CN106250712A - A kind of ureteral calculus Forecasting Methodology based on increment type neural network model and prognoses system - Google Patents

A kind of ureteral calculus Forecasting Methodology based on increment type neural network model and prognoses system Download PDF

Info

Publication number
CN106250712A
CN106250712A CN201610861253.7A CN201610861253A CN106250712A CN 106250712 A CN106250712 A CN 106250712A CN 201610861253 A CN201610861253 A CN 201610861253A CN 106250712 A CN106250712 A CN 106250712A
Authority
CN
China
Prior art keywords
neuron
ureteral calculus
network model
neural network
data
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
Application number
CN201610861253.7A
Other languages
Chinese (zh)
Inventor
杨滨
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hunan Old Code Information Technology Co Ltd
Original Assignee
Hunan Old Code Information Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hunan Old Code Information Technology Co Ltd filed Critical Hunan Old Code Information Technology Co Ltd
Priority to CN201610861253.7A priority Critical patent/CN106250712A/en
Publication of CN106250712A publication Critical patent/CN106250712A/en
Pending legal-status Critical Current

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
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients

Abstract

The invention discloses a kind of ureteral calculus Forecasting Methodology based on increment type neural network model, comprise the steps: to set up the daily data database of ureteral calculus；Neural network model is trained；Gather daily life data to send to server, preserve to the daily data logger of user；The daily data logger of user extracts data on the same day, forms n-dimensional vector, input in ureteral calculus pathology neural network model after doing normalized and carry out ureteral calculus probabilistic forecasting；Wired home ureteral calculus care appliances judges that whether ureteral calculus probit is more than 0.5；User is judged to ureteral calculus, and user removes examination in hospital voluntarily, and by wired home ureteral calculus care appliances, inspection result is sent back server, and server judges to check that result is the most correct；Perform increasable algorithm when checking result mistake, neural network model is carried out dynamic corrections.The present invention predicts accurately, makes neural network model to measure for each user.

Description

A kind of ureteral calculus Forecasting Methodology based on increment type neural network model and prediction System
Technical field
The invention belongs to field of medical technology, particularly relate to a kind of ureter based on increment type neural network model knot Stone Forecasting Methodology and prognoses system.
Background technology
Present Domestic is each health management system arranged is respectively provided with ureteral calculus prediction and evaluation, and its prediction mode used is data Coupling.Its principle is that by system matches fixed data, then personal lifestyle data entry system is shown ill probability.But due to Human body and the complexity of disease, unpredictability, in the bio signal form of expression with information, Changing Pattern (Self-variation Change with after medical intervention) on, it being detected and signal representation, the data of acquisition and the analysis of information, decision-making etc. are many All there is extremely complex non-linear relationship in aspect.So using traditional Data Matching can only be data examination blindly, nothing Method judges the logic association between data and data and variable, and the codomain deviation obtained is big, causes the specificity of system prediction The poorest, domestic health management system arranged effectively the ureteral calculus of individual cannot be carried out Accurate Prediction so current.
Major part is all to use BP neural network model to ureteral calculus prediction before this, but when new detection data are produced The when of raw, it is necessary to again training neural network model, operation efficiency is extremely low.And after system user scale increases, clothes Business device will be unable to complete in time training mission.
Summary of the invention
The purpose of the present invention is that and overcomes the deficiencies in the prior art, it is provided that a kind of based on increment type neural network model Ureteral calculus Forecasting Methodology and prognoses system, the present invention by neural network model training predict a large amount of patient in hospital pathology Data, find ureteral calculus pathology and ureteral calculus earlier life variations in detail, clinical symptoms, examination criteria value, high-risk Crowd characteristic, the logic association between these several the causes of disease and variable, ultimately form probability Accurate Prediction ill to ureteral calculus Ureteral calculus pathology neural network model, the present invention by gather user's daily life data, its data of active analysis Periodically, regular suffer from ureteral calculus probability eventually through ureteral calculus pathology Neural Network model predictive user, with The mode of visual effect reminds user's instant hospitalizing, is constantly repaiied by increasable algorithm when Neural Network model predictive is inaccurate Positive neural network model, trains the neural network model for this user, along with use to set up for each equipment user The increase of time, the neural network model made this user to measure with foundation, accuracy rate is greatly improved.
