CN110136795A - It is a kind of for recognizing the construction method of the time series database of early warning - Google Patents
It is a kind of for recognizing the construction method of the time series database of early warning Download PDFInfo
- Publication number
- CN110136795A CN110136795A CN201910421470.8A CN201910421470A CN110136795A CN 110136795 A CN110136795 A CN 110136795A CN 201910421470 A CN201910421470 A CN 201910421470A CN 110136795 A CN110136795 A CN 110136795A
- Authority
- CN
- China
- Prior art keywords
- list
- behavior
- rehabilitation
- measure
- sensor
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
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
- G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2474—Sequence data queries, e.g. querying versioned data
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Medical Informatics (AREA)
- Physics & Mathematics (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- General Engineering & Computer Science (AREA)
- Epidemiology (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Fuzzy Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Medical Treatment And Welfare Office Work (AREA)
- Alarm Systems (AREA)
Abstract
The present invention relates to a kind of for recognizing the construction method of the time series database of early warning, by acquiring the real-time action information of user and being integrated to it with configuration information data set, the construction method further includes following steps: expertise database is constructed in a manner of obtaining several classification data according to executing the message data set to higher level classification；Based on the classification data, the building of process performing prediction classifier is predicted with the behavior that the following setting moment is occurred；The behavior obtained for prediction carries out scoring processing, and warning information is generated in the case where scoring decrease beyond given threshold.
Description
Technical field
The present invention relates to database construction technology fields more particularly to a kind of for recognizing the structure of the time series database of early warning
Construction method.
Background technique
The assessment for the cognitive ability of old man relies primarily on cognitive disorder Screening Scale at present, by staff to old
People evaluates and tests, and then realizes and assess the cognitive ability of old man.But there are following a problems for traditional scale assessment: one
Person it is longer to pay a return visit time interval, it is difficult to realize regular return visit；The two, the assessment time is longer, requires staff high, it is difficult to
It realizes a wide range of universal；Three, the low old man of education level are difficult to complete most of evaluation and test content.
Design a kind of cognition assessment system in patent of invention CN106327049A, including information module, test module and point
Analyse module；Information module is used for the data according to object, obtains the medical information to match with test module, and foundation is completely recognized
Know assessment database；Test module obtains the recognition tests data of object by test, includes following five submodules: attention
With execution functional test module, recall tests module, mathematics and computing capability test module, language testing module, movement and row
For control and scenario test module；The cognition that the medical information and test module that analysis module is obtained according to information module obtain
Test data determines the cognition assessment result of object.But the invention, which measures realization cognition there is still a need for staff, to be commented
Estimate, staff is required high, it is difficult to realize automation.The cognition decline of old man is a slow and imperceptible mistake simultaneously
Journey, cognition decline is often embodied among the daily behavior of old man, therefore the present invention is desirable to provide a kind of method, passes through sense
Know that technology realizes the behavioral data long term monitoring to old man, and then realizes the assessment to old man's cognitive ability.
In addition, the inevitable difference of those skilled in the art and examination department that one side is understood by applicant；Separately
On the one hand due to having studied lot of documents and patent when inventor makes the present invention, but length limit do not enumerate in detail it is all
Details and content, however this feature of the invention absolutely not for not having these prior arts, present invention have been provided with existing skill
All features of art, and applicant retains foundation guidelines for examination relevant regulations and increases related existing skill in the background technology at any time
The right of art.
Summary of the invention
Word " module " as used herein describes any hardware, software or combination thereof, is able to carry out and " mould
The associated function of block ".
For the deficiencies of the prior art, the present invention provides a kind of for recognizing the construction method of the time series database of early warning,
By acquiring the real-time action information of user and being integrated to it with configuration information data set, the construction method further includes
Following steps: the structure in a manner of obtaining several classification data is executed to higher level classification according to the message data set
Build expertise database；Based on the classification data, the building of process performing prediction classifier is to set moment institute to future
The behavior of generation is predicted；The behavior obtained for prediction carries out scoring processing, decrease beyond the feelings of given threshold in scoring
Warning information is generated under condition.
According to a kind of preferred embodiment, the scoring processing includes at least following steps: the n for selecting daily triggering most
Item behavior is as Standard Task X=(X1, X2..., Xn)；Standard of the time of the n task as this behavior is completed in test
Time T is simultaneously expressed as T=(T1, T2..., Tn)；Time t=(t needed for determining user's completion Standard Task1, t2...,
tn), user complete Standard Task task completeness α=(α1, α2..., αn), and according to the importance of each Standard Task
Determine its respective weight w=(w1, w2..., wn)；Establish the scoring formula of cognitive abilityAnd the scoring of user's cognitive ability is determined according to the scoring formula.
According to a kind of preferred embodiment, the building of the message data set includes at least following steps: user's
Several are arranged in activity space can acquire the sensor of user's difference action message, and at least for all the sensors configuration
One server；The server is configured to the sensor device letter according to the basic information at least constructed for stating sensor
The mode of breath table and the sensor data information table of the information content for stating sensor acquisition completes the message data set
Building.
According to a kind of preferred embodiment, the building of the expertise database includes at least following steps: being based on base
The classification that relevance between this life-form structure, sensor and attribute sensor completes message data set is included at least with constructing
It is table that behavior title, behavior, which define, trigger article, trigger sensor, triggered time and the sensor row of position, in which: is based on
The sensor row can be according to the article and article for triggering basic living behavior decomposition to specific movement, movement by table
Basic living behavior is limited several sensors that can be triggered to it by the mode of associated sensor.
