, and then standardize and convert the original scores in the comprehensive evaluation subunits according to the comprehensive norm parameters of healthy users to generate user
  • the correlation score wuk between u and ability k the formula is as follows:
  • Wuk is the comprehensive evaluation sub-unit score after standardized conversion, that is, the correlation score between user u and ability k. If the user is measured to have poor ability, the more the deviation from the norm, the greater the value of Wuk;
  • Xi is the original score in the comprehensive assessment subunit
  • ⁇ i is the standard deviation of the original score of the subunit of the comprehensive assessment of the healthy population matched with the subject's age, gender, occupation, and education level;
  • ⁇ i are also called comprehensive norm parameters of healthy users
  • wmn represents the weight of scale lm ⁇ L to ability kn ⁇ K.
  • G(u,i) represents the ability shared by user u and training task i;
  • Wuk represents the correlation score between user u and ability k
  • rfj represents the weight of the training task tf ⁇ T to the capability kj ⁇ K.
  • S1-4 Establish a training list of recommended tasks for the initial test of the machine: sort all the training tasks in the set I(u) according to the degree of user matching. Take the Top-N items and assign explanations to each training task. Save the Top-N training tasks to the recommended task training list for the initial test of the machine. For example, the ranking of moving point click is higher than that of whack-a-mole.
  • the establishment of the manual evaluation recommendation task training list includes the following steps.
  • S2-1 Establish a disease training model, determine the correlation degree wid between training task i and disease d according to disease pathological characteristics and rehabilitation training literature, construct a disease training task weight matrix (I ⁇ D) and perform standardized processing.
  • I ⁇ D disease training task weight matrix
  • I represents the set of training tasks
  • D represents the disease set
  • wid represents the weight of training task i ⁇ I to disease d ⁇ D, wid ⁇ I ⁇ D;
  • H(u) represents the collection of diseases diagnosed by users
  • Wid represents the correlation score between training task i and disease d
  • Qui represents the association score between user u and disease d.
  • S2-4 Establish a training list for manual evaluation and recommendation tasks: weight, deduplicate, and sort all training tasks in the set I'(u) according to the user matching degree. Take the Top-N items and assign explanations to each training task. Save the Top-N training tasks to the manual evaluation recommendation task training list, for example, the ranking of moving point click is higher than that of whack-a-mole.
  • the training tasks are recommended through double evaluation, and usually the user will accumulate a large number of training tasks.
  • the matching degrees Pui and P'ui of user u to training task i are weighted and summed respectively, and the recommendation score S of each training task is calculated.
  • a represents the weight corresponding to the initial test of the machine
  • b represents the weight corresponding to the manual evaluation.
  • the embodiment of the present invention also provides a personalized cognitive training task recommendation system based on user capabilities, including a processor module, a power supply module, a human-computer interaction module, a communication module, an input module, an evaluation module, a storage module, calculation module and output module, the above-mentioned power supply module, human-computer interaction module, communication module, input module, evaluation module, storage module, calculation module and output module are respectively connected to the processor module, and the evaluation module includes the machine initial test module , a manual evaluation module and a comprehensive difficulty recommendation module.
  • the machine preliminary test module is used to establish a machine preliminary test recommended task training list, and the machine preliminary test recommended task training list is established through the following steps:
  • S1-1 Establish a general cognitive ability model, and construct an ability weight matrix (L ⁇ K) and a training task ability weight matrix (T ⁇ K) according to the general cognitive ability model, and then compare the ability weight matrix and the training task ability The weights in the weight matrix are standardized, where,
  • x is a vector composed of individual scales or comprehensive scale items
  • ⁇ LK is a vector composed of cognitive ability
  • ⁇ X is the relationship between the scale and cognitive ability, and is the factor of the scale on cognitive ability Load matrix
  • ⁇ LK is the error of scale measurement
  • y is the vector composed of training tasks
  • ⁇ TK is the vector composed of cognitive abilities
  • ⁇ y is the relationship between training tasks and cognitive abilities, and is the cognitive ability of training tasks
  • the factor loading matrix of ; ⁇ TK is the error in the calculation of the training task
  • L represents a set of scales
  • K represents a set of capabilities
  • wmn represents the weight of scale lm ⁇ L to capability kn ⁇ K, wmn ⁇ L ⁇ K
  • T the training task set
  • K the ability set
  • rfj the weight of the training task tf ⁇ T to the ability kj ⁇ K, rfj ⁇ T ⁇ K;
  • S1-2 Refers to administering the test through a pre-set scale, collecting data according to the answers to the scale, and constructing the correlation score wuk between user u and ability k,
  • G(u,i) represents the ability shared by user u and training task i
  • Wuk represents the association score between user u and ability k
  • rfj represents the weight of training task tf ⁇ T to ability kj ⁇ K
  • S1-4 Establish a training list of recommended tasks for the initial test of the machine: sort all the training tasks in the set I(u) according to the matching degree of the user.
  • the set I(u) refers to the set of training for all tasks in the user's current system
  • the manual evaluation module is used to establish a manual evaluation recommended task training list, and the manual evaluation recommended task training list is established through the following steps:
  • S2-1 Establish a disease training model, determine the degree of association wid between training task i and disease d, construct a disease training task weight matrix (I ⁇ D) and perform standardized processing, where,
  • I represents the training task set
  • D represents the disease set
  • wid represents the weight of training task i ⁇ I to disease d ⁇ D, wid ⁇ I ⁇ D;
  • H(u) represents the set of diseases diagnosed by the user
  • Wid represents the correlation score of training task i and disease d
  • Qui represents the correlation score of user u and disease d
  • S2-4 Establish a manual evaluation and recommendation task training list, and perform weighting, deduplication, and sorting on all training tasks in the set I'(u) according to the user matching degree;
  • the comprehensive recommendation module is used to set up the optimal recommended task training list, and establishes the optimal recommended task training list through the following steps:
  • the matching degrees Pui and P’ui of user u to training task i are weighted and summed respectively, and the recommendation score S of each training task is calculated.
  • a represents the weight corresponding to the initial test of the machine
  • b represents the weight corresponding to the manual evaluation.
  • Each training task is arranged according to the score. This order rule is the optimal recommendation.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Epidemiology (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Biomedical Technology (AREA)
  • Pathology (AREA)
  • Theoretical Computer Science (AREA)
  • Psychiatry (AREA)
  • Developmental Disabilities (AREA)
  • Psychology (AREA)
  • Social Psychology (AREA)
  • Hospice & Palliative Care (AREA)
  • Child & Adolescent Psychology (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Medical Treatment And Welfare Office Work (AREA)