To achieve these goals, the invention provides a kind of ureteral calculus based on increment type neural network model pre- Survey method, comprises the steps:
Step (1), obtain hospital lithureteria because of pathology data source and patient's daily monitoring data, thus set up defeated The daily data database of ureteral calculi；
Step (2), the daily data database of ureteral calculus set up according to step (1) are off-line manner to nerve net Network model is trained, to obtain the ureteral calculus pathology neural network model trained；
Step (3), by intelligent monitoring device, the daily life data of user are acquired, and the daily life that will gather Live data sends to server, and the daily life data of user are preserved to the daily data logger of user by server；
Step (4), from the daily data logger of user, extract data on the same day, form n-dimensional vector, and n-dimensional vector is done The ureteral calculus pathology neural network model trained in input step (2) after normalized carries out ureteral calculus general Rate is predicted, obtains ureteral calculus probability, and ureteral calculus probability is sent to the nursing of wired home ureteral calculus by server Equipment；
After step (5), wired home ureteral calculus care appliances receive the ureteral calculus probability that server transmits, sentence Disconnected ureteral calculus probit, whether more than 0.5, if greater than 0.5, is then judged to that this user obtained ureteral calculus, attention device Warning is to remind user, if less than 0.5, is then judged to that this user does not obtain ureteral calculus；
Step (6), when user is judged to ureteral calculus, user removes examination in hospital voluntarily, and will check result Sending back server by wired home ureteral calculus care appliances, server judges to check that result is the most correct, if inspection Come to an end fruit mistake, then explanation ureteral calculus pathology Neural Network model predictive is inaccurate, if checking that result is correct, then illustrates Ureteral calculus pathology Neural Network model predictive is accurate；
Step (7), when checking result mistake, extract from the daily data logger of user the record in m days preserve to In incremental data table, when the record quantity in incremental data table is more than h bar, perform increasable algorithm, to lithureteria Reason neural network model carries out dynamic corrections；
Step (8), repetition step (3)～(7).
Further, the input layer of neural network model is n node, and hidden layer number is n*2+1, and output layer is 1 Node, extracts k bar record from ureteral calculus daily data database table and is trained, and every record is a n-dimensional vector, All data are before use first through normalized so that it is numerical value is interval in [0,1], then perform following steps to neutral net Model is trained:
1) one n-dimensional vector of input is to neural network model, calculates all of weight vector in neural network model defeated to this Entering the distance of n-dimensional vector, closest neuron is triumph neuron, and its computing formula is as follows:
| | x i - W k | | = min j ( | | x i - W j | | ) ;
Wherein: WkIt is the weight vector of triumph neuron, | | ... | | for Euclidean distance；
2) adjusting the weight vector of neuron in triumph neuron and triumph neuron field, formula is as follows:
Wherein: WjT () is neuron；Wj(t+1) weight vector before being adjustment and after adjustment；J belongs to triumph neuron neck Territory；α (t) is learning rate, and it is as the function that the increase of iterations is gradually successively decreased, and span is [0 1], through repeatedly It is 0.62 that Optimal learning efficiency is chosen in experiment；DjIt it is the distance of neuron j and triumph neuron；σ (t) is as the letter that the time successively decreases Number；All input n-dimensional vectors all are input in neural network model be trained by iteration each time, when the iteration reaching regulation After number of times, neural network model training terminates.
Further, wired home ureteral calculus care appliances will check that result sends back the object information of server Form is: { checking whether correct, blood glucose value }, and server is after receiving object information, it is judged that check that result is the most correct.