According to a kind of preferred embodiment, in the case where generating warning information, health is generated based on the behavior that prediction obtains
Multiple Training strategy generates the rehabilitation training strategy including at least following steps: building predictive behavior list, interest list, life
Habit list, measure of rehabilitation list living；Based on predictive behavior list, interest list, living habit list and measure of rehabilitation list
Between correlation analysis as a result, by the measure of rehabilitation in specific measure of rehabilitation list according to insertion predictive behavior list in
Mode constructs behavior guiding list.
According to a kind of preferred embodiment, generating the rehabilitation training strategy further includes following steps: in setting time week
Phase constructs several behavior guiding lists different from each other in the way of being inserted into measure of rehabilitation different from each other；It is set described
Fix time in the period, configure interactive voice device or display by instruct user according to behavior guiding list is alternately performed in a manner of
Complete cognitive rehabilitative training.
According to a kind of preferred embodiment, the behavior guiding list is constructed including at least following steps: to the rehabilitation
Measure list and the living habit list carry out first order association analysis to sift out undesirable measure of rehabilitation, thus raw
At the first measure of rehabilitation list；Second level association analysis is carried out with most to the first measure of rehabilitation list and the interest list
The hobby for meeting user of big degree, to generate the second measure of rehabilitation list；To the second measure of rehabilitation list and in advance
Survey behavior list carries out third level association analysis, and predictive behavior list is inserted into the measure of rehabilitation in the second measure of rehabilitation list
In to form behavior guiding list.
According to a kind of preferred embodiment, the measure of rehabilitation list is reduced by the cognitive ability for being able to suppress user
Several behavior measures are constituted, wherein server can establish measure of rehabilitation list in such a way that customized or networking obtains.
According to a kind of preferred embodiment, the behavior prediction classifier can carry out structure based on Hidden Markov Model
It builds, in which: the element value that can be arranged by defining the i-th row jth of matrix A is by i-th of state qiIt is transferred to j-th of state qj's
Probability aij=P (qj|qi), and define matrix B row k jth column value state be qjWhen observe the probability b of data kj(k)
=P (k | qj) mode construct Hidden Markov Model λ=(A, B, π)；It can be by using the sensing data collected in advance
The Hidden Markov Model is trained.
According to a kind of preferred embodiment, completed for the sensing data being collected into set time period using training
Hidden Markov Model carry out behavior prediction, in which: using the Hidden Markov Model carry out behavior prediction when, will pass
Sensor data O=(o0, o1..., oT-1) it is used as observation sequence, by behavior I=(i0, i1..., iT-1) it is used as hidden variable.
Advantageous effects of the invention: the present invention collects the daily life of user by establishing implicit perception environment
Live data establishes behavior model, can meet long-term implicit data acquisition, and reduction is bothered user's daily life, obtained
Data steady in a long-term are obtained, and thus set up behavior model.Cognition energy is carried out by the permanently effective monitoring of daily life data
The assessment of power reduces the fault of human subjective's judgement.Simultaneously by real-time monitoring and prediction, scores cognitive ability, recognizing
Early warning is provided with during decline.
Detailed description of the invention
Fig. 1 is the flow diagram of the construction method of currently preferred time series database；With
Fig. 2 is the modularization connection relationship diagram of currently preferred each module.
Reference signs list
1: sensor 2: server 3: interactive voice device
4: display
Specific embodiment
It is described in detail with reference to the accompanying drawing.
Embodiment 1
As depicted in figs. 1 and 2, the present invention provides a kind of for recognizing the construction method of the time series database of early warning, at least
Include the following steps:
S1: acquiring the real-time action information of user and is integrated it with configuration information data set.
Specifically, several are arranged in the activity space of user can acquire the sensing of user's difference action message
Device 1, and at least one server 2 is configured for sensor 1, so that the real-time action that server 2 can at least acquire sensor 1
Information is stored.For example, server 2 can be cloud server, sensor 1 can be by way of wireless connection and service
Device 2 is communicatively coupled.Activity space refers to the living space of user.For example, when user lives in the house of oneself
When life, activity space is the living space that house is constituted.When nurse ward of the user in hospital, activity space is just
It is nurse ward.Preferably, it is acquired for the ease of the different action messages to user, it can be according to side shown in table 1
Formula arranges different sensors in e.g. house.
Table 1
Preferably, the configuration of server 2 is according to the side at least constructing sensor device information table and sensor data information table
The building of formula completion message data set.Sensor device information table is used to state the basic information of sensor.Basic information is at least
Including sensor name, MAC Address, effect description, sensor type, data unit and installation site.Sensor data information
Table is used to state the information content of sensor acquisition.The information content includes at least sensor device, data content and uplink time.
