CN109767836A - A kind of medical diagnosis artificial intelligence system, device and its self-teaching method - Google Patents
A kind of medical diagnosis artificial intelligence system, device and its self-teaching method Download PDFInfo
- Publication number
- CN109767836A CN109767836A CN201811639163.9A CN201811639163A CN109767836A CN 109767836 A CN109767836 A CN 109767836A CN 201811639163 A CN201811639163 A CN 201811639163A CN 109767836 A CN109767836 A CN 109767836A
- Authority
- CN
- China
- Prior art keywords
- data
- new
- module
- algorithm
- medical
- 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.)
- Pending
Links
Landscapes
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The present invention provides a kind of medical diagnosis artificial intelligence system, device and its self-teaching method, comprising: carries out raw data acquisition；Data after being classified；Data substitute into data target algorithm after classifying, and carry out data conversion, form visual data；And constantly carry out the verifying of data target algorithm；Using visual data, building user draws a portrait；According to user's portrait judge whether that matched medical model can be found, carries out risk prompting；New medical model is constructed by machine learning algorithm according to user's sample characteristics；New data index algorithm is voluntarily constructed according to new medical model；New data index algorithm is replaced into former data target algorithm.The present invention can collect the data of processing patient with 24 hours, and not influence the normal life of user, and medical diagnosis artificial intelligence platform of the invention can carry out self-teaching, so that diagnosis is more and more accurate, diagnosable diseases range is more and more.
Description
Technical field
The present invention relates to a kind of artificial intelligence platform and its operation methods, and in particular, to it is a kind of can self-teaching doctor
Learn diagnosis artificial intelligence system, device and its self-teaching method.
Background technique
Current online medical diagnosis platform, rests on doctor mostly and resides on website, and patient passes through network communication skill
Art is linked up with doctor, and the illness of oneself is described to doctor in a manner of allowing doctor to make diagnosis.This mode has following
Disadvantage: (1) doctor cannot actually touch patient, less can be carried out the four methods of diagnosis, if patient symptom cannot be described it is clear,
It is then likely to misleading doctor and makes false judgment；(2) since the information of acquisition is extremely limited, even if online medical diagnosis platform
On doctor be made that diagnosis, patient still will go to hospital to check finally to make a definite diagnosis in person, this makes the online medical diagnosis flat
Platform has actually become chicken ribs, and all causes waste of time to patient and doctor；(3) doctor is not any in diagnosis
It can refer to data, easily judge by accident, generate irreversible consequence；(4) since doctor's night will rest, online medical diagnosis platform exists
When evening, be it is in paralyzed state, be unable to 24-hour service sufferer.
Therefore, being badly in need of one kind in the market, can voluntarily to collect data, diagnosis illness, diagnosis accurate, indefatigable artificial
Intelligent disease diagnosing system.
Summary of the invention
For the defects in the prior art, the object of the present invention is to provide a kind of medical diagnosis artificial intelligence systems, device
And self-teaching method, user's data, diagnosis illness can be voluntarily collected, and diagnose accurately, be less prone to diagnostic error,
New data index algorithm can be generated in time, according to new illness to adapt to huger audience.
According to the first aspect of the invention, a kind of self-teaching method of medical diagnosis artificial intelligence system is provided, comprising:
Acquire medical diagnosis initial data；
The collected initial data is subjected to classification processing, data after being classified；
Data after the classification are substituted into data target algorithm, data conversion is carried out, form visual data；And no
The disconnected verifying for carrying out the data target algorithm；
Using the visual data, building user draws a portrait；
According to user portrait judge whether that matched medical model can be found, be mentioned if so, carrying out risk to user
It wakes up；Otherwise, inform that user does not find matched medical model, new by machine learning algorithm building according to user's sample characteristics
Medical model；
New data index algorithm is voluntarily constructed according to the new medical model, the new data index algorithm is replaced into former number
According to index algorithm.
Optionally, it is described classified after after data further include: data after the classification are stored to big data platform number
According to pond.
Optionally, new data index algorithm is voluntarily constructed according to the new medical model, comprising:
The data after extracting the classification in the big data platform data pool；
New data index algorithm is voluntarily constructed in conjunction with data after the classification according to the new medical model.
Optionally, after the raw data acquisition further include: remove the obvious deviation data in the initial data.
Optionally, the raw data acquisition passes through intelligent terminal uninterrupted sampling initial data.