Abstract

本发明公开了一种基于用户能力的个性化认知训练任务推荐方法及系统。该方法包括如下步骤：建立机器初测推荐任务训练列表；建立人工评价推荐任务训练列表；以及合并排序，建立最优推荐任务训练列表。本发明可以基于用户的能力进行个性化的推荐认知训练任务，并通过机器算法和人工评价相结合的方式做到治疗方案的互为补充，平衡机器和人为的问题，减少决策误差。

Description

基于用户能力的个性化认知训练任务推荐方法及系统 技术领域
本发明涉及一种基于用户能力的个性化认知训练任务推荐方法，同时也涉及相应的个性化认知训练任务推荐系统，属于个性化推荐技术领域。
背景技术
个性化推荐技术作为一种重要的信息过滤技术和手段，吸引了大量学者的研究。目前，该项推荐技术已经被广泛应用在各种大型多媒体和电子商务网站上，如亚马逊、京东、谷歌新闻和淘宝等。个性化推荐技术主要分为三类：基于内容的推荐、基于协同过滤的推荐和基于混合的推荐。其中，基于协同过滤的推荐方法是应用最广泛的。
为了向用户及时准确地推荐满足其个性化需求的认知训练，研究人员尝试将个性化推荐技术引入到医疗康复行业中。例如，有研究人员提出了一种基于疾病的个性化推荐技术。该技术可以通过人工判定用户的疾病状况，然后为其推荐训练任务。但是，现有的认知训练推荐方法没有考虑用户认知功能的具体受损情况，在准确度方面仍然存在很大的提升空间。
发明内容
本发明所要解决的首要技术问题在于提供一种设计合理、综合考虑多因素且推荐结果准确度高的基于用户能力的个性化认知训练任务推荐方法。
本发明所要解决的另一技术问题在于提供一种基于用户能力的个性化认知训练任务推荐系统。
为了实现上述目的，本发明采用下述的技术方案：
根据本发明实施例的第一方面，提供一种基于用户能力的个性化认知训练任务推荐方法，包括以下步骤：
建立机器初测推荐任务训练列表；
建立人工评价推荐任务训练列表；以及
合并排序，建立最优推荐任务训练列表，
其中，所述建立机器初测推荐任务训练列表包括以下步骤：
S1-1：建立一般认知能力模型，并根据所述一般认知能力模型构建能力权重矩阵(L×K)以及训练任务能力权重矩阵(T×K)，然后对能力权重矩阵和训练任务能力权重矩阵中的权重进行标准化处理，其中，
建立一般认知能力结构方程模型的公式如下：
x＝ΛXηLK+εLK
y＝ΛyηTK+εTK
其中，x是单项量表或综合性量表分项组成的向量，ηLK是认知能力组成的向量，ΛX是量表与认知能力之间的关系，是量表在认知能力上的因子负荷矩阵，εLK是量表测量上的误差，y是训练任务组成的向量，ηTK是认知能力组成的向量，Λy是训练任务与认知能力之间的关系，是训练任务在认知能力上的因子负荷矩阵；εTK是训练任务计算上的误差，
对于所述能力权重矩阵(L×K)，L表示量表集合，K表示能力集合；wmn表示量表lm∈L对能力kn∈K的权重，wmn∈L×K，
对于所述训练任务能力权重矩阵(T×K)，T表示训练任务集合，K表示能力集合，rfj表示训练任务tf∈T对能力kj∈K的权重，rfj∈T×K；
S1-2：指通过系统预先设置好的量表进行施测，根据量表作答情况来收集数据，构建用户u和能力k的关联分数wuk，
S1-3：计算用户u对训练任务i的匹配度Pui，公式如下：
Pui＝∑Wukrfj
k∈G(u,i)
其中，G(u,i)表示用户u和训练任务i共有的能力，Wuk表示用户u和能力k的关联分数，rfj表示训练任务tf∈T对能力kj∈K的权重；
S1-4：建立机器初测推荐任务训练列表：对集I(u)中的所有训练任务进行按照用户匹配度进行排序，其中，
集I(u)是指针对用户目前系统中所有任务训练的集合，这个集会随着系统升级不断增加，如2.0系统共有77个任务训练；
所述建立人工评价推荐任务训练列表包括以下步骤：
S2-1：建立疾病训练模型：确定训练任务i和疾病d的关联度wid，构建疾病训练任务权重矩阵(I×D)并进行标准化处理，其中，
I表示训练任务集合，D表示疾病集合，
wid表示训练任务i∈I对疾病d∈D的权重，wid∈I×D；
S2-2：基于用户已持有的医学检测结果，确定用户u和疾病d的关联性分数Qui分数，分数范围在1～100分，du∈D；
S2-3计算用户u对训练任务i的匹配度P’ui，公式如下：
P’ui＝∑QuiWid
d∈H(u)
其中，H(u)表示用户确诊的疾病的集合，Wid表示训练任务i和疾病d的关联性分数，Qui表示用户u和疾病d的关联性分数，
S2-4：建立人工评价推荐任务训练列表：对集I’(u)中的所有训练任务进行按照用户匹配度进行加权、去重、排序，其中，
集I’(u)是指基于文献或经验的疾病对应的一般任务训练的集合，可以是多个疾病的并集，这个集会随着对疾病认知的加深而不断变化，如偏侧忽略可训练的任务包括长空捕雀、一锤定音、寻枝摘果等；
所述建立最优推荐任务训练列表包括步骤：训练任务经过双重测评推荐，将用户u对训练任务i的匹配度Pui和P’ui分别进行加权再求和，计算每个训练任务的推荐得分S，
S＝aPui+bP’ui
其中a表示机器初测对应的权重；b表示人工评价对应的权重，将每个训练任务按得分高低排列，该得分排列顺序即为推荐训练任务的排序。
其中较优地，通过以下步骤构建所述用户u和能力k的关联分数wuk：
用户依次完成机器初测预先设定的n个综合性测评子单元，并分别生成原始分数Xn；
提取受试者在n个综合性测评子单元中的原始分数，然后根据健康用户综合常模参数对综合性测评子单元中的原始分数进行标准化转 换，生成用户u和能力k的关联分数wuk，公式如下：
Figure PCTCN2022107482-appb-000001
其中，i的取值范围为1～n；Wuk是进行标准化转换后的综合性测评子单元分数，即用户u和能力k的关联分数，如果用户被测量出能力较差，偏离常模越多，Wuk的值越大；Xi是综合性测评子单元中的原始分；
Figure PCTCN2022107482-appb-000002
是与受试者年龄、性别、职业、教育程度匹配的健康人群综合性测评子单元原始得分的平均值；σi是与受试者年龄、性别、职业、教育程度匹配的健康人群综合性测评子单元原始得分的标准差；
Figure PCTCN2022107482-appb-000003
和σi又叫做健康用户综合常模参数；wmn表示量表lm∈L对能力kn∈K的权重。
根据本发明实施例的第二方面，提供一种基于用户能力的个性化认知训练任务推荐系统，包括机器初测模块、人工评价模块和综合推荐模块，
其中，所述机器初测模块用于建立机器初测推荐任务训练列表，通过以下步骤建立机器初测推荐任务训练列表：
S1-1：建立一般认知能力模型，并根据所述一般认知能力模型构建能力权重矩阵(L×K)以及训练任务能力权重矩阵(T×K)，然后对能力权重矩阵和训练任务能力权重矩阵中的权重进行标准化处理，其中，