For example, in order to be acquired to real-time action information that e.g. kitchen tools use of the user in home environment, it can be in kitchen
Flame sensor is set on room hearth.Flame sensor can be far infrared flame sensor or ultraviolet flame sensor.This
When, unique identification of the MAC Address as sensor is used to distinguish different sensors, and different sensors has
Different MAC Address.Sensor name is flame sensor.Effect is described as judging whether using combustion gas.Sensor type is
The type of sensor.Data unit is to detect the unit of the data-signal obtained.For example, when sensor is flame sensor,
It can acquire the wavelength that flame is discharged, and then data unit is nanometer.Installation site is kitchen range.Sensor device is
The concrete model of sensor.It is infrared that data content includes at least the opening time of combustion gas, the shut-in time of combustion gas and combustion gas generation
The intensity of light.Uplink time refers to that the real-time action information that flame sensor is acquired is uploaded to the time of server.For example,
The data of acquisition are just uploaded to server to carry out unified storage in real time when sensor is triggered and executes data acquisition.Clothes
Data periodically can be carried out filing storage and data processing according to uplink time by business device.For example, server can be daily
Data are filed in ten two points of timings at night, to complete the packing storage that the same day collects data.Meanwhile data are according to binary system
Mode be stored in database.I.e. for door status switch sensor, flame sensor, laser sensor, water flow sensor, micro-
Dynamic switch sensor, pressure sensor and intelligent switch are stored as 1 when the numerical value of the data of sensor acquisition is greater than zero, no
Then it is stored as 0.For temperature sensor and humidity sensor, by sensor values divided by sensor measurement range before storage
Maximum temperature values or maximal humidity value, and stored what is obtained in the numerical value between 0 and 1 into database.
S2: it is constructed in a manner of obtaining several classification data to higher level classification specially according to being executed to message data set
Family's knowledge data base.The priori knowledge for relationship between sensor and behavior is provided by expertise to increase behavior prediction
Precision, and then can increase cognition early warning precision.Higher level classification is executed to refer to data progress further
Subdivision, makes it possible to for behavior and sensor being associated.
Specifically, completing classification based on the relevance between basic living behavior, sensor and attribute sensor with structure
Build expertise database.It is table that expertise database, which at least has sensor row,.Sensor row is that table includes at least behavior
Title, behavior definition, triggering article, trigger sensor, triggered time and position.By to the basic living row in daily life
It, can be by biography associated by basic living behavior decomposition to the triggered article of specific movement, movement, article to be defined
Sensor.By the way that behavior is limited several sensors that may be triggered to it, the accuracy rate of behavior prediction can be effectively improved.Example
It such as, can be to be away from home by behavior name definition when user needs to be away from home outgoing.User, which is away from home, at least have been needed
At the unlatching at gate and the shutoff operation at gate, and then behavior can be defined as opening gate, close gate and without indoor row
For triggering.It analyzes to obtain user by the data that sensor acquires when server and successively performs and open gate and close big
Door, and without other indoor behaviors triggerings when then predicts that user is in the state being away from home.Triggering article is gate at this time.Touching
Hair sensor is mounted in the door status switch sensor on gate.Triggered time is not construed as limiting, i.e. user's section at any time
The behavior being away from home is executed to belong to normally.Position refers to the address that behavior occurs, such as the position being away from home can be defined
For parlor.Preferably, it can be constructed based on the relevance between different behaviors and different sensors as shown in Table 2 comprising several
The sensor row of a difference behavior title is table.By building sensor row be table can specify different behavior and sensor it
Between many-to-many relationship.It is to collectively form that i.e. one complete behavior, which needs multiple sub-line, so as to trigger different biographies
Sensor.For example, the behavior of preparation breakfast as shown in table 2, needs to complete refrigerator operation, gas stove operation and micro-wave oven behaviour
Work can be completed, wherein the door status switch sensor of refrigerator, the flame sensor of gas stove and the door sensor of micro-wave oven can be triggered
Switch sensor.Preferably, sensor row is that every data line in table can be used as a classification data.Preferably, belong to
Property include at least sensor installation site, the article of binding and the time of triggering.
Table 2
S3: the building based on classification data process performing prediction classifier with behavior that the following setting moment will be occurred into
Row prediction.
Specifically, behavior prediction classifier can be constructed based on Hidden Markov Model.For example, being directed to sensor number
According to the timing that implies in behavior, future will sometime be occurred according to sensing data using Hidden Markov Model
Behavior is predicted.Hidden Markov Model introduce an implicit variable, and think state in data change be by and only by
Caused by the implicit variable at previous time point and this implicit variable, but this implicit variable can not observed person directly observe
It arrives.In Hidden Markov Model, row is corresponded to by implying transfer matrix, initial state probabilities matrix and hidden variable between variable
For emission matrix solve the behavior at the time point to be predicted.When carrying out behavior prediction using Hidden Markov Model, will go
For the hidden variable for regarding sensor as, sensing data is observable and behavioral data is hiding.
Preferably, the Hidden Markov mould completed for the sensing data being collected into set time period using training
Type carries out behavior prediction.Such as Hidden Markov Model can be trained by using the sensing data collected in advance,
And behavior was carried out using the Hidden Markov Model that training is completed to the sensing data that this day is collected into for the period with 24 hours
Prediction.Specifically, Hidden Markov Model training and prediction include at least following steps:
A1: the element value of the i-th row jth column of matrix A is defined by i-th of state qiIt is transferred to j-th of state qjProbability
aij=P (qj|qi).The value of the row k jth column of definition matrix B is q in statejWhen, observe the probability b of data kj(k)=P (k
|qj).So far, Hidden Markov Model λ=(A, B, π) building is completed.