According to the second aspect of the invention, provide it is a kind of can self-teaching medical diagnosis artificial intelligence system, comprising:
Raw data acquisition module, for carrying out medical diagnosis raw data acquisition；
Data categorization module, for carrying out the raw data acquisition module collected initial data at classification
Reason, data after being classified；
Data conversion module, data substitute into data target and calculate after the classification for generating the data categorization module
In method, data conversion is carried out, forms visual data；The data conversion module is also used to constantly carry out the data target algorithm
Verifying；
User, which draws a portrait, constructs module, and the visual data building user for being formed using the data conversion module is drawn
Picture；
Medical model matching judgment module is sentenced for being drawn a portrait to construct the user that module constructs and draw a portrait according to the user
It is disconnected whether to find matched medical model；
Risk reminding module, for carrying out when the medical model matching judgment module finds matched medical model
Risk is reminded；
New medical model generation module, for not finding matched medicine mould in the medical model matching judgment module
When type, inform that user does not find matched medical model, and new by machine learning algorithm building according to user's sample characteristics
Medical model；
New data index algorithm constructs module, the new medicine for being generated according to the new medical model generation module
Model voluntarily constructs new data index algorithm；
Data target algorithm replacement module, for replacing with the data target algorithm in the data conversion module
The new data index algorithm that the new data index algorithm building module constructs.
Optionally, it is described can self-teaching medical diagnosis artificial intelligence system further include: big data platform data pool, use
The data after storing the classification that the data categorization module generates.
Optionally, the new data index algorithm building module includes:
Data extracting unit after classification, for being put down from the big data before the new data index algorithm constructs module
Data after the classification are proposed in platform data pool；
New data index algorithm construction unit, the new medicine for being generated according to the new medical model generation module
Data voluntarily construct new data index algorithm after the classification that data extracting unit is extracted after classifying described in models coupling.
Optionally, it is described can self-teaching medical diagnosis artificial intelligence system further include: cleaning module, for removing
State the obvious deviation data in the initial data of raw data acquisition module acquisition.
Optionally, the raw data acquisition module, uninterrupted sampling initial data.
According to the third aspect of the invention we, provide it is a kind of can self-teaching medical diagnosis artificial intelligence device, including deposit
Reservoir, processor and storage on a memory and the computer program that can run on a processor, described in the processor execution
It can be used for executing the self-teaching method of the medical diagnosis artificial intelligence system when program.
Compared with prior art, the present invention have it is following the utility model has the advantages that
The present invention it is above-mentioned can self-teaching method, self-teaching can be carried out, data target algorithm is voluntarily improved, be used for
Medical diagnosis artificial intelligence system, device, can improve the accuracy of diagnostic result, and corresponding health type is more and more.
The present invention it is above-mentioned can the medical diagnosis artificial intelligence system of self-teaching, device, can be handled with collection in 24 hours
The data of patient diagnose without user's initiative, do not influence the normal life of user, there may be health detecting user
When problem, thinks that user sends alarm at once, inform user's unsoundness risk.
Detailed description of the invention
Upon reading the detailed description of non-limiting embodiments with reference to the following drawings, other feature of the invention,
Objects and advantages will become more apparent upon:
Fig. 1 is the flow chart of one embodiment of the invention method；
Fig. 2 is the flow chart of another embodiment of the present invention；
Fig. 3 is the flow chart of another embodiment of the present invention；
Fig. 4 is the flow chart of another embodiment of the present invention；
Fig. 5 is the heart rate density accounting sequence chart of one embodiment of the invention whole day analysis；
Fig. 6 a, Fig. 6 b are the relational graph in one embodiment of the invention between average heart rate and oxygen demand, relative metabolic rate；
Fig. 7 a, Fig. 7 b are centre of motion rate and oxygen demand, the relational graph of relative metabolic rate respectively；
Fig. 8~Figure 10 is that traditional Chinese medicine diagnoses artificial intelligence system monitoring display figure in one embodiment of the invention；
Figure 11 is new medical model training functional block diagram in one embodiment of the invention.
Specific embodiment
The present invention is described in detail combined with specific embodiments below.Following embodiment will be helpful to the technology of this field
Personnel further understand the present invention, but the invention is not limited in any way.It should be pointed out that the ordinary skill of this field
For personnel, without departing from the inventive concept of the premise, various modifications and improvements can be made.These belong to the present invention
Protection scope.