建立一般认知能力结构方程模型的公式如下：
x＝ΛXηLK+εLK
y＝ΛyηTK+εTK
其中，x是单项量表或综合性量表分项组成的向量，ηLK是认知能力组成的向量，ΛX是量表与认知能力之间的关系，是量表在认知能力上的因子负荷矩阵，εLK是量表测量上的误差，y是训练任务组成的向量，ηTK是认知能力组成的向量，Λy是训练任务与认知能力之间的关系，是训练任务在认知能力上的因子负荷矩阵；εTK是训练任务计算上的误差，
对于所述能力权重矩阵(L×K)，L表示量表集合，K表示能力集合；wmn表示量表lm∈L对能力kn∈K的权重，wmn∈L×K，
对于所述训练任务能力权重矩阵(T×K)，T表示训练任务集合，K表示能力集合，rfj表示训练任务tf∈T对能力kj∈K的权重，rfj∈T×K；
S1-2：指通过预先设置好的量表进行施测，根据量表作答情况来收集数据，构建用户u和能力k的关联分数wuk，
S1-3：计算用户u对训练任务i的匹配度Pui，公式如下：
Pui＝∑Wukrfj
k∈G(u,i)
其中，G(u,i)表示用户u和训练任务i共有的能力，Wuk表示用户u和能力k的关联分数，rfj表示训练任务tf∈T对能力kj∈K的权重；
S1-4：建立机器初测推荐任务训练列表：对集I(u)中的所有训练任务进行按照用户匹配度进行排序，集I(u)是指针对用户目前系统中所有任务训练的集合，
所述人工测评模块用于建立人工评价推荐任务训练列表，通过以下步骤建立人工评价推荐任务训练列表：
S2-1：建立疾病训练模型，确定训练任务i和疾病d的关联度wid，构建疾病训练任务权重矩阵(I×D)并进行标准化处理，其中，
I表示训练任务集合，D表示疾病集合，
wid表示训练任务i∈I对疾病d∈D的权重，wid∈I×D；
S2-2：基于用户已持有的医学检测结果，确定用户u和疾病d的关联性分数Qui分数，分数范围在1～100分，du∈D；
S2-3计算用户u对训练任务i的匹配度P’ui，公式如下：
P’ui＝∑QuiWid
d∈H(u)
其中，H(u)表示用户确诊的疾病的集合，Wid表示训练任务i和疾病d的关联性分数，Qui表示用户u和疾病d的关联性分数，
S2-4：建立人工评价推荐任务训练列表，对集I’(u)中的所有训练任务进行按照用户匹配度进行加权、去重、排序；
所述综合推荐模块用于建立最优推荐任务训练列表，通过以下步骤建立最优推荐任务训练列表：
训练任务经过双重测评推荐，将用户u对训练任务i的匹配度Pui和P’ui分别进行加权再求和，计算每个训练任务的推荐得分S，
S＝aPui+bP’ui
其中a表示机器初测对应的权重，b表示人工评价对应的权重，将每个训练任务按得分高低排列，该得分配列顺序即为推荐训练任务的排序。
本发明通过优化算法提高了认知训练推荐方法的准确度，它能够基于用户的能力进行个性化的推荐认知训练任务，并通过机器算法和人工评价相结合的方式做到治疗方案的互为补充，平衡机器和人为的问题，减少决策误差。本发明基于认知任务训练的不断丰富，具有较强的兼容性和可拓展性。
附图说明
图1为本发明实施例中，基于用户能力的个性化认知训练任务推荐方法的流程图；
图2为根据本发明的实施例的构建认知能力的实例图；
图3为根据本发明的实施例的另一构建认知能力的实例图；
图4为本发明实施例中，基于用户能力的个性化认知训练任务推荐系统的结构示意图。
具体实施方式
下面结合附图和具体实施例对本发明的技术内容进行详细具体地说明。
为使本发明的目的、技术方案和优点更加清楚，下面将结合本发明具体实施例及相应的附图对本发明技术方案进行清楚、完整地描述。显然，所描述的实施例仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
本发明实施例所提供的基于用户能力的个性化认知训练任务推荐方法，至少包括以下步骤：
建立机器初测推荐任务训练列表；
建立人工评价推荐任务训练列表；以及
合并排序，建立最优推荐任务训练列表，
其中，所述建立机器初测推荐任务训练列表包括以下步骤：
S1-1：建立一般认知能力模型，并根据所述一般认知能力模型构建能力权重矩阵(L×K)以及训练任务能力权重矩阵(T×K)，然后对能力权重矩阵和训练任务能力权重矩阵中的权重进行标准化处理，其中，
建立一般认知能力结构方程模型的公式如下：
x＝ΛXηLK+εLK
y＝ΛyηTK+εTK
其中，x是单项量表或综合性量表分项组成的向量，ηLK是认知能力组成的向量，ΛX是量表与认知能力之间的关系，是量表在认知能力上的因子负荷矩阵，εLK是量表测量上的误差，y是训练任务组成的向量，ηTK是认知能力组成的向量，Λy是训练任务与认知能力之间的关系，是训练任务在认知能力上的因子负荷矩阵；εTK是训练任务计算上的误差，
对于所述能力权重矩阵(L×K)，L表示量表集合，K表示能力集合；wmn表示量表lm∈L对能力kn∈K的权重，wmn∈L×K，
对于所述训练任务能力权重矩阵(T×K)，T表示训练任务集合，K表示能力集合，rfj表示训练任务tf∈T对能力kj∈K的权重，rfj∈T×K；
S1-2：指通过系统预先设置好的量表进行施测，根据量表作答情况来收集数据，构建用户u和能力k的关联分数wuk，
S1-3：计算用户u对训练任务i的匹配度Pui，公式如下：
Pui＝∑Wukrfj
k∈G(u,i)
其中，G(u,i)表示用户u和训练任务i共有的能力，Wuk表示用户u和能力k的关联分数，rfj表示训练任务tf∈T对能力kj∈K的权重；
S1-4：建立机器初测推荐任务训练列表：对集I(u)中的所有训练任务进行按照用户匹配度进行排序，
集I(u)是指针对用户目前系统中所有任务训练的集合，这个集会随着系统升级不断增加，如2.0系统共有77个任务训练。
所述建立人工评价推荐任务训练列表，包括以下步骤：
S2-1：建立疾病训练模型：确定训练任务i和疾病d的关联度wid，构建疾病训练任务权重矩阵(I×D)并进行标准化处理，其中，
I表示训练任务集合，D表示疾病集合，
wid表示训练任务i∈I对疾病d∈D的权重，wid∈I×D；
S2-2：基于用户已持有的医学检测结果，确定用户u和疾病d的关联性分数Qui分数，分数范围在1～100分，du∈D；
S2-3计算用户u对训练任务i的匹配度P’ui，公式如下：
P’ui＝∑QuiWid
d∈H(u)
其中，H(u)表示用户确诊的疾病的集合，Wid表示训练任务i和疾病d的关联性分数，Qui表示用户u和疾病d的关联性分数。
S2-4：建立人工评价推荐任务训练列表：对集I’(u)中的所有训练任务进行按照用户匹配度进行加权、去重、排序；
集I’(u)是指基于文献或经验的疾病对应的一般任务训练的集合，可以是多个疾病的并集，这个集会随着对疾病认知的加深而不断变化，如偏侧忽略可训练的任务包括长空捕雀、一锤定音、寻枝摘果等。