A2: A and B belong to unknown parameter in the Hidden Markov Model of building, need hidden using viterbi algorithm solution
Markov model.Specifically, viterbi algorithm need be maximumlly hidden variable sequence after given observation data probability
maxP(I/O).Viterbi algorithm is defined as
The available recurrence formula of formula defined based on viterbi algorithm Wherein, Ψ is definedt(i)=argmax1≤j≤Nδt-1(j)aji.And then viterbi algorithm
Process are as follows:
Input: model λ=(A, B, π) observes data sequence O=(o0, o1..., oT-1)。
Output: hidden variable sequence I=(i0, i1..., iT-1)。
Initialize local state:
δ0=πibi(o0), i=0,1 ..., N-1
Ψ0(i)=0, i=0,1 ..., N-1
Dynamic Programming recursion t=1 is carried out, the local state at 2 ..., T-1 moment:
Calculate the maximum Ψ of moment TT(i), Ψ at this timeTIt (i) is the hidden variable state most possibly occurred at the t=T moment
I=(i is obtained using Ψ (i) backtracking0, i1..., iT-1):
it=Ψt+1(it+1)
When being predicted using Hidden Markov Model, by sensing data O=(o0, o1..., oT-1) regard sight as
Sequence is examined, by behavior I=(i0, i1..., iT-1) regard hidden variable as.It is corresponding that each moment can be acquired using viterbi algorithm
Behavior.
Preferably, in order to make it easy to understand, the concrete meaning of parameter involved in the above process is defined.T is indicated
Data sequence length.A indicates the transfer matrix between state (hidden variable).B indicates emission matrix.π indicates initial state probabilities
Matrix.The quantity of N expression hidden variable value.Q={ q0, q1..., qN-1Indicate hidden variable possible values set.I=(i0,
i1..., iT-1) indicate hidden variable sequence.O=(o0, o1..., oT-1) indicate the data sequence observed.
S4: the behavior obtained for prediction carries out scoring processing, generates in the case where scoring decrease beyond given threshold
Warning information.The decline of cognitive ability shows as the mistake of behavior in daily life, skips when completing a certain activity
Or incorrect important step is executed, so that this activity loses the meaning realized originally.For example, the mistake of behavior may wrap
It includes and forgets to close coal gas, refrigerator doors is kept to open, taken a long time to complete a certain simple work for a long time.Pass through
By cognitive ability score normalization, to quantify the quality of activity completion.According to daily score information, when the cognition energy of user
Power scoring just generates warning information when decreaseing beyond e.g. 10% threshold value.
Specifically, carrying out scoring processing to behavior includes at least following steps:
B1: select the most n behaviors of daily triggering as Standard Task X=(X1, X2..., Xn)。
B2: test obtains standard time T of the time as this behavior that common people complete the n task, is expressed as T=
(T1, T2..., Tn)。
B3: time t=(t needed for determining user's completion Standard Task1, t2..., tn), user complete standard appoint
Task completeness α=(α of business1, α2..., αn), and its respective weight w=is determined according to the importance of each Standard Task
(w1, w2..., wn)。
Specifically, the task completeness of Standard Task can pass through the subtask quantity that actually accomplishes and subtask quantity
The ratio of sum is determined.For example, behavior of washing the dishes as shown in table 2, includes altogether 5 subtasks, i.e., opening water valve,
Close water valve, open and close water valve several times, open kitchen cabinet and close kitchen cabinet.When user completes 3 subtasks therein
When, the task completeness for the behavior of washing the dishes is (3/5) * 100%=60%.
Preferably, the importance of Standard Task is divided by the frequency that task occurs.For example, by the important of Standard Task
Property is divided into three grades.The weight of the first estate is established as 0.5, and the weight of the second grade is established as 0.3, the power of the tertiary gradient
Recasting is set to 0.2.Statistics calculating is carried out to the frequency of the different behaviors in database, the daily behavior occurred more than 3 times
It is divided into the first estate.1~3 behavior occurs and is divided into the second grade.Behavior less than 1 time is divided into the tertiary gradient.
B4: the scoring formula of cognitive ability is establishedCommenting based on cognitive ability
The scoring for dividing formula that can calculate user cognition ability.
Embodiment 2
The present embodiment is the further improvement to embodiment 1, and duplicate content repeats no more.
Preferably, in the case where generating warning information, rehabilitation training strategy is generated based on the behavior that prediction obtains.It generates
Rehabilitation training strategy includes at least following steps:
C1: building predictive behavior list, interest list, living habit list, measure of rehabilitation list.
Specifically, predictive behavior list refers to that each moment in future generated by Hidden Markov Model prediction can be general
The list that the concrete behavior that rate generates is constituted, can be constructed in the way of timing.I.e. predictive behavior is according to the time
Sequencing is arranged.
Preferably, interest list refers to the concrete behavior list being made of the interests of user.For example, user can
There is but be not limited to listen song, dance, listen opera, a variety of different hobbies such as song of acting in an opera.User or its people that accompanies and attends to
Member can will be in the interests typing server of user by external input terminal.For example, server 2 can be configured with example
The external input terminal of keyboard in this way or voice recording device, the entourages such as user either children of user can
The interests of user to be inputted in server to construct interest list by external input terminal.
Preferably, living habit list refers to the user analyzed according to the collected data of sensor in setting
Between the concrete behavior that executes in section.Specifically, all the sensors data collected in 24 hours can be analyzed with
Determine the concrete behavior that user executes in different time period.For example, using object, morning for e.g. the elderly
7 points are got up, and first thing is to go to toilet after getting up, and are washed one's face and rinsed one's mouth later.So as to be carried out according to by time and concrete behavior
Associated mode obtains the living habit list of user.The partial content of living habit list can be as shown in table 3.It is preferred that
, the living habit list of user can be adjusted in the way of regularly updating.For example, in summer and winter,
User is, for example, time generation variation of getting up since climate reasons will lead to it, to need to determine living habit list
Phase updates.