Shown in referring to Fig.1, one embodiment of the invention traditional Chinese medicine diagnoses the process of the self-teaching method of artificial intelligence system
Figure, includes the following steps:
S100 carries out raw data acquisition by intelligent terminal, and intelligent terminal can be mobile phone, bracelet or smartwatch
Etc. equipment be in period continuous acquisition, such as the initial data on entire daytime, entire night when acquiring initial data
Initial data, the initial data of whole day, also or initial data when tranquillization, initial data can be PPG sensor and detect
Consecutive variations volumetric blood, be also possible to the walking number etc. that six axle sensors detect；
Collected initial data is carried out classification processing, data after being classified by S200, and classification processing refers to: according to
Everyone time cycle carries out heart rate acquisition, and time cycle point sleep (deep sleep and either shallow sleep), early morning, (morning was awake
Do not get up), tranquillization, movement, and according to the twenty four hours time carry out time single-point and each segment classification；
S300 substitutes into data after classification in data target algorithm, carries out data conversion, forms visual data；And no
The disconnected verifying for carrying out data target algorithm；Data target algorithm is that different types of heart rate value is compared and reconstructed, and is obtained
Heart rate to the same type at the same time is same more in the ratio calculated variation of different cycles, distribution and a people
The comparison of each serial heart rate value of number people and the degree of deviation calculate.
Specifically, such as time array index analysis:
The wherein heart rate density accounting sequence chart of whole day analysis can watch out people's entirety heart rate Density Distribution and cover
Whether section and heart rate density quality.As shown in Figure 5.
S400, using visual data, building user draws a portrait, and so-called user portrait includes: that age of user, gender, height are (dynamic
State), weight (dynamic), the information such as waistline (dynamic)；
S500 according to user's portrait judges whether that matched medical model can be found, if so, thening follow the steps S600；It is no
Then, step S700 is executed；
Wherein, medical model refers to: various disease under different conditions the linear regression analysis of corresponding Heart Rate States with
Mathematical statistics analysis, i.e., in different illnesss, the mathematical model of the Rule of Change of Heart Rate of patient and normal person.
As two groups of heart rate variability values of following table compare:
In the case where there is crowd's label bar part, normal person is linear with certain disease type patient heart rate [pulse frequency] change conditions
Correlation.Such as 55 years old or more the elderly's 1:00 AM can go out to 9 points to 12 heart rate mean values of heart rate mean value and the morning between 4 points
Now row 10%-20% does not occur, illustrates Cardiovascular abnormality.
It is drawn a portrait according to user and finds matched medical model, referred to: the linear regression point in disease of different user characteristics
The correspondence situation of analysis is different, and is average heart rate and oxygen demand, energy in one embodiment of the invention as shown in Fig. 6 a, Fig. 6 b
Relational graph between metabolic rate.Such as the morning breath increased heart rate degree and cardiovascular and cerebrovascular disease incidence of 55 years old or more age Hypertensive Population
It is in a linear relationship, age bracket crowd but non-linear relation within 55 years old, in this way, being found by information that S400 user draws a portrait pair
The medical model answered.
S600 such as finds matched medical model, carries out risk prompting to user according to matched medical model；
S700 does not such as find matched medical model, then is constructed according to user's sample characteristics by machine learning algorithm
New medical model；
In some embodiments, new medical model is constructed by machine learning algorithm, can carried out according to the following steps:
S701, raw data acquisition and cleaning: when finding that user has the specific undefined problem of body and mind aspect or disease, first
General character label is constructed with this.Then all sign information data of these full users, age, gender, heart rate, blood pressure, blood are collected
The parameter informations such as oxygen, constitutional index, waist-to-hipratio, heart rate variability, heart rate decelerations power, and these data are subjected to duplicate removal, standard
Change and error correction.
S702, data preliminary analysis and screening: find out maximum (small) values of the every class data of above-mentioned S701, average value, variance,
Median, certain particular values (such as zero) proportion or regularity of distribution etc., so as to these data have one it is preliminary
Analysis and understanding；It on the other hand is determining independent variable (x1, x2,x3,...,xn；Feature vector, i.e. input vector) and dependent variable y
(object vector, i.e. output vector), finds out the correlation of these independents variable and dependent variable, determines related coefficient；It will finally determine
Independent variable screened according to significance level, screening can select by hand or model selection, it is right after choosing suitable feature
These independents variable are named preferably to mark.
S703, data conversion:
This step be in order to which features described above data are converted to the data that machine learning algorithm can be easily recognized, such as in
The heart, normalization, vectorization etc., and the reprocessing to Feature Selection result indicate ability with Enhanced feature, prevent model
Excessively complicated and difficulty of learning.