Further, the increasable algorithm that ureteral calculus pathology neural network model carries out dynamic corrections is:
Vectorial for every in incremental data table V{V1,V2,…,Vn, it is sent in neural network model learning function enter Row study, learning procedure is as follows:
1) first each to output layer weight vector is composed little random number and does normalized, then utilizes input mode vector V Meansigma methods Avg (V), be initialized as in neural network model the 0th layer the weights of unique neuron, and be set to nerve of winning Unit, calculates its quantization error QE；
2) from the neuron of the 0th layer, expand out 2 × 2 structures SOM, and its level identities Layer is set to 1；
3) for each 2 × 2 structure SOM subnet expanded out in Layer layer, the power of these 4 neurons is initialized Value；The input vector set Ci of i-th neuron is set to sky, and main label is set to the main label ratio r of NULL, neuron ii It is set to 0；The abnormity early warning data vector V of new SOM inherits the triumph input vector set VX of his father's neuron；
4) from VX, select a vectorial VXiDo following judgement:
If VXiFor the data of not tape label, then calculating the Euclidean distance of it and each neuron, chosen distance is the shortest Neuron is as triumph neuron；
If VXiFor the data of tape label, then select main label and VXiLabel is identical and riThe neuron work that value is maximum For triumph neuron, update this triumph neuron main label；
If can not find main label and VXiThe identical neuron of label, then find and VXiClosest neuron i makees For triumph neuron；
5) weights of neuron in triumph neuron and neighborhood thereof are adjusted, update the vector set W=W ∪ that wins {VXi, calculate the main label of triumph neuron, main label ratio riWith comentropy EiIf. the most predetermined frequency of training, then Go to step 4)；
6) calculate adjusted after this neural network model in quantization error QE of each neuroni, neuronal messages entropy Ei With the average quantization error MQE of subnet, formula is as follows:
QE i = Σ X j ∈ C i { | | W i - X j | | } ;
Wherein: WiFor the weight vector of neuron i, CiFor being mapped to the set that all input vectors of neuron i are constituted；
E i = Σ i ∈ T ( - l b n i m ) × n i m ;
Wherein: niRepresenting to fall that label is the number of samples of i on neuron, m represents to fall label data on neuron Sum, T represents to fall the sample label kind set on neuron；
Then judge:
If QE × threshold value q of MQE > father node, wherein q=0.71, then in this SOM, insert a line neuron, turn step Rapid 4)；
If Ei> E of father nodei× threshold value p, wherein p=0.42, then grow one layer of new subnet from this neuron, will The subnet newly grown increases in the subnet queue of Layer+1 layer；
If SOM is not inserted into the not longest subnet made new advances of new neuron, illustrate that this subnet has been trained；
7) all 2 × 2 structures SOM of the Layer+1 layer for newly expanding out, iteration operating procedure 3)～5) to it again Being trained, until neural network model no longer produces new neuron and new layering, whole training terminates.
Further, if user includes health check-up by other means and checks oneself, learn and oneself have suffered from ureteral calculus, and The attention device of wired home ureteral calculus care appliances does not warn, then it represents that wired home ureteral calculus care appliances is sentenced Disconnected inaccurate, now perform step (6)～(7), wired home ureteral calculus care appliances is sent to service object information On device.
Present invention also offers the prognoses system of a kind of described ureteral calculus Forecasting Methodology, including intelligent monitoring device, Smart machine data acquisition unit, server and wired home ureteral calculus care appliances, described intelligent monitoring device is with described Smart machine data acquisition unit is connected, and described smart machine data acquisition unit is by communication device one and described server network Communication, described wired home ureteral calculus care appliances is by communication device two and described server network communication.
Further, described wired home ureteral calculus care appliances is provided with attention device.
Further, described intelligent monitoring device include Intelligent worn device, Intelligent water cup, Intelligent weight claim, intelligence horse Bucket and Intelligent light sensing equipment.