Table 3
| Serial number | Time | Concrete behavior |
| 1 | 7:00~7:10 | It gets up |
| 2 | 7:10~7:30 | It goes to toilet |
| 3 | 7:30~7:50 | It washes one's face and rinses one's mouth |
| 4 | 7:50~8:30 | Prepare breakfast |
| 5 | 8:30~9:00 | It has breakfast |
Preferably, several behavior measure structures that measure of rehabilitation list is reduced by the cognitive ability for being able to suppress user
At.Server can establish measure of rehabilitation list in such a way that customized, networking obtains.Specifically, server can be with example
Hospital expert system in this way establishes access relation, so as to obtain the row for inhibiting cognitive ability to reduce from hospital's expert system
For measure.In general, hospital expert system has expertise database, wherein being stored with being effectively relieved about different syndromes
Measure.For example, being directed to senile dementia, it can be recorded in expertise database and e.g. carry out finger training, improve diet
Quality, improvement mood etc. can be effectively improved or prevent the measure of senile dementia.Server 2 and hospital expert system, which are established, visits
Ask that the measure that can be directly recorded after relationship is included in measure of rehabilitation list.Preferably, server can also be by outer
The mode for connecing input is customized to measure of rehabilitation list progress.For example, user, after the professional diagnosis through doctor, doctor understands root
Specific behavior measure is formulated for it according to the actual conditions of user, and then user or its caregiver can pass through
In the behavior measure input measure of rehabilitation list that the mode of external input provides doctor.
C2: based on the degree of association between predictive behavior list, interest list, living habit list and measure of rehabilitation list point
Analysis is as a result, construct row in the way of in radom insertion predictive behavior list for the measure of rehabilitation in specific measure of rehabilitation list
To instruct list.
Specifically, server is configured to carry out first order association analysis to measure of rehabilitation list and living habit list to sieve
Undesirable measure of rehabilitation out, to generate the first measure of rehabilitation list.Server can be true based on living habit list
Determine the animation feature of user.Animation feature is mainly used for determining the behavior that user can not execute.For example, raw
User, which can be recorded, in habit list living has the living habit to say one's prayer at night.In conjunction with the e.g. language of arrangement in the room
Sound sensor or imaging sensor can be determined the content of pray, so that analysis show that user has about some ancestor
Teach religious belief.Server is then able to obtain the taboo thing about the religion by way of extraneous input or networking
, so that the measure of rehabilitation relevant to taboo item in measure of rehabilitation list be deleted.Server can carry e.g. grey
The association analysis algorithms such as color correlation analysis algorithm, FP-Growth algorithm or Apriori algorithm with realize measure of rehabilitation list with
The first order association analysis of living habit list.
Preferably, server be additionally configured to carry out the first measure of rehabilitation list and interest list second level association analysis with
Meet the hobby of user to the greatest extent.There may be with user's for measure of rehabilitation in first measure of rehabilitation list
The case where hobby conflicts completely.For example, measure of rehabilitation may include the movements such as execution running or square dance to reinforce moving.
User may cause its execution that can not or have no intention to move relevant measure of rehabilitation due to fat or legs and feet inconvenience etc., this
When server can be formed deleting in the first measure of rehabilitation list with the conflicting measure of rehabilitation of interest list
Two measure of rehabilitation lists.Server can also according to e.g. grey relational grade analysis algorithm, FP-Growth algorithm or
The association analysis such as Apriori algorithm algorithm is to realize the second level association analysis of the first measure of rehabilitation list and interest list.
Preferably, server is additionally configured to carry out the second measure of rehabilitation list and predictive behavior list the third level and is associated with point
Analysis, and the measure of rehabilitation in the second measure of rehabilitation list is inserted into predictive behavior list to form behavior guiding list.Rehabilitation
Measure needs time of origin, scene and behavior property based on the concrete behavior in predictive behavior list to determine whether there is
Execute conflict.Specifically, behavior property can be divided into restricted limbs, attention concentration class and without restricted, wherein limb
The restricted expression concrete behavior of body needs user's four limbs that could complete.Need temporarily to occupy the double of user for example, going to toilet
Hand needs persistently to occupy the both legs of user to unlock trousers.Also for example during preparing breakfast, need for a long time frequently
The both hands of temporary user operate, and the both legs of temporary user is needed to walk about to shift one's position once in a while, thus
Realize the pick-and-place of article.Attention concentrates class to show that concrete behavior needs user's significant attention, cannot be by the external world
Excessive interference.For example, user should avoid the interference of external sound when praying as far as possible.It is no restricted to show user
In leisure state is loosened, state can be adjusted at any time to execute the item of any completion needed for it.For example, user is seeing
When TV, after TV need to only being opened by remote controler can whole process lie down or sitting is on sofa, it is in and loosens at this time
Leisure state, attention does not need high concentration and four limbs are also at unappropriated state, makes it possible at any time
Handle remaining item.Preferably, server can also according to e.g. grey relational grade analysis algorithm, FP-Growth algorithm or
The association analysis such as Apriori algorithm algorithm is associated with to realize the second measure of rehabilitation list with the third level of predictive behavior list point
Analysis.For example, the time of origin of the concrete behavior is 7 points of morning for the concrete behavior gone to toilet is executed after getting up 7 points of morning,
Scene is toilet.Behavior property is that limbs are restricted.For the measure of rehabilitation for playing music or opera, due to spot
Point belongs to the personal air of user, scene can be determined according to association analysis algorithm and play music or opera will not
It clashes.Time of origin belongs to the period earlier, user periphery be, for example, neighbours may also in sleep state, because
This according to correlation analysis algorithm can determine time of origin, and there are certain to conflict with music or opera is played, but can pass through
Volume is reduced to weaken conflict.The measure of rehabilitation of broadcasting music or opera can't occupy the attention or limb of user
Body can determine that it will not be clashed between behavior property according to correlation analysis algorithm.Final server can control
In toilet is, for example, that player executes in the way of the volume (i.e. 20 decibels~50 decibels) that the mankind normally speak
Play the measure of rehabilitation of music or opera.