S704, data set is split:
The data converted in S703 are randomly divided into two parts --- training set and verifying collection, training set is for new
Medical model training, verifying collection will be used to assess the performance situation after new medical model is trained.In a particular embodiment, training set
It is generally 8:2 or 7:3 with verifying collection proportion, it is of course also possible to select other ratios according to the actual situation.
S705, model training:
Before new medical model training, it is thus necessary to determine that suitable algorithm.By cross validation to logistic regression, decision tree,
Random forest, neural network, gradient are promoted and SVM these algorithms are tested and compared one by one, and it is each to adjust ginseng to ensure
Algorithm is optimal solution, then selects best one.After determining good algorithm, in conjunction with the training set data in S704 to new medicine
Model is trained optimization.
S706, model evaluation:
It completes after training, the verifying collection data in S705 to be used to assess trained model.Model evaluation is common
Method has confusion matrix, promotion figure, Lorentz figure, Gini coefficient, roc curve.After completing assessment, if model needs to optimize,
It can then be realized by adjusting ginseng, constantly repeat the process of S705 training and S706 assessment, until model meets the expected requirements.It is complete
At the building of new medical model.
It, can be online with further progress package interface and model, it may be assumed that new medicine mould after above-mentioned new medical model building
After type is up to standard, it can be called with packing service interface with implementation model, to be returned to prediction result.Medical model new in this way
It is online to use.
S800 voluntarily constructs new data index algorithm according to new medical model, specifically, for the heart rate in disease model
Change mathematical law model, it is pairs of to calculate pointer type and canonical form in classification as do not found and having been built up in changes in heart rate
It should be related to, then need to re-establish new array correspondence and criterion；
Crowd for different portraits and heart rate statistical analysis in different time points are sorted out, the class studied at present
Type be can not cover it is all, and now user portrait subdivision also not enough, such as the elderly classification, -65 years old 55 years old be one
A old age type, but with the increase of data volume, 55-65 grades sections are can to do more fining classification and portrait information
Recombination.Such as in study population's changes in heart rate and organ oxygen demand Long-term change trend, by original crowd be subdivided into male and female it
Afterwards, it can not find corresponding Segmentation Model in original system, but convolutional neural networks artificial intelligence can be with self learning type by corresponding crowd
Data are done after label subdivision to sort out and calculate.
Each parameter statistics of the men and women subject as shown in below table under different load:
To obtain the crowd portrayal more segmented and linear analysis as a result, then determining new index.
As shown in Fig. 7 a, Fig. 7 b, centre of motion rate and oxygen demand, the relationship of relative metabolic rate, data target are respectively indicated
Algorithm, which can simulate, increases the new major class index etc. of motion analysis, as shown in the table:
Certainly, above is only the case where being likely to occur in one embodiment, the invention is not limited to above situation,
When some news occur, the design or improvement of data target algorithm can be carried out according to actual needs, formed a kind of new
Data target algorithm.
New data index algorithm is replaced former data target algorithm by S900.
It further include S210 after step S200 on the basis of the above embodiments referring to shown in Fig. 2: for number after classifying
According to storing to big data platform data pool.
Referring to shown in Fig. 3, based on any of the above embodiments, step S800 includes: S801 from big data platform number
According to data after extraction classification in pond；S802 voluntarily constructs new data index algorithm according to data after new medical model combining classification.
Referring to shown in Fig. 4, based on any of the above embodiments, further comprising the steps of after step S100: S110 is gone
Except the obvious deviation data in initial data.In step S100, the acquisition of initial data is uninterrupted sampling initial data round the clock.
Corresponding to above-mentioned method, the present invention also provides it is a kind of can self-teaching medical diagnosis artificial intelligence system reality
Example is applied, which includes: raw data acquisition module, for carrying out raw data acquisition；Data categorization module, being used for will be original
The initial data that data collecting module collected arrives carries out classification processing, data after being classified；Data conversion module, for that will count
Data substitute into data target algorithm after the classification generated according to categorization module, carry out data conversion, form visual data；Data turn
Block is changed the mold, is also used to constantly carry out the verifying of data target algorithm；User, which draws a portrait, constructs module, for using data conversion module
Visual data building user's portrait of formation；Medical model matching judgment module constructs module building for drawing a portrait according to user
User portrait judge whether that matched medical model can be found；Risk reminding module, in medical model matching judgment mould
When block finds matched medical model, risk prompting is carried out to user；New medical model generation module, in medical model
When not finding matched medical model with judgment module, inform that user does not find matched medical model, and according to user's sample
Eigen constructs new medical model by machine learning algorithm；New data index algorithm constructs module, for according to new medicine
The new medical model that model generation module generates voluntarily constructs new data index algorithm；Data target algorithm replacement module, is used for
Data target algorithm in data conversion module is replaced with into the new data index that new data index algorithm building module constructs
Algorithm.