所述建立最优推荐任务训练列表包括步骤：训练任务经过双重测评推荐，将用户u对训练任务i的匹配度Pui和P’ui分别进行加权再求和，计算每个训练任务的推荐得分S，
S＝aPui+bP’ui
其中a表示机器初测对应的权重；b表示人工评价对应的权重，将每个训练任务按得分高低排列，该顺序规则为最优推荐。
权重“a”和“b”系统默认值为0.5,0.5，表示任务训练推荐尊重机器算法和人工评价的双重意见；如a＝1,b＝0，则表示完全参照机器算法的推荐；如a＝0,b＝1，则表示完全参照人工评价的推荐。在实际运用中可根据有经验的治疗师对患者的了解程度自行设定，机器与人工互为补充，以此来减少单一机器算法或人工评价的决策误差。
根据本发明实施例提供的基于用户能力的个性化认知训练任务推荐方法，其中，
其中，通过以下步骤构建所述用户u和能力k的关联分数wuk：
用户依次完成机器初测预先设定的n个综合性测评子单元，并分别生成原始分数Xn；
提取受试者在n个综合性测评子单元中的原始分数，然后根据健康用户综合常模参数对综合性测评子单元中的原始分数进行标准化转换，生成用户u和能力k的关联分数wuk，公式如下：
Figure PCTCN2022107482-appb-000004
其中，i的取值范围为1～n；Wuk是进行标准化转换后的综合性测评子单元分数，即用户u和能力k的关联分数，如果用户被测量出能力较差，偏离常模越多，Wuk的值越大；Xi是综合性测评子单元中的原始分；
Figure PCTCN2022107482-appb-000005
是与受试者年龄、性别、职业、教育程度匹配的健康人群综合性测评子单元原始得分的平均值；σi是与受试者年龄、性别、职业、教育程度匹配的健康人群综合性测评子单元原始得分的标准差；
Figure PCTCN2022107482-appb-000006
和σi又叫做健康用户综合常模参数；wmn表示量表lm∈L对能力kn∈K的权重。
以下结合附图，详细说明本发明的各个实施例提供的技术方案。
如图1所示，本发明实施例提供的基于用户能力的个性化认知训练任务推荐方法，至少包括如下步骤：
S1.建立机器初测推荐任务训练列表；
S2.建立人工评价推荐任务训练列表；以及
S3.合并排序，建立最优推荐任务训练列表。
其中，步骤S1建立机器初测推荐任务训练列包括以下步骤
S1-1建立一般认知能力模型，依据智力理论和脑功能网络的研究成果，构建能力权重矩阵，训练任务能力权重矩阵，然后对能力权重矩阵和训练任务能力权重矩阵中的权重进行标准化处理。
建立一般认知能力结构方程模型的公式如下：
x＝ΛXηLK+εLK
y＝ΛyηTK+εTK
其中，x是单项量表或综合性量表分项组成的向量；ηLK是认知能力组成的向量；ΛX是量表与认知能力之间的关系，是量表在认知能力上的因子负荷矩阵；εLK是量表测量上的误差。y是训练任务组成的向量；ηTK是认知能力组成的向量；Λy是训练任务与认知能力 之间的关系，是训练任务在认知能力上的因子负荷矩阵；εTK是训练任务计算上的误差。
例如，如图2所示，在本发明的一个实施例中x BNT即x，BNT为MoCA测试中命名子测试的缩写；ΛX＝0.81；ηBNT-Gc即ηLK，BNT为MoCA测试中命名子测试的缩写，Gc为能力理解知识的缩写；εBNT-Gc即εLK，值为0，MoCA测试中命名子测试与能力理解知识的表达式为：
x BNT＝0.81*ηBNT-Gc+εBNT-Gc
如图3所示，在本发明的一个实施例中y 30001即y，30001为打地鼠的ID；ΛX＝0.88；η30001-Gv即ηTK，30001为打地鼠的ID，Gv为能力视觉处理的缩写；ε30001-Gv即εTK，值为0，打地鼠与能力视觉处理的表达式为：
y30001＝0.88*η30001-Gv+ε30001-Gv
其中，根据一般认知能力结构方程模型，构建能力权重矩阵(L×K),其中L表示量表集合；K表示能力集合；wmn表示量表lm∈L对能力kn∈K的权重，wmn∈L×K。
表1能力权重矩阵
Figure PCTCN2022107482-appb-000007
注：量表集合L为MoCA测试中的命名子测试、数字顺背子测试、句子复述子测试，共3个元素。能力集合K为理解知识、视觉处理、流体推理、短期记忆、处理速度，共5个元素。
其中，根据一般认知能力结构方程模型，构建训练任务能力权重矩阵(T×K),其中，
T表示训练任务集合；K表示能力集合；rfj表示训练任务tf∈T对能力kj∈K的权重，rfj∈T×K。
表2训练任务能力权重矩阵
Figure PCTCN2022107482-appb-000008
注：量表集合T为打地鼠、动点点击，共2个元素。能力集合K为理解知识、视觉处理、流体推理、短期记忆、处理速度，共5个元素。
S1-2：通过机器初测计算用户自身能力状况，其中，机器初测是指通过系统预先设置好的量表进行施测，根据量表作答情况来收集数据，测量结果会被智能分析，并比对常模以确定用户自身能力状况，构建用户u和能力k的关联分数wuk。通过系统预先设置好的量表进行的测试包括MoCA、MMSE、ADL、PHQ-9、GAD-7、生活满意度指数等临床应用较广的量表和任务式测评工具。
其中，通过以下步骤构建所述用户u和能力k的关联分数wuk：
用户依次完成机器初测预先设定的综合性测评子单元1、2、3、......n个测评，并生成原始分数X01、X02、X03、......Xn。
例如用户张某进行了MoCA测试，生成了MoCA测评得分表，原始分数在最后一列显示，包括总分和分项分数，如以下表3所示。
表3 MoCA测评得分表
Figure PCTCN2022107482-appb-000009
本发明所提供的推荐系统会自动提取受试者在n个综合性测评子单元中的原始分数，然后根据健康用户综合常模参数对综合性测评子单元中的原始分数进行标准化转换，生成用户u和能力k的关联分数wuk，公式如下：
Figure PCTCN2022107482-appb-000010
其中，i的取值范围为1～n；
Wuk是进行标准化转换后的综合性测评子单元分数，即用户u和能力k的关联分数，如果用户被测量出能力较差，偏离常模越多，Wuk的值越大；
Xi是综合性测评子单元中的原始分；
Figure PCTCN2022107482-appb-000011
是与受试者年龄、性别、职业、教育程度匹配的健康人群综合性测评子单元原始得分的平均值；
σi是与受试者年龄、性别、职业、教育程度匹配的健康人群综合性测评子单元原始得分的标准差；
Figure PCTCN2022107482-appb-000012
和σi又叫做健康用户综合常模参数；
wmn表示量表lm∈L对能力kn∈K的权重。