C3: in set time period, it is different from each other in the way of being inserted into measure of rehabilitation different from each other to construct several
Behavior guiding list.
Specifically, can be in one week, as unit of day, every day constructs a behavior guiding list, so that each
It behavior guiding list is different.The difference of behavior guiding list be presented as same time period execute measure of rehabilitation each other
It is different.For example, the rehabilitation training measure for playing music can be inserted into for user first day 7 points of morning.At second day
Can be inserted into the rehabilitation training measure for playing morning news in the morning for user.
C4: in set time period, interactive voice device 3 or display 4 are configured to instruct user according to being alternately performed
The mode of behavior guiding list completes cognitive rehabilitative training.
Specifically, several interactive voice devices or several displays can be set in the living space of user.Example
Such as, interactive voice device and display can be set in parlor, bedroom and toilet.Interactive voice device can be broadcast to user
It puts voice and the voice signal of user's input can be received, consequently facilitating server judges the feedback content of user.Language
Sound interaction device and display are connected to server, so that server can carry out each interactive voice device and display
Control.Interactive voice device can according to voice broadcast by way of instruct user to execute measure of rehabilitation, or pass through display
Device can intuitively instruct user to execute measure of rehabilitation.
Preferably, with one week for the time cycle, 7 parts of behavior guiding lists different from each other can be formed.It is alternately performed row
Refer to for list in the first part of behavior guiding list of execution in first day, in the second part of behavior guiding list of execution in second day, successively
Analogize, in the 7th part of behavior guiding list of execution in the 7th day, so that the rehabilitation training content that user is daily in one week
It is different, it can be avoided user and generate and be weary of, and then effectively improve rehabilitation training effect.
It should be noted that above-mentioned specific embodiment is exemplary, those skilled in the art can disclose in the present invention
Various solutions are found out under the inspiration of content, and these solutions also belong to disclosure of the invention range and fall into this hair
Within bright protection scope.It will be understood by those skilled in the art that description of the invention and its attached drawing are illustrative and are not
Constitute limitations on claims.Protection scope of the present invention is defined by the claims and their equivalents.
Claims (10)
1. a kind of for recognizing the construction method of the time series database of early warning, by acquiring the real-time action information of user and right
It is integrated with configuration information data set, which is characterized in that the construction method further includes following steps:
It is constructed in a manner of obtaining several classification data to higher level classification specially according to being executed to the message data set
Family's knowledge data base；
Based on the classification data building of process performing prediction classifier with behavior that the following setting moment will be occurred into
Row prediction；
The behavior obtained for prediction carries out scoring processing, and early warning letter is generated in the case where scoring decrease beyond given threshold
Breath.
2. construction method according to claim 1, which is characterized in that the scoring processing includes at least following steps:
The n item behavior for selecting daily triggering most is as Standard Task X=(X1,X2,…,Xn)；
Test complete the n task time as this behavior standard time T and be expressed as T=(T1,T2,…,Tn)；
Time t=(t needed for determining user's completion Standard Task1,t2,…,tn), user complete Standard Task task
Completeness α=(α1,α2,…,αn), and its respective weight w=(w is determined according to the importance of each Standard Task1,w2,…,
wn)；
Establish the scoring formula of cognitive abilityAnd made according to scoring formula determination
The scoring of user's cognitive ability.
3. construction method according to claim 2, which is characterized in that the building of the message data set includes at least as follows
Step:
Several sensors (1) that can acquire user's difference action message are set in the activity space of user, and are
All the sensors (1) configure at least one server (2)；
The server (2) is configured to according to the sensor device information table at least constructed for stating the basic information of sensor
The structure of the message data set is completed with the mode of the sensor data information table of the information content for stating sensor acquisition
It builds.
4. construction method according to claim 3, which is characterized in that the building of the expertise database includes at least
Following steps:
The classification of message data set is completed based on the relevance between basic living behavior, sensor and attribute sensor with structure
Build including at least behavior title, behavior defines, trigger article, trigger sensor, triggered time and the sensor row of position as table,
Wherein:
It can be according to the object for triggering basic living behavior decomposition to specific movement, movement by table based on the sensor row
Basic living behavior is limited several sensors that can be triggered to it by the mode of sensor associated by product and article.
5. construction method according to claim 4, which is characterized in that in the case where generating warning information, based on prediction
The behavior of acquisition generates rehabilitation training strategy, generates the rehabilitation training strategy including at least following steps:
Construct predictive behavior list, interest list, living habit list, measure of rehabilitation list；
Based on the correlation analysis between predictive behavior list, interest list, living habit list and measure of rehabilitation list as a result,
Measure of rehabilitation in specific measure of rehabilitation list is constructed into behavior guiding list in the way of in insertion predictive behavior list.