Beneficial effects of the present invention:
1, the present invention predicts a large amount of patient in hospital pathological data by neural network model training, finds lithureteria Reason and ureteral calculus earlier life variations in detail, clinical symptoms, examination criteria value, high-risk group's feature, these several causes of disease it Between logic association and variable, ultimately form the ureteral calculus pathology nerve net of probability Accurate Prediction ill to ureteral calculus Network model, the present invention is by gathering user's daily life data, and the periodicity of its data of active analysis, regularity are eventually through defeated Ureteral calculi pathology Neural Network model predictive user suffers from ureteral calculus probability, reminds user i.e. in the way of visual effect Time seek medical advice and prevent.
2, constantly revise neural network model when Neural Network model predictive is inaccurate by increasable algorithm, with for Each equipment user sets up and trains the neural network model for this user, along with the increase of the time of use, to set up this The neural network model that user makes to measure, accuracy rate is greatly improved.
Accompanying drawing explanation
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing In having technology to describe, the required accompanying drawing used is briefly described, it should be apparent that, the accompanying drawing in describing below is only this Some embodiments of invention, for those of ordinary skill in the art, on the premise of not paying creative work, it is also possible to Other accompanying drawing is obtained according to these accompanying drawings.
Fig. 1 is the flow chart of the embodiment of the present invention.
Detailed description of the invention
Below in conjunction with the accompanying drawings invention is further illustrated, but be not limited to the scope of the present invention.
Embodiment
As it is shown in figure 1, a kind of based on increment type neural network model the ureteral calculus Forecasting Methodology that the present invention provides, Comprise the steps:
Step (1), obtain hospital lithureteria because of pathology data source and patient's daily monitoring data, thus set up defeated The daily data database of ureteral calculi；
The most daily monitoring data are 12 item data, and its 12 item data is body temperature, heart beating, heart rate, body fat, the amount of drinking water and frequency Rate, total urination time, urine color, body weight, the length of one's sleep and quality, every day 12 item data such as travel distance, the present invention is with 12 Data set up 12 dimensional vectors；
Step (2), the daily data database of ureteral calculus set up according to step (1) are off-line manner to nerve net Network model is trained, to obtain the ureteral calculus pathology neural network model trained；
Step (3), by intelligent monitoring device, the daily life data of user are acquired, and the daily life that will gather Live data sends to server, and the daily life data of user are preserved to the daily data logger of user by server；
Step (4), from the daily data logger of user, extract data on the same day, form 12 dimensional vectors, and to 12 dimensional vectors The ureteral calculus pathology neural network model trained in input step (2) after doing normalized carries out ureteral calculus Probabilistic forecasting, obtains ureteral calculus probability, and server sends ureteral calculus probability to wired home ureteral calculus and protects Reason equipment；
After step (5), wired home ureteral calculus care appliances receive the ureteral calculus probability that server transmits, sentence Disconnected ureteral calculus probit, whether more than 0.5, if greater than 0.5, is then judged to that this user obtained ureteral calculus, attention device Warning is to remind user, if less than 0.5, is then judged to that this user does not obtain ureteral calculus；
Step (6), when user is judged to ureteral calculus, user removes examination in hospital voluntarily, and will check result Sending back server by wired home ureteral calculus care appliances, server judges to check that result is the most correct, if inspection Come to an end fruit mistake, then explanation ureteral calculus pathology Neural Network model predictive is inaccurate, if checking that result is correct, then illustrates Ureteral calculus pathology Neural Network model predictive is accurate；
Step (7), when checking result mistake, extract from the daily data logger of user the record in 7 days preserve to In incremental data table, when the record quantity in incremental data table is more than 100, perform increasable algorithm, to ureteral calculus Pathology neural network model carries out dynamic corrections；
Step (8), repetition step (3)～(7).