在本发明的一个实施例中，根据常模参数和计算公式，得出用户张某在MoCA测试的数字顺背子测试对流体推理能力的关联分数为Wuk＝46.2，对短期记忆能力的关联分数为Wuk＝96.8。
S1-3：计算用户u对训练任务i的匹配度Pui，公式如下：
Pui＝∑Wukrfj
k∈G(u,i)
其中，G(u,i)表示用户u和训练任务i共有的能力；
Wuk表示用户u和能力k的关联分数；
rfj表示训练任务tf∈T对能力kj∈K的权重。
在本发明的一个实施例中，得出用户张某对打地鼠训练任务的匹配度为Pui＝86.4，对动点点击训练任务的匹配度为Pui＝94.6。
S1-4：建立机器初测推荐任务训练列表：对集I(u)中的所有训练任务进行按照用户匹配度进行排序。取Top-N个物品，为每个训练任务赋予解释。保存Top-N个训练任务到机器初测推荐任务训练列表中，例如动点点击的排名高于打地鼠。
其中，所述建立人工评价推荐任务训练列表，包括以下步骤。
S2-1：建立疾病训练模型，依据疾病病理特征和康复训练文献，确定训练任务i和疾病d的关联度wid，构建疾病训练任务权重矩阵(I×D)并进行标准化处理。其中，
I表示训练任务集合；
D表示疾病集合；
wid表示训练任务i∈I对疾病d∈D的权重，wid∈I×D；
表4推荐任务训练列表
Figure PCTCN2022107482-appb-000013
S2-2：通过医生诊断结果(人工评价)，包括对受试者主述、病史、查体情况、影像学检查结果、日常生活能力评估结果以及其它量表评估结果，综合确定用户自身(疾病)状况，用Qui分数表示，分数范围在1～100分，分数越高表示患疾病d的风险越高，du∈D；在本发明的一个实施例中，对用户张某诊断结果为患轻度认知障碍的风险较高，Qui＝98。
S2-3：算用户u对训练任务i的匹配度P’ui，公式如下：
P’ui＝∑QuiWid
d∈H(u)
其中，H(u)表示用户确诊的疾病的集合，
Wid表示训练任务i和疾病d的关联性分数；
Qui表示用户u和疾病d的关联性分数。
在本发明的一个实施例中，得出用户张某对打地鼠训练任务的匹配度为P’ui＝72.52，对动点点击训练任务的匹配度为P’ui＝90.16。
S2-4：建立人工评价推荐任务训练列表：对集I’(u)中的所有训练任务进行按照用户匹配度进行加权、去重、排序。取Top-N个物品，为每个训练任务赋予解释。保存Top-N个训练任务到人工评价推荐任 务训练列表中，例如动点点击的排名高于打地鼠。
其中，在步骤“合并排序，建立最优推荐任务训练列表”中，训练任务经过双重测评推荐，通常用户会累计大量的训练任务。将用户u对训练任务i的匹配度Pui和P’ui分别进行加权再求和，计算每个训练任务的推荐得分S。
S＝aPui+bP’ui
其中a表示机器初测对应的权重；b表示人工评价对应的权重。
将每个训练任务按得分高低排列，该顺序规则为最优推荐。
在本发明的一个实施例中，得出用户张某对打地鼠训练任务的推荐得分为S＝79.46，对动点点击训练任务的推荐得分为S＝92.38。动点点击的排名高于打地鼠。
如图4所示，本发明实施例还提供一种基于用户能力的个性化认知训练任务推荐系统，包括处理器模块、电源模块、人机交互模块、通信模块、录入模块、测评模块、存储模块、计算模块和输出模块，上述电源模块、人机交互模块、通信模块、录入模块、测评模块、存储模块、计算模块和输出模块各自与处理器模块连接，所述测评模块包括机器初测模块，人工评价模块和综合难度推荐模块。
其中，所述机器初测模块用于建立机器初测推荐任务训练列表，通过以下步骤建立机器初测推荐任务训练列表：
S1-1：建立一般认知能力模型，并根据所述一般认知能力模型构建能力权重矩阵(L×K)以及训练任务能力权重矩阵(T×K)，然后对能力权重矩阵和训练任务能力权重矩阵中的权重进行标准化处理，其中，
建立一般认知能力结构方程模型的公式如下：
x＝ΛXηLK+εLK
y＝ΛyηTK+εTK
其中，x是单项量表或综合性量表分项组成的向量，ηLK是认知能力组成的向量，ΛX是量表与认知能力之间的关系，是量表在认知能力上的因子负荷矩阵，εLK是量表测量上的误差，y是训练任务组成的向量，ηTK是认知能力组成的向量，Λy是训练任务与认知能力之间的关系，是训练任务在认知能力上的因子负荷矩阵；εTK是训练任务计算上的误差，
对于所述能力权重矩阵(L×K)，L表示量表集合，K表示能力集合；wmn表示量表lm∈L对能力kn∈K的权重，wmn∈L×K，
对于所述训练任务能力权重矩阵(T×K)，T表示训练任务集合，K表示能力集合，rfj表示训练任务tf∈T对能力kj∈K的权重，rfj∈T×K；
S1-2：指通过预先设置好的量表进行施测，根据量表作答情况来收集数据，构建用户u和能力k的关联分数wuk，
S1-3：计算用户u对训练任务i的匹配度Pui，公式如下：
Pui＝∑Wukrfj
k∈G(u,i)
其中，G(u,i)表示用户u和训练任务i共有的能力，Wuk表示用户u和能力k的关联分数，rfj表示训练任务tf∈T对能力kj∈K的权重；
S1-4：建立机器初测推荐任务训练列表：对集I(u)中的所有训练任务进行按照用户匹配度进行排序，集I(u)是指针对用户目前系统中所有任务训练的集合，
所述人工测评模块用于建立人工评价推荐任务训练列表，通过以下步骤建立人工评价推荐任务训练列表：
S2-1：建立疾病训练模型，确定训练任务i和疾病d的关联度wid，构建疾病训练任务权重矩阵(I×D)并进行标准化处理，其中，
I表示训练任务集合，D表示疾病集合，
wid表示训练任务i∈I对疾病d∈D的权重，wid∈I×D；
S2-2：基于用户已持有的医学检测结果，确定用户u和疾病d的关联性分数Qui分数，分数范围在1～100分，du∈D；
S2-3计算用户u对训练任务i的匹配度P’ui，公式如下：
P’ui＝∑QuiWid
d∈H(u)
其中，H(u)表示用户确诊的疾病的集合，Wid表示训练任务i和疾病d的关联性分数，Qui表示用户u和疾病d的关联性分数，
S2-4：建立人工评价推荐任务训练列表，对集I’(u)中的所有训练任务进行按照用户匹配度进行加权、去重、排序；
所述综合推荐模块用于建立最优推荐任务训练列表，通过以下步骤建立最优推荐任务训练列表：
训练任务经过双重测评推荐，将用户u对训练任务i的匹配度Pui和P’ui分别进行加权再求和，计算每个训练任务的推荐得分S，
S＝aPui+bP’ui
其中a表示机器初测对应的权重，b表示人工评价对应的权重，将每个训练任务按得分高低排列，该顺序规则为最优推荐。
以上所述仅为本发明的实施例而已，并不用于限制本发明。对于本领域技术人员来说，本发明可以有各种更改和变化。凡在本发明的精神和原理之内所作的任何修改、等同替换、改进等，均应包含在本发明的权利要求范围之内。