6. construction method according to claim 5, which is characterized in that generating the rehabilitation training strategy further includes walking as follows
It is rapid:
In set time period, several behaviors different from each other are constructed in the way of being inserted into measure of rehabilitation different from each other and are referred to
Lead list；
In the set time period, interactive voice device (3) or display (4) are configured to instruct user according to being alternately performed
The mode of behavior guiding list completes cognitive rehabilitative training.
7. construction method according to claim 6, which is characterized in that construct the behavior guiding list including at least as follows
Step:
It is undesirable to sift out that first order association analysis is carried out to the measure of rehabilitation list and the living habit list
Measure of rehabilitation, to generate the first measure of rehabilitation list；
Carry out second level association analysis to the first measure of rehabilitation list and the interest list is made with maximum satisfaction
The hobby of user, to generate the second measure of rehabilitation list；
Third level association analysis is carried out to the second measure of rehabilitation list and predictive behavior list, and will be in the second measure of rehabilitation list
Measure of rehabilitation insertion predictive behavior list in form behavior guiding list.
8. the construction method according to one of preceding claims, which is characterized in that the measure of rehabilitation list is by that can press down
Several behavior measures that the cognitive ability of user processed reduces are constituted, wherein server (2) can pass through customized or networking
The mode of acquisition establishes measure of rehabilitation list.
9. construction method according to claim 8, which is characterized in that the behavior prediction classifier can be based on hidden Ma Er
Section's husband's model constructs, in which:
The element value that can be arranged by defining the i-th row jth of matrix A is by i-th of state qiIt is transferred to j-th of state qjProbability
aij=P (qj|qi), and define matrix B row k jth column value state be qjWhen observe the probability b of data kj(k)=P
(k|qj) mode construct Hidden Markov Model λ=(A, B, π)；
The Hidden Markov Model can be trained by using the sensing data collected in advance.
10. construction method according to claim 9, which is characterized in that for the sensing being collected into set time period
Device data carry out behavior prediction using the Hidden Markov Model that training is completed, in which:
When carrying out behavior prediction using the Hidden Markov Model, by sensing data O=(o0,o1,…,oT-1) as sight
Sequence is examined, by behavior I=(i0,i1,…,iT-1) it is used as hidden variable.
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210285110.1A CN114512211A (en) | 2019-05-20 | 2019-05-20 | Cognitive training strategy generation method and system based on time sequence database |
| CN201910421470.8A CN110136795B (en) | 2019-05-20 | 2019-05-20 | Construction method of time sequence database for cognitive early warning |
| CN202210308288.3A CN114664405A (en) | 2019-05-20 | 2019-05-20 | Cognitive early warning system and method based on time sequence database |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910421470.8A CN110136795B (en) | 2019-05-20 | 2019-05-20 | Construction method of time sequence database for cognitive early warning |
Related Child Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210285110.1A Division CN114512211A (en) | 2019-05-20 | 2019-05-20 | Cognitive training strategy generation method and system based on time sequence database |
| CN202210308288.3A Division CN114664405A (en) | 2019-05-20 | 2019-05-20 | Cognitive early warning system and method based on time sequence database |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110136795A true CN110136795A (en) | 2019-08-16 |
| CN110136795B CN110136795B (en) | 2022-02-01 |
Family
ID=67571839
Family Applications (3)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910421470.8A Active CN110136795B (en) | 2019-05-20 | 2019-05-20 | Construction method of time sequence database for cognitive early warning |
| CN202210285110.1A Pending CN114512211A (en) | 2019-05-20 | 2019-05-20 | Cognitive training strategy generation method and system based on time sequence database |
| CN202210308288.3A Pending CN114664405A (en) | 2019-05-20 | 2019-05-20 | Cognitive early warning system and method based on time sequence database |
Family Applications After (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210285110.1A Pending CN114512211A (en) | 2019-05-20 | 2019-05-20 | Cognitive training strategy generation method and system based on time sequence database |
| CN202210308288.3A Pending CN114664405A (en) | 2019-05-20 | 2019-05-20 | Cognitive early warning system and method based on time sequence database |
Country Status (1)
| Country | Link |
|---|---|
| CN (3) | CN110136795B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111400284A (en) * | 2020-03-20 | 2020-07-10 | 广州咨元信息科技有限公司 | Method for establishing dynamic anomaly detection model based on performance data |
| CN115485749A (en) * | 2020-08-28 | 2022-12-16 | 日立环球生活方案株式会社 | Monitoring device, monitoring method, and monitoring program |
| CN116168805A (en) * | 2023-01-20 | 2023-05-26 | 北京瑞帆科技有限公司 | Thinking training device and cognitive training system for cognitive training |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1564219A (en) * | 2004-03-25 | 2005-01-12 | 浙江工业大学 | Household security device for lonely living edged people |
| CN107886103A (en) * | 2016-09-29 | 2018-04-06 | 日本电气株式会社 | For identifying the method, apparatus and system of behavior pattern |