The input layer of the neural network model of the present invention is 12 nodes, and hidden layer number is 25, and output layer is 1 node (i.e. the probability of ureteral calculus), extracts 400000 records from ureteral calculus daily data database table and is trained, Every record is 12 dimensional vectors, and all data are before use first through normalized so that it is numerical value is interval in [0,1], so Neural network model is trained by rear execution following steps:
1) one 12 dimensional vector of input are to neural network model, calculate all of weight vector in neural network model defeated to this Entering the distance of 12 dimensional vectors, closest neuron is triumph neuron, and its computing formula is as follows:
| | x i - W k | | = min j ( | | x i - W j | | ) ;
Wherein: WkIt is the weight vector of triumph neuron, | | ... | | for Euclidean distance；
2) adjusting the weight vector of neuron in triumph neuron and triumph neuron field, formula is as follows:
Wherein: WjT () is neuron；Wj(t+1) weight vector before being adjustment and after adjustment；J belongs to triumph neuron neck Territory；α (t) is learning rate, and it is as the function that the increase of iterations is gradually successively decreased, and span is [0 1], through repeatedly It is 0.62 that Optimal learning efficiency is chosen in experiment；DjIt it is the distance of neuron j and triumph neuron；σ (t) is as the letter that the time successively decreases Number；All input n-dimensional vectors all are input in neural network model be trained by iteration each time, when the iteration reaching regulation After number of times, neural network model training terminates.
The wired home ureteral calculus care appliances of the present invention will check that result sends back the object information of server Form is: { checking whether correct, blood glucose value }, and server is after receiving object information, it is judged that check that result is the most correct.
The increasable algorithm that ureteral calculus pathology neural network model carries out dynamic corrections of the present invention is:
Vectorial for every in incremental data table V{V1,V2,…,Vn, it is sent in neural network model learning function enter Row study, learning procedure is as follows:
1) first each to output layer weight vector is composed little random number and does normalized, then utilizes input mode vector V Meansigma methods Avg (V), be initialized as in neural network model the 0th layer the weights of unique neuron, and be set to nerve of winning Unit, calculates its quantization error QE；
2) from the neuron of the 0th layer, expand out 2 × 2 structures SOM, and its level identities Layer is set to 1；
3) for each 2 × 2 structure SOM subnet expanded out in Layer layer, the power of these 4 neurons is initialized Value；The input vector set Ci of i-th neuron is set to sky, and main label is set to the main label ratio r of NULL, neuron ii It is set to 0；The abnormity early warning data vector V of new SOM inherits the triumph input vector set VX of his father's neuron；
4) from VX, select a vectorial VXiDo following judgement:
If VXiFor the data of not tape label, then calculating the Euclidean distance of it and each neuron, chosen distance is the shortest Neuron is as triumph neuron；
If VXiFor the data of tape label, then select main label and VXiLabel is identical and riThe neuron work that value is maximum For triumph neuron, update this triumph neuron main label；
If can not find main label and VXiThe identical neuron of label, then find and VXiClosest neuron i makees For triumph neuron；
5) weights of neuron in triumph neuron and neighborhood thereof are adjusted, update the vector set W=W ∪ that wins {VXi, calculate the main label of triumph neuron, main label ratio riWith comentropy EiIf. the most predetermined frequency of training, then Go to step 4)；
6) calculate adjusted after this neural network model in quantization error QE of each neuroni, neuronal messages entropy Ei With the average quantization error MQE of subnet, formula is as follows:
QE i = Σ X j ∈ C i { | | W i - X j | | } ;
Wherein: WiFor the weight vector of neuron i, CiFor being mapped to the set that all input vectors of neuron i are constituted；
E i = Σ i ∈ T ( - l b n i m ) × n i m ;
Wherein: niRepresenting to fall that label is the number of samples of i on neuron, m represents to fall label data on neuron Sum, T represents to fall the sample label kind set on neuron；
Then judge:
If QE × threshold value q of MQE > father node, wherein q=0.71, then in this SOM, insert a line neuron, turn step Rapid 4)；
If Ei> E of father nodei× threshold value p, wherein p=0.42, then grow one layer of new subnet from this neuron, will The subnet newly grown increases in the subnet queue of Layer+1 layer；
If SOM is not inserted into the not longest subnet made new advances of new neuron, illustrate that this subnet has been trained；
7) all 2 × 2 structures SOM of the Layer+1 layer for newly expanding out, iteration operating procedure 3)～5) to it again Being trained, until neural network model no longer produces new neuron and new layering, whole training terminates.