Claims (10)

  1. 一种基于用户能力的个性化认知训练任务推荐方法，其特征在于包括以下步骤：
    建立机器初测推荐任务训练列表；
    建立人工评价推荐任务训练列表；以及
    合并排序，建立最优推荐任务训练列表。
  2. 如权利要求1所述的个性化认知训练任务推荐方法，其特征在于所述建立机器初测推荐任务训练列表，包括以下步骤：
    S1-1：建立一般认知能力模型，并根据所述一般认知能力模型构建能力权重矩阵(L×K)以及训练任务能力权重矩阵(T×K)，然后对能力权重矩阵和训练任务能力权重矩阵中的权重进行标准化处理；
    建立一般认知能力结构方程模型的公式如下：
    x＝ΛXηLK+εLK
    y＝ΛyηTK+εTK
    其中，x是单项量表或综合性量表分项组成的向量，ηLK是认知能力组成的向量，ΛX是量表与认知能力之间的关系，是量表在认知能力上的因子负荷矩阵，εLK是量表测量上的误差，y是训练任务组成的向量，ηTK是认知能力组成的向量，Λy是训练任务与认知能力之间的关系，是训练任务在认知能力上的因子负荷矩阵；εTK是训练任务计算上的误差，
    对于所述能力权重矩阵(L×K)，L表示量表集合，K表示能力集合；wmn表示量表lm∈L对能力kn∈K的权重，wmn∈L×K，
    对于所述训练任务能力权重矩阵(T×K)，T表示训练任务集合，K表示能力集合，rfj表示训练任务tf∈T对能力kj∈K的权重，rfj∈T×K；
    S1-2：指通过预先设置好的量表进行施测，根据量表作答情况来收集数据，构建用户u和能力k的关联分数wuk，
    S1-3：计算用户u对训练任务i的匹配度Pui，公式如下：
    Pui＝∑Wukrfj
    k∈G(u,i)
    其中，G(u,i)表示用户u和训练任务i共有的能力，Wuk表示用户u和能力k的关联分数，rfj表示训练任务tf∈T对能力kj∈K的权重；
    S1-4：建立机器初测推荐任务训练列表：对集I(u)中的所有训练任务进行按照用户匹配度进行排序，集I(u)是指针对用户目前系统中所有任务训练的集合。
  3. 如权利要求1所述的个性化认知训练任务推荐方法，其特征在于所述建立人工评价推荐任务训练列表，包括以下步骤：
    S2-1：建立疾病训练模型，确定训练任务i和疾病d的关联度wid，构建疾病训练任务权重矩阵(I×D)并进行标准化处理，其中，
    I表示训练任务集合，D表示疾病集合，
    wid表示训练任务i∈I对疾病d∈D的权重，wid∈I×D；
    S2-2：基于用户已持有的医学检测结果，确定用户u和疾病d的关联性分数Qui分数，分数范围在1～100分，du∈D；
    S2-3计算用户u对训练任务i的匹配度P’ui，公式如下：
    P’ui＝∑QuiWid
    d∈H(u)
    其中，H(u)表示用户确诊的疾病的集合，Wid表示训练任务i和疾病d的关联性分数，Qui表示用户u和疾病d的关联性分数，
    S2-4：建立人工评价推荐任务训练列表，对集I’(u)中的所有训练任务进行按照用户匹配度进行加权、去重、排序。
  4. 如权利要求1所述的个性化认知训练任务推荐方法，其特征在于所述建立最优推荐任务训练列表，包括如下步骤：
    训练任务经过双重测评推荐，将用户u对训练任务i的匹配度Pui和P’ui分别进行加权再求和，计算每个训练任务的推荐得分S，
    S＝aPui+bP’ui
    其中a表示机器初测对应的权重，b表示人工评价对应的权重，将每个训练任务按得分高低排列，该得分排列顺序即为推荐训练任务的排序。
  5. 如权利要求1所述的基于用户能力的个性化认知训练任务推 荐方法，其特征在于，通过以下步骤构建所述用户u和能力k的关联分数wuk：
    用户依次完成机器初测预先设定的n个综合性测评子单元，并分别生成原始分数Xn；
    提取受试者在n个综合性测评子单元中的原始分数，然后根据健康用户综合常模参数对综合性测评子单元中的原始分数进行标准化转换，生成用户u和能力k的关联分数wuk，公式如下：
    Figure PCTCN2022107482-appb-100001
    其中，i的取值范围为1～n；Wuk是进行标准化转换后的综合性测评子单元分数，即用户u和能力k的关联分数，如果用户被测量出能力较差，偏离常模越多，Wuk的值越大；Xi是综合性测评子单元中的原始分；
    Figure PCTCN2022107482-appb-100002
    是与受试者年龄、性别、职业、教育程度匹配的健康人群综合性测评子单元原始得分的平均值；σi是与受试者年龄、性别、职业、教育程度匹配的健康人群综合性测评子单元原始得分的标准差；
    Figure PCTCN2022107482-appb-100003
    和σi又叫做健康用户综合常模参数；wmn表示量表lm∈L对能力kn∈K的权重。
  6. 一种基于用户能力的个性化认知训练任务推荐系统，其特征在于包括机器初测模块、人工评价模块和综合推荐模块；其中，
    所述机器初测模块用于建立机器初测推荐任务训练列表，
    所述人工测评模块用于建立人工评价推荐任务训练列表，
    所述综合推荐模块用于建立最优推荐任务训练列表。