In one embodiment, it is above-mentioned can self-teaching medical diagnosis artificial intelligence system further include: big data platform
Data pool, data after the classification that categorization module generates for storing data.
In one embodiment, above-mentioned new data index algorithm building module includes: data extracting unit after classification, is used for
Before new data index algorithm constructs module, the data after proposing classification in big data platform data pool；New data index algorithm
Construction unit, data extracting unit is extracted after the new medical model combining classification for being generated according to new medical model generation module
To classification after data voluntarily construct new data index algorithm.
In one embodiment, it is above-mentioned can self-teaching medical diagnosis artificial intelligence system further include: cleaning module is used
Obvious deviation data in the initial data of removal raw data acquisition module acquisition.Raw data acquisition module round the clock not between
Disconnected acquisition initial data.
It is above-mentioned can each module and unit in the medical diagnosis artificial intelligence system embodiment of self-teaching, implement skill
Art can using can technology in step corresponding to medical diagnosis artificial intelligence approach of self-teaching, details are not described herein.
In conjunction with it is above-mentioned can self-teaching medical diagnosis artificial intelligence system and method, the embodiment of the present invention also provides one
Kind can self-teaching medical diagnosis artificial intelligence device, the device include memory, processor and storage on a memory simultaneously
The computer program that can be run on a processor, processor can be used for the medical diagnosis artificial intelligence system executed when executing program
Self-teaching method.
The above embodiment of the present invention can be collected the data of processing patient with 24 hours, and not influence the normal life of user,
And medical diagnosis artificial intelligence platform of the invention can carry out self-teaching, so that diagnosing more and more accurate, diagnosable disease
Range is more and more.Above-mentioned medical diagnosis artificial intelligence platform carries out intelligent diagnostics and study according to the data of collection, can be with
The monitor state of user is monitored, and reference or suggestion are provided according to the health data of monitoring, as shown in Figure 8,9, 10.
As shown in figure 11, in the preferred embodiment of part, the model training of above-mentioned S705 or new medical model generation module,
Neural network algorithm be can choose to be trained to model, trained and Forecasting recognition process be referred to following steps into
Row:
1. determining training set: the data (train data) in training set need to include characteristic value (input) and correspond to simultaneously
Label value (label), label value (label) are also referred to as target value (target).
2. planned network structure: determine the network number of plies, the number of nodes and activation primitive of each hidden layer and output layer
Activation primitive and loss function.Figure 11's is two layers of hidden layer, and activation primitive a () used in hidden layer is ReLu function, output
The function of layer is linear linear function (can also regard no activation primitive as).Loss function L () is for comparing distance
MSE:mean ((output-target)2), the smaller expression prediction effect of MSE is better, and training process is exactly continuous reduction MSE
Process.
3. weights initialisation: the W in Figure 11h1、Wh2、W0It cannot be sky before training, loss can be calculated by initializing,
Wh1、Wh1、W0Initialization determines which loss put initially as starting point in loss function from and trains network.
4. training network: training process is exactly to go out to export by network query function with the characteristic value (input) of training set data
It is worth (output), then calculates loss with output valve (output) and target value (target), then calculate gradient
(gradients) process of Lai Gengxin weight (weights):
A, positive transmitting: calculating the predicted value of current network,
Output=linear (W0*ReLu(Wh2*ReLu(Wh1*input+bh1)+bh2)+b0)
B, loss is calculated:
Loss=mean ((output-target)2)
C, calculate gradient: backpropagation calculates the corresponding gradient of each parameter, Figure 11 application since loss
For Stochastic Gradient Descent (SGD) to calculate gradient, i.e., updating gradient calculated every time all is from a sample
Originally it calculates.
D, update weight: the gradient of all parameters will all update:
W=W-Learningrate*gradient
E, it predicts new value: after training all samples, then upsetting sample order training several times.After training, when again
Carry out new data input, so that it may be predicted with the network that training is completed, output at this moment is exactly that effect is fine
Predicted value.Pre- flow gauge is on the right side of Figure 11 shown in Forecasting recognition end.
It should be noted that the step in method provided by the invention, can use corresponding module in system, device,
Unit etc. is achieved, and those skilled in the art are referred to the step process of the technical solution implementation method of system, that is, system
In embodiment can be regarded as the preference of implementation method, it will not be described here.