If the user of the present invention includes health check-up by other means and checks oneself, learn and oneself have suffered from ureteral calculus, and The attention device of wired home ureteral calculus care appliances does not warn, then it represents that wired home ureteral calculus care appliances is sentenced Disconnected inaccurate, now perform step (6)～(7), wired home ureteral calculus care appliances is sent to service object information On device.
Present invention also offers the prognoses system of a kind of described ureteral calculus Forecasting Methodology, including intelligent monitoring device, Smart machine data acquisition unit, server and wired home ureteral calculus care appliances, described intelligent monitoring device is with described Smart machine data acquisition unit is connected, and described smart machine data acquisition unit is by communication device one and described server network Communication, described wired home ureteral calculus care appliances is by communication device two and described server network communication.
It is provided with attention device on the described wired home ureteral calculus care appliances of the present invention.
The described intelligent monitoring device of the present invention include Intelligent worn device, Intelligent water cup, Intelligent weight claim, intelligent closestool With Intelligent light sensing equipment etc..
The present invention predicts a large amount of patient in hospital pathological data by neural network model training, finds ureteral calculus pathology With ureteral calculus earlier life variations in detail, clinical symptoms, examination criteria value, high-risk group's feature, between these several the causes of disease Logic association and variable, ultimately form the ureteral calculus pathology neutral net of probability Accurate Prediction ill to ureteral calculus Model, the present invention is by gathering user's daily life data, and the periodicity of its data of active analysis, regularity are eventually through urine output Pipe calculus pathology Neural Network model predictive user suffers from ureteral calculus probability, reminds user instant in the way of visual effect Seek medical advice and prevent.
All data of the present invention preserve to server, can significantly save calculating cost, and hardware configuration is low, thus sells Valency is the lowest.
The present invention carries communication device one and communication device two, by wifi from being dynamically connected the Internet, and can protect for a long time Hold online.Various intelligent monitoring devices can easily access present device by the mode such as network or bluetooth, sets in acquisition The daily life data of the monitoring of intelligent monitoring device, the data that therefore present device obtains can be automatically uploaded after standby mandate Be real-time, accurately, polynary.
Owing to everyone physical trait is different, the data characteristics shown during ureteral calculus morbidity can not yet With.The most conventional is the highest by the method accuracy rate of neural network prediction ureteral calculus.The present invention is directed to each equipment use Family is set up and is trained the neural network model for this user, is running after a period of time, and generation is fixed to this customer volume body Doing neural network prediction model, accuracy rate is greatly improved.
When neural network model is judged by accident, error message be will be feedbacked to service by wired home ureteral calculus care appliances Device, for this user's dynamic corrections neural network model, when next time, similar characteristics data occurred in this user, will not miss again Sentence.Therefore, along with the increase of the time of use, the judgement of the wired home ureteral calculus care appliances of the present invention will be increasingly Accurately.
The ultimate principle of the present invention, principal character and advantages of the present invention have more than been shown and described.The technology of the industry Personnel, it should be appreciated that the present invention is not restricted to the described embodiments, simply illustrating this described in above-described embodiment and description The principle of invention, the present invention also has various changes and modifications without departing from the spirit and scope of the present invention, and these become Change and improvement both falls within scope of the claimed invention.Claimed scope by appending claims and Equivalent defines.