  7. 如权利要求6所述的基于用户能力的个性化认知训练任务推荐系统，其特征在于所述机器初测模块通过以下步骤建立机器初测推荐任务训练列表：
    S1-1：建立一般认知能力模型，并根据所述一般认知能力模型构建能力权重矩阵(L×K)以及训练任务能力权重矩阵(T×K)，然后对能力权重矩阵和训练任务能力权重矩阵中的权重进行标准化处理，其中，
    建立一般认知能力结构方程模型的公式如下：
    x＝ΛXηLK+εLK
    y＝ΛyηTK+εTK
    其中，x是单项量表或综合性量表分项组成的向量，ηLK是认知能力组成的向量，ΛX是量表与认知能力之间的关系，是量表在认知能力上的因子负荷矩阵，εLK是量表测量上的误差，y是训练任务组成的向量，ηTK是认知能力组成的向量，Λy是训练任务与认知能力之间的关系，是训练任务在认知能力上的因子负荷矩阵；εTK是训练任务计算上的误差，
    对于所述能力权重矩阵(L×K)，L表示量表集合，K表示能力集合；wmn表示量表lm∈L对能力kn∈K的权重，wmn∈L×K，
    对于所述训练任务能力权重矩阵(T×K)，T表示训练任务集合，K表示能力集合，rfj表示训练任务tf∈T对能力kj∈K的权重，rfj∈T×K；
    S1-2：指通过预先设置好的量表进行施测，根据量表作答情况来收集数据，构建用户u和能力k的关联分数wuk，
    S1-3：计算用户u对训练任务i的匹配度Pui，公式如下：
    Pui＝∑Wukrfj
    k∈G(u,i)
    其中，G(u,i)表示用户u和训练任务i共有的能力，Wuk表示用户u和能力k的关联分数，rfj表示训练任务tf∈T对能力kj∈K的权重；
    S1-4：建立机器初测推荐任务训练列表：对集I(u)中的所有训练任务进行按照用户匹配度进行排序，集I(u)是指针对用户目前系统中所有任务训练的集合。
  8. 如权利要求6所述的基于用户能力的个性化认知训练任务推荐系统，其特征在于所述人工测评模块通过以下步骤建立人工评价推荐任务训练列表：
    S2-1：建立疾病训练模型，确定训练任务i和疾病d的关联度wid，构建疾病训练任务权重矩阵(I×D)并进行标准化处理，其中，
    I表示训练任务集合，D表示疾病集合，
    wid表示训练任务i∈I对疾病d∈D的权重，wid∈I×D；
    S2-2：基于用户已持有的医学检测结果，确定用户u和疾病d的 关联性分数Qui分数，分数范围在1～100分，du∈D；
    S2-3计算用户u对训练任务i的匹配度P’ui，公式如下：
    P’ui＝∑QuiWid
    d∈H(u)
    其中，H(u)表示用户确诊的疾病的集合，Wid表示训练任务i和疾病d的关联性分数，Qui表示用户u和疾病d的关联性分数，
    S2-4：建立人工评价推荐任务训练列表，对集I’(u)中的所有训练任务进行按照用户匹配度进行加权、去重、排序。
  9. 如权利要求6所述的基于用户能力的个性化认知训练任务推荐系统，其特征在于所述综合推荐模块通过以下步骤建立最优推荐任务训练列表：
    训练任务经过双重测评推荐，将用户u对训练任务i的匹配度Pui和P’ui分别进行加权再求和，计算每个训练任务的推荐得分S，
    S＝aPui+bP’ui
    其中a表示机器初测对应的权重，b表示人工评价对应的权重，将每个训练任务按得分高低排列，该得分配列顺序即为推荐训练任务的排序。
  10. 如权利要求6所述的基于用户能力的个性化认知训练任务推荐系统，其特征在于，通过以下步骤构建所述用户u和能力k的关联分数wuk：
    用户依次完成机器初测预先设定的n个综合性测评子单元，并分别生成原始分数Xn；
    提取受试者在n个综合性测评子单元中的原始分数，然后根据健康用户综合常模参数对综合性测评子单元中的原始分数进行标准化转换，生成用户u和能力k的关联分数wuk，公式如下：
    Figure PCTCN2022107482-appb-100004
    其中，i的取值范围为1～n；Wuk是进行标准化转换后的综合性测评子单元分数，即用户u和能力k的关联分数，如果用户被测量出能力较差，偏离常模越多，Wuk的值越大；Xi是综合性测评子单元中的原始分；
    Figure PCTCN2022107482-appb-100005
    是与受试者年龄、性别、职业、教育程度匹配的健康人 群综合性测评子单元原始得分的平均值；σi是与受试者年龄、性别、职业、教育程度匹配的健康人群综合性测评子单元原始得分的标准差；
    Figure PCTCN2022107482-appb-100006
    和σi是健康用户综合常模参数；wmn表示量表lm∈L对能力kn∈K的权重。
PCT/CN2022/107482 2021-07-23 2022-07-22 基于用户能力的个性化认知训练任务推荐方法及系统 WO2023001301A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US18/420,648 US20240212825A1 (en) 2021-07-23 2024-01-23 User ability-based personalized cognitive training task recommendation method and system