One skilled in the art will appreciate that in addition to realizing system provided by the invention in a manner of pure computer readable program code
And its other than each device, completely can by by method and step carry out programming in logic come so that system provided by the invention and its
Each device is in the form of logic gate, switch, specific integrated circuit, programmable logic controller (PLC) and embedded microcontroller etc.
To realize identical function.So system provided by the invention and its every device are considered a kind of hardware component, and it is right
The device for realizing various functions for including in it can also be considered as the structure in hardware component；It can also will be for realizing each
The device of kind function is considered as either the software module of implementation method can be the structure in hardware component again.
Specific embodiments of the present invention are described above.It is to be appreciated that the invention is not limited to above-mentioned
Particular implementation, those skilled in the art can make various deformations or amendments within the scope of the claims, this not shadow
Ring substantive content of the invention.
Claims (10)
1. a kind of self-teaching method of medical diagnosis artificial intelligence system characterized by comprising
Acquire medical diagnosis initial data；
The collected initial data is subjected to classification processing, data after being classified；
Data after the classification are substituted into data target algorithm, data conversion is carried out, form visual data；And constantly into
The verifying of the row data target algorithm；
Using the visual data, building user draws a portrait；
According to user portrait judge whether that matched medical model can be found, if so, carrying out risk prompting to user；It is no
Then, it informs that user does not find matched medical model, constructs new medicine by machine learning algorithm according to user's sample characteristics
Model；
New data index algorithm is voluntarily constructed according to the new medical model, the new data index algorithm is replaced into former data and is referred to
Mark algorithm.
2. a kind of self-teaching method of medical diagnosis artificial intelligence system according to claim 1, which is characterized in that institute
It states after being classified after data further include:
Data after the classification are stored to big data platform data pool.
3. a kind of self-teaching method of medical diagnosis artificial intelligence system according to claim 2, which is characterized in that root
New data index algorithm is voluntarily constructed according to the new medical model, comprising:
The data after extracting the classification in the big data platform data pool；
New data index algorithm is voluntarily constructed in conjunction with data after the classification according to the new medical model.
4. a kind of self-teaching method of medical diagnosis artificial intelligence system according to claim 1, which is characterized in that institute
State raw data acquisition, further includes: remove the obvious deviation data in the initial data；And/or the initial data is adopted
Collection, passes through intelligent terminal uninterrupted sampling initial data.
5. one kind can self-teaching medical diagnosis artificial intelligence system characterized by comprising
Raw data acquisition module, for carrying out medical diagnosis raw data acquisition；
Data categorization module, for the collected initial data of the raw data acquisition module to be carried out classification processing,
Data after being classified；
Data conversion module, data substitute into data target algorithm after the classification for generating the data categorization module
In, data conversion is carried out, visual data is formed；The data conversion module is also used to constantly carry out the data target algorithm
Verifying；
User, which draws a portrait, constructs module, visual data building user's portrait for being formed using the data conversion module；
Medical model matching judgment module, for the user portrait searching for constructing module building of being drawn a portrait according to the user
The medical model matched；
Risk reminding module, for carrying out risk when the medical model matching judgment module finds matched medical model
It reminds；
New medical model generation module, for not finding matched medical model in the medical model matching judgment module
When, inform that user does not find matched medical model, and new doctor is constructed by machine learning algorithm according to user's sample characteristics
Learn model；
New data index algorithm constructs module, the new medical model for being generated according to the new medical model generation module
Voluntarily construct new data index algorithm；
Data target algorithm replacement module, it is described for replacing with the data target algorithm in the data conversion module
The new data index algorithm that new data index algorithm building module constructs.
6. one kind according to claim 5 can self-teaching medical diagnosis artificial intelligence system, which is characterized in that also wrap
It includes: big data platform data pool, data after the classification for storing the data categorization module generation.
7. one kind according to claim 5 can self-teaching medical diagnosis artificial intelligence system, which is characterized in that it is described
New data index algorithm constructs module
Data extracting unit after classification is used for before the new data index algorithm constructs module, from the big data platform number
According to proposing data after the classification in pond；
New data index algorithm construction unit, the new medical model for being generated according to the new medical model generation module
Data voluntarily construct new data index algorithm after the classification extracted in conjunction with data extracting unit after the classification.
8. one kind according to claim 5 can self-teaching medical diagnosis artificial intelligence system, which is characterized in that also wrap
It includes: cleaning module, the obvious deviation data in the initial data for removing the raw data acquisition module acquisition.