Claims (8)

1. a ureteral calculus Forecasting Methodology based on increment type neural network model, it is characterised in that comprise the steps:
Step (1), obtain hospital ureteral calculus and cure the disease etiology and pathology data source and patient's daily monitoring data, thus set up defeated The daily data database of ureteral calculi；
Step (2), the daily data database of ureteral calculus set up according to step (1) are off-line manner to neutral net mould Type is trained, to obtain the ureteral calculus pathology neural network model trained；
Step (3), by intelligent monitoring device, the daily life data of user are acquired, and the daily life number that will gather According to sending to server, the daily life data of user are preserved to the daily data logger of user by server；
Step (4), from the daily data logger of user, extract data on the same day, form n-dimensional vector, and n-dimensional vector is done normalizing The ureteral calculus pathology neural network model trained in input step (2) after change process carries out ureteral calculus probability pre- Surveying, obtain ureteral calculus probability, ureteral calculus probability is sent to wired home ureteral calculus care appliances by server；
After step (5), wired home ureteral calculus care appliances receive the ureteral calculus probability that server transmits, it is judged that defeated Whether ureteral calculi probit is more than 0.5, if greater than 0.5, is then judged to that this user obtained ureteral calculus, and attention device warns To remind user, if less than 0.5, then it is judged to that this user does not obtain ureteral calculus；
Step (6), when user is judged to ureteral calculus, user removes examination in hospital voluntarily, and inspection result is passed through Wired home ureteral calculus care appliances sends back server, and server judges to check that result is the most correct, if checking knot Really mistake, then explanation ureteral calculus pathology Neural Network model predictive is inaccurate, if checking that result is correct, then urine output is described Pipe calculus pathology Neural Network model predictive is accurate；
Step (7), when checking result mistake, the record extracted from the daily data logger of user in m days preserves to increment In tables of data, when the record quantity in incremental data table is more than h bar, perform increasable algorithm, to ureteral calculus pathology god Dynamic corrections is carried out through network model；
Step (8), repetition step (3)～(7).
A kind of ureteral calculus Forecasting Methodology based on increment type neural network model the most according to claim 1, it is special Levying and be, the input layer of neural network model is n node, and hidden layer number is n*2+1, and output layer is 1 node, from urine output Extracting k bar record in pipe calculus daily data database table to be trained, every record is a n-dimensional vector, and all data exist First through normalized before using so that it is numerical value is interval in [0,1], then perform following steps and neural network model is carried out Training:
1) one n-dimensional vector of input is to neural network model, calculates all of weight vector in neural network model and ties up to this input n The distance of vector, closest neuron is triumph neuron, and its computing formula is as follows:
Wherein: WkIt is the weight vector of triumph neuron, | | ... | | for Euclidean distance；
2) adjusting the weight vector of neuron in triumph neuron and triumph neuron field, formula is as follows:
Wherein: WjT () is neuron；wj(t+1) weight vector before being adjustment and after adjustment；J belongs to triumph neuron field；α T () is learning rate, it is as the function that the increase of iterations is gradually successively decreased, and span is [0 1], through many experiments Choosing Optimal learning efficiency is 0.62；DjIt it is the distance of neuron j and triumph neuron；σ (t) is as the function that the time successively decreases； All input n-dimensional vectors all are input in neural network model be trained by iteration each time, when the iteration time reaching regulation After number, neural network model training terminates.
A kind of ureteral calculus Forecasting Methodology based on increment type neural network model the most according to claim 1, it is special Levying and be, the form of the object information that inspection result sends back server is by wired home ureteral calculus care appliances: { inspection Look into the most correct, blood glucose value }, server is after receiving object information, it is judged that check that result is the most correct.