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110833758.3 2021-07-23
CN202110833758.3A CN113284623B (zh) 2021-07-23 2021-07-23 基于用户能力的个性化认知训练任务推荐算法及系统

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US18/420,648 Continuation US20240212825A1 (en) 2021-07-23 2024-01-23 User ability-based personalized cognitive training task recommendation method and system

Publications (1)

Publication Number Publication Date
WO2023001301A1 true WO2023001301A1 (zh) 2023-01-26

Family

ID=77287033

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/107482 WO2023001301A1 (zh) 2021-07-23 2022-07-22 基于用户能力的个性化认知训练任务推荐方法及系统

Country Status (3)

Country Link
US (1) US20240212825A1 (zh)
CN (1) CN113284623B (zh)
WO (1) WO2023001301A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117612712A (zh) * 2024-01-23 2024-02-27 首都医科大学宣武医院 一种检测和提升认知测评诊断精度的方法及系统

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113284623B (zh) * 2021-07-23 2021-11-05 北京智精灵科技有限公司 基于用户能力的个性化认知训练任务推荐算法及系统
CN113741702B (zh) * 2021-11-03 2022-03-04 北京智精灵科技有限公司 基于情绪监测的认知障碍人机交互方法及系统
CN115295153B (zh) * 2022-09-30 2023-01-20 北京智精灵科技有限公司 基于深度学习的认知评估方法及认知任务推送方法
CN116738240B (zh) * 2023-08-14 2023-12-01 北京智精灵科技有限公司 认知训练的建模方法、建模系统及认知训练任务推送方法
CN118628309A (zh) * 2024-08-07 2024-09-10 北京智精灵科技有限公司 认知能力训练系统、方法、电子设备、计算机程序及存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112017788A (zh) * 2020-09-07 2020-12-01 平安科技（深圳）有限公司 基于强化学习模型的疾病排序方法、装置、设备及介质
US20200402663A1 (en) * 2019-06-18 2020-12-24 Canon Medical Systems Corporation Medical information processing apparatus and medical information processing method
JP2021047504A (ja) * 2019-09-17 2021-03-25 国立大学法人千葉大学 医師のための漢方医学における自動診断支援装置、自動診断支援方法、プログラム、学習済みモデルおよび学習済みモデル作成方法
CN112599241A (zh) * 2020-12-16 2021-04-02 深圳市唐仁医疗科技有限公司 一种基于智慧终端的大健康医疗疾病辅助诊断系统及方法
CN113284623A (zh) * 2021-07-23 2021-08-20 北京智精灵科技有限公司 基于用户能力的个性化认知训练任务推荐算法及系统

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107562795A (zh) * 2017-08-01 2018-01-09 广州市香港科大霍英东研究院 基于异构信息网络的推荐方法及装置
CN108280596A (zh) * 2018-03-02 2018-07-13 北京乐星球教育科技有限公司 学习品质测评方法
CN109902222B (zh) * 2018-11-30 2022-05-13 华为技术有限公司 一种推荐方法及装置
US11640403B2 (en) * 2019-07-03 2023-05-02 Kpn Innovations, Llc. Methods and systems for automated analysis of behavior modification data
CN111639860A (zh) * 2020-06-02 2020-09-08 安庆师范大学 一种基于数学模型的在线教育系统及其分班方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200402663A1 (en) * 2019-06-18 2020-12-24 Canon Medical Systems Corporation Medical information processing apparatus and medical information processing method
JP2021047504A (ja) * 2019-09-17 2021-03-25 国立大学法人千葉大学 医師のための漢方医学における自動診断支援装置、自動診断支援方法、プログラム、学習済みモデルおよび学習済みモデル作成方法
CN112017788A (zh) * 2020-09-07 2020-12-01 平安科技（深圳）有限公司 基于强化学习模型的疾病排序方法、装置、设备及介质
CN112599241A (zh) * 2020-12-16 2021-04-02 深圳市唐仁医疗科技有限公司 一种基于智慧终端的大健康医疗疾病辅助诊断系统及方法
CN113284623A (zh) * 2021-07-23 2021-08-20 北京智精灵科技有限公司 基于用户能力的个性化认知训练任务推荐算法及系统

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117612712A (zh) * 2024-01-23 2024-02-27 首都医科大学宣武医院 一种检测和提升认知测评诊断精度的方法及系统

Also Published As

Publication number Publication date
US20240212825A1 (en) 2024-06-27
CN113284623B (zh) 2021-11-05
CN113284623A (zh) 2021-08-20

Similar Documents

Publication Publication Date Title
WO2023001301A1 (zh) 基于用户能力的个性化认知训练任务推荐方法及系统
Paraschiakos et al. Activity recognition using wearable sensors for tracking the elderly
Me et al. Machine learning approach to model sport training
Pratap et al. Predictive assessment of autism using unsupervised machine learning models
US12020818B1 (en) Cross care matrix based care giving intelligence
CN109192312B (zh) 一种心力衰竭患者不良事件智能管理系统及方法
Masih et al. Application of data mining techniques for early detection of heart diseases using Framingham heart study dataset
Uddin et al. Cooperative learning for personalized context-aware pain assessment from wearable data
Yeh et al. An intelligent model for the classification of children’s occupational therapy problems
Grevet et al. Modeling the acceptability of bcis for motor rehabilitation after stroke: a large scale study on the general public
Yu et al. CR-LCRP: Course recommendation based on Learner–Course Relation Prediction with data augmentation in a heterogeneous view
Zhu Decision model of design for elderly wearable health monitoring products based on quality function deployment and analytic network process
Fontecha et al. Mobile services infrastructure for frailty diagnosis support based on Gower's similarity coefficient and treemaps
Agrawal et al. Healthcare recommendation system using patients' review
EP4281976A1 (en) Intelligent telehealth platform
Lin et al. System design of an intelligent nutrition consultation and recommendation model
Huang Exploration on psychological well-being education of college athletes
Kang Artificial Intelligence Network Embedding, Entrepreneurial Intention, and Behavior Analysis for College Students’ Rural Tourism Entrepreneurship
Ding et al. Design of Elderly Assisted Wheelchair Based on KJ-Technique and Analytic Hierarchy Process (AHP)
Li et al. Research on the Smart Home-care services System of Chengdu Community based on “Internet+”
Wang Innovative Strategies for the Development of International Chinese Language Education Based on Deep Learning Models
Hamam Guest Editorial Achieving Health Equity Through AI for Diagnosis and Treatment and Patient Monitoring
Yu Performance assessment and fitness analysis of athletes using decision tree and data mining techniques
Rahman et al. Review on food recommender system for diabetes mellitus
Zhao et al. A Risk Assessment Approach of Hypertension Based on Mobile Crowd Sensing.

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22845464

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 22845464

Country of ref document: EP

Kind code of ref document: A1