| US20180151254A1 (en) * | 2016-11-30 | 2018-05-31 | Electronics And Telecommunications Research Institute | High-speed similar case search method and device through reduction of large scale multi-dimensional time series health data to multiple dimensions |
| CN108348194A (en) * | 2015-09-15 | 2018-07-31 | 联邦科学和工业研究机构 | Mobility monitors |
| CN109473173A (en) * | 2018-09-30 | 2019-03-15 | 华中科技大学 | A kind of the elderly's Cognitive deficiency assessment system and device based on video |
| CN109545330A (en) * | 2018-11-30 | 2019-03-29 | 北京京师脑力科技有限公司 | It is a kind of to improve the cognitive training method and system for executing function |
-
2019
- 2019-05-20 CN CN201910421470.8A patent/CN110136795B/en active Active
- 2019-05-20 CN CN202210285110.1A patent/CN114512211A/en active Pending
- 2019-05-20 CN CN202210308288.3A patent/CN114664405A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1564219A (en) * | 2004-03-25 | 2005-01-12 | 浙江工业大学 | Household security device for lonely living edged people |
| CN108348194A (en) * | 2015-09-15 | 2018-07-31 | 联邦科学和工业研究机构 | Mobility monitors |
| CN107886103A (en) * | 2016-09-29 | 2018-04-06 | 日本电气株式会社 | For identifying the method, apparatus and system of behavior pattern |
| US20180151254A1 (en) * | 2016-11-30 | 2018-05-31 | Electronics And Telecommunications Research Institute | High-speed similar case search method and device through reduction of large scale multi-dimensional time series health data to multiple dimensions |
| CN109473173A (en) * | 2018-09-30 | 2019-03-15 | 华中科技大学 | A kind of the elderly's Cognitive deficiency assessment system and device based on video |
| CN109545330A (en) * | 2018-11-30 | 2019-03-29 | 北京京师脑力科技有限公司 | It is a kind of to improve the cognitive training method and system for executing function |
Non-Patent Citations (1)
| Title |
|---|
| 杨帅: "基于室内轨迹分析的老人行为识别方法", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111400284A (en) * | 2020-03-20 | 2020-07-10 | 广州咨元信息科技有限公司 | Method for establishing dynamic anomaly detection model based on performance data |
| CN111400284B (en) * | 2020-03-20 | 2023-09-12 | 广州咨元信息科技有限公司 | Method for establishing dynamic anomaly detection model based on performance data |
| CN115485749A (en) * | 2020-08-28 | 2022-12-16 | 日立环球生活方案株式会社 | Monitoring device, monitoring method, and monitoring program |
| CN115485749B (en) * | 2020-08-28 | 2024-04-23 | 日立环球生活方案株式会社 | Monitoring device, monitoring method and monitoring program |
| CN116168805A (en) * | 2023-01-20 | 2023-05-26 | 北京瑞帆科技有限公司 | Thinking training device and cognitive training system for cognitive training |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114512211A (en) | 2022-05-17 |
| CN110136795B (en) | 2022-02-01 |
| CN114664405A (en) | 2022-06-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jung et al. | Occupant-centered real-time control of indoor temperature using deep learning algorithms | |
| CN110136795A (en) | It is a kind of for recognizing the construction method of the time series database of early warning | |
| Ha et al. | Goal recognition with Markov logic networks for player-adaptive games | |
| van Kasteren | Activity recognition for health monitoring elderly using temporal probabilistic models | |
| Kormányos et al. | Multilevel simulation of daily activities: Why and how? | |
| Casaccia et al. | Measurement of users’ well-being through domotic sensors and machine learning algorithms | |
| JP2009502216A (en) | Methods, systems, and devices that utilize, regulate, trigger, and synchronize global bionetwork effects of optimized self-organizing adaptive capacity | |
| Elbayoudi et al. | The human behaviour indicator: A measure of behavioural evolution | |
| Thielke et al. | Associations between observed in‐home behaviors and self‐reported low mood in community‐dwelling older adults | |
| Papamatthaiakis et al. | Monitoring and modeling simple everyday activities of the elderly at home | |
| Bourret et al. | Dynamics of parent-adolescent interactions during a discussion on career choice: The role of parental behaviors and emotions | |
| Snell et al. | Child effects and child care: Implications for risk and adjustment | |
| Palermo et al. | Designing a clinically applicable deep recurrent model to identify neuropsychiatric symptoms in people living with dementia using in-home monitoring data | |
| JP2021067618A (en) | Program and system for determining meat quality | |
| Culmone et al. | Human activity recognition using a semantic ontology-based framework | |
| JP7085266B2 (en) | Devices, programs and methods that can estimate satisfaction based on behavioral information | |
| Zhang et al. | Predicting mental health problems with personality, behavior, and social networks | |
| Duchene et al. | An hybrid refinement methodology for multivariate simulation in home health telecare | |
| Virone et al. | Integration of an environmental sound module to an existing in-home activity simulator | |
| Casagrande | Sensor event and activity prediction using binary sensors in real homes with older adults | |
| Spyrou et al. | Cognitive game for OCD effects minimisation using IoT | |
| Ali et al. | Novel Respiratory Diseases Diagnosis by Using Fuzzy Logic | |
| Duchêne et al. | An hybrid knowledge-based methodology for multivariate simulation in home health telecare | |
| Shao | Smart Home Platform for Human Care Based on Healthcare as a Service | |
| Bragard et al. | Communication skills training for residents: which variables predict learning of skills? |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230505 Address after: Room 215-E, 2nd Floor, Podium Building, No. 33 Yueming Road, Xixing Street, Binjiang District, Hangzhou City, Zhejiang Province, 310000 China (Zhejiang) Free Trade Pilot Zone Patentee after: Hangzhou Zhilan health Co.,Ltd. Address before: 230009 No. 193, Tunxi Road, Hefei, Anhui Patentee before: Hefei University of Technology |