9. one kind according to claim 8 can self-teaching medical diagnosis artificial intelligence system, which is characterized in that it is described
Raw data acquisition module uninterrupted sampling medical diagnosis initial data.
10. a kind of medical diagnosis artificial intelligence device, which is characterized in that on a memory including memory, processor and storage
And the computer program that can be run on a processor, which is characterized in that the processor can be used for executing when executing described program
Any method of claim 1-4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811639163.9A CN109767836A (en) | 2018-12-29 | 2018-12-29 | A kind of medical diagnosis artificial intelligence system, device and its self-teaching method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811639163.9A CN109767836A (en) | 2018-12-29 | 2018-12-29 | A kind of medical diagnosis artificial intelligence system, device and its self-teaching method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109767836A true CN109767836A (en) | 2019-05-17 |
Family
ID=66453036
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811639163.9A Pending CN109767836A (en) | 2018-12-29 | 2018-12-29 | A kind of medical diagnosis artificial intelligence system, device and its self-teaching method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109767836A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111428882A (en) * | 2020-03-27 | 2020-07-17 | 联想(北京)有限公司 | Processing method and computer equipment |
| CN111540471A (en) * | 2020-05-12 | 2020-08-14 | 西安交通大学医学院第一附属医院 | Health state tracking and early warning method and system based on user health data |
| CN111714135A (en) * | 2020-06-05 | 2020-09-29 | 安徽华米信息科技有限公司 | Method and device for determining blood oxygen saturation |
| CN114556241A (en) * | 2019-10-14 | 2022-05-27 | 西门子股份公司 | AI companion that integrates Artificial Intelligence (AI) into function blocks in a Programmable Logic Controller (PLC) program in automation |
| WO2022206641A1 (en) * | 2021-03-31 | 2022-10-06 | 华为技术有限公司 | Hypertension risk measurement method and related apparatus |
| WO2023281424A1 (en) * | 2021-07-09 | 2023-01-12 | Ayur.Ai (Opc) Private Limited | Integrative system and method for performing medical diagnosis using artificial intelligence |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104095615A (en) * | 2014-07-17 | 2014-10-15 | 上海翰临电子科技有限公司 | Human sleep monitoring method and system |
| CN105559754A (en) * | 2015-12-29 | 2016-05-11 | 西北工业大学 | Sleep-disordered breathing detection method and device based on heart rate and breathing signal |
| CN106097015A (en) * | 2016-06-20 | 2016-11-09 | 荆伟 | A kind of market prediction system and method |
| CN107403061A (en) * | 2017-07-07 | 2017-11-28 | 中北大学 | User's medical assessment model building method and medical assessment server |
| CN107731309A (en) * | 2017-08-31 | 2018-02-23 | 武汉百药联科科技有限公司 | A kind of Forecasting Methodology of pharmaceutical activity and its application |
| CN108460318A (en) * | 2017-02-22 | 2018-08-28 | 清华大学深圳研究生院 | Authentication/recognition methods based on electrocardiosignal and equipment |
| CN108597609A (en) * | 2018-05-04 | 2018-09-28 | 华东师范大学 | A kind of doctor based on LSTM networks is foster to combine health monitor method |
| CN109087706A (en) * | 2018-07-13 | 2018-12-25 | 浙江清华长三角研究院 | Human health appraisal procedure and assessment system based on sleep big data |
-
2018
- 2018-12-29 CN CN201811639163.9A patent/CN109767836A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104095615A (en) * | 2014-07-17 | 2014-10-15 | 上海翰临电子科技有限公司 | Human sleep monitoring method and system |
| CN105559754A (en) * | 2015-12-29 | 2016-05-11 | 西北工业大学 | Sleep-disordered breathing detection method and device based on heart rate and breathing signal |
| CN106097015A (en) * | 2016-06-20 | 2016-11-09 | 荆伟 | A kind of market prediction system and method |
| CN108460318A (en) * | 2017-02-22 | 2018-08-28 | 清华大学深圳研究生院 | Authentication/recognition methods based on electrocardiosignal and equipment |
| CN107403061A (en) * | 2017-07-07 | 2017-11-28 | 中北大学 | User's medical assessment model building method and medical assessment server |
| CN107731309A (en) * | 2017-08-31 | 2018-02-23 | 武汉百药联科科技有限公司 | A kind of Forecasting Methodology of pharmaceutical activity and its application |
| CN108597609A (en) * | 2018-05-04 | 2018-09-28 | 华东师范大学 | A kind of doctor based on LSTM networks is foster to combine health monitor method |
| CN109087706A (en) * | 2018-07-13 | 2018-12-25 | 浙江清华长三角研究院 | Human health appraisal procedure and assessment system based on sleep big data |
Non-Patent Citations (1)
| Title |
|---|
| 肖琴: ""基于互联网数据的个人信用风险评估的研究与应用"", 《中国优秀硕士学位论文全文数据库 经济与管理科学辑》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114556241A (en) * | 2019-10-14 | 2022-05-27 | 西门子股份公司 | AI companion that integrates Artificial Intelligence (AI) into function blocks in a Programmable Logic Controller (PLC) program in automation |
| CN114556241B (en) * | 2019-10-14 | 2024-05-07 | 西门子股份公司 | AI companion for integrating Artificial Intelligence (AI) into functional blocks in Programmable Logic Controller (PLC) programs in automation |
| CN111428882A (en) * | 2020-03-27 | 2020-07-17 | 联想(北京)有限公司 | Processing method and computer equipment |
| CN111540471A (en) * | 2020-05-12 | 2020-08-14 | 西安交通大学医学院第一附属医院 | Health state tracking and early warning method and system based on user health data |
| CN111540471B (en) * | 2020-05-12 | 2024-01-26 | 西安交通大学医学院第一附属医院 | Health state tracking and early warning method and system based on user health data |
| CN111714135A (en) * | 2020-06-05 | 2020-09-29 | 安徽华米信息科技有限公司 | Method and device for determining blood oxygen saturation |
| CN111714135B (en) * | 2020-06-05 | 2022-06-10 | 合肥华米微电子有限公司 | Method and device for determining blood oxygen saturation |
| WO2022206641A1 (en) * | 2021-03-31 | 2022-10-06 | 华为技术有限公司 | Hypertension risk measurement method and related apparatus |
| WO2023281424A1 (en) * | 2021-07-09 | 2023-01-12 | Ayur.Ai (Opc) Private Limited | Integrative system and method for performing medical diagnosis using artificial intelligence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109767836A (en) | A kind of medical diagnosis artificial intelligence system, device and its self-teaching method | |
| CN104287706B (en) | A kind of health status real time monitor and diagnosis and treatment commending system | |
| Abdali-Mohammadi et al. | Development of an IoT-based and cloud-based disease prediction and diagnosis system for healthcare using machine learning algorithms | |
| CN111951975B (en) | Sepsis early warning method based on deep learning model GPT-2 | |
| Ahmad et al. | Comparison between neural networks against decision tree in improving prediction accuracy for diabetes mellitus | |
| CN101916334B (en) | A kind of skin Forecasting Methodology and prognoses system thereof | |
| CN111081379B (en) | Disease probability decision method and system thereof | |
| CN114512239B (en) | Cerebral apoplexy risk prediction method and system based on transfer learning | |
| CN109872819A (en) | A kind of acute kidney injury incidence rate forecasting system based on Intensive Care Therapy detection | |
| CN111631688A (en) | Algorithm for automatic sleep staging | |
| CN110659677A (en) | Human body falling detection method based on movable sensor combination equipment | |
| CN110046757B (en) | Outpatient clinic volume prediction system and prediction method based on LightGBM algorithm | |
| CN108877928A (en) | Patient information acquisition method, device, computer equipment and storage medium | |
| CN110893100A (en) | Device and method for monitoring posture change based on plantar pressure sensor | |
| Arifoglu et al. | Abnormal behaviour detection for dementia sufferers via transfer learning and recursive auto-encoders | |
| Gan et al. | Human-computer interaction based interface design of intelligent health detection using PCANet and multi-sensor information fusion | |
| CN112768074A (en) | Artificial intelligence-based serious disease risk prediction method and system | |
| CN114550299A (en) | System and method for evaluating daily life activity ability of old people based on video | |
| CN116844720B (en) | Intelligent monitoring management system for debilitating health of old people | |
| CN106650206A (en) | Prediction method of high blood pressure based on incremental neural network model and prediction system | |
| CN113096792A (en) | Intelligent health monitoring and early warning method and system based on machine learning | |
| CN117542474A (en) | Remote nursing monitoring system and method based on big data | |
| CN115147768B (en) | Fall risk assessment method and system | |
| CN106295238A (en) | A kind of hypertensive nephropathy Forecasting Methodology based on increment type neural network model and prognoses system | |
| CN110598789A (en) | Human fatigue state prediction method and system based on fuzzy perceptron |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190517 |