A kind of ureteral calculus Forecasting Methodology based on increment type neural network model the most according to claim 1, it is special Levying and be, the increasable algorithm that ureteral calculus pathology neural network model carries out dynamic corrections is:
Vectorial for every in incremental data table V{V1,V2,…,Vn, it is sent in neural network model learning function learn Practising, learning procedure is as follows:
1) first each to output layer weight vector is composed little random number and does normalized, then utilizes the flat of input mode vector V Average Avg (V), is initialized as in neural network model the 0th layer the weights of unique neuron, and is set to triumph neuron, meter Calculate its quantization error QE；
2) from the neuron of the 0th layer, expand out 2 × 2 structures SOM, and its level identities Layer is set to 1；
3) for each 2 × 2 structure SOM subnet expanded out in Layer layer, the weights of these 4 neurons are initialized；Will The input vector set Ci of i-th neuron is set to sky, and main label is set to the main label ratio r of NULL, neuron iiIt is set to 0；The abnormity early warning data vector V of new SOM inherits the triumph input vector set VX of his father's neuron；
4) from VX, select a vectorial VXiDo following judgement:
If VXiFor the data of not tape label, then calculate the Euclidean distance of it and each neuron, the nerve that chosen distance is the shortest Unit is as triumph neuron；
If VXiFor the data of tape label, then select main label and VXiLabel is identical and riThe neuron of value maximum is as obtaining Victory neuron, updates this triumph neuron main label；
If can not find main label and VXiThe identical neuron of label, then find and VXiClosest neuron i is as obtaining Victory neuron；
5) weights of neuron in triumph neuron and neighborhood thereof are adjusted, update the vector set W=W ∪ { VX that winsi, Calculate the main label of triumph neuron, main label ratio riWith comentropy EiIf. the most predetermined frequency of training, then go to step 4)；
6) calculate adjusted after this neural network model in quantization error QE of each neuroni, neuronal messages entropy EiAnd son The average quantization error MQE of net, formula is as follows:
Wherein: WiFor the weight vector of neuron i, CiFor being mapped to the set that all input vectors of neuron i are constituted；
Wherein: niRepresenting to fall that label is the number of samples of i on neuron, m represents to fall the total of label data on neuron Number, T represents to fall the sample label kind set on neuron；
Then judge:
If QE × threshold value q of MQE > father node, wherein q=0.71, then in this SOM, insert a line neuron, go to step 4)；
If Ei> E of father nodei× threshold value p, wherein p=0.42, then grow one layer of new subnet from this neuron, will be the longest The subnet gone out increases in the subnet queue of Layer+1 layer；
If SOM is not inserted into the not longest subnet made new advances of new neuron, illustrate that this subnet has been trained；
7) all 2 × 2 structures SOM of the Layer+1 layer for newly expanding out, iteration operating procedure 3)～5) it is re-started Training, until neural network model no longer produces new neuron and new layering, whole training terminates.
A kind of ureteral calculus Forecasting Methodology based on increment type neural network model the most according to claim 1, it is special Levy and be, if user includes health check-up by other means and checks oneself, learn and oneself have suffered from ureteral calculus, and wired home is defeated The attention device of ureteral calculi care appliances does not warn, then it represents that wired home ureteral calculus care appliances judges inaccurate, Now performing step (6)～(7), wired home ureteral calculus care appliances is sent to object information on server.
6. one kind uses the prognoses system of ureteral calculus Forecasting Methodology described in claim 1～6, it is characterised in that include intelligence Energy monitoring device, smart machine data acquisition unit, server and wired home ureteral calculus care appliances, described intelligent monitoring Equipment is connected with described smart machine data acquisition unit, and described smart machine data acquisition unit passes through communication device one with described Server network communication, described wired home ureteral calculus care appliances is led to described server network by communication device two News.
The prognoses system of ureteral calculus Forecasting Methodology the most according to claim 7, it is characterised in that described wired home is defeated It is provided with attention device on ureteral calculi care appliances.
The prognoses system of ureteral calculus Forecasting Methodology the most according to claim 7, it is characterised in that described intelligent monitoring sets Standby include Intelligent worn device, Intelligent water cup, Intelligent weight claim, intelligent closestool and Intelligent light sensing equipment.
CN201610861253.7A 2016-09-28 2016-09-28 A kind of ureteral calculus Forecasting Methodology based on increment type neural network model and prognoses system Pending CN106250712A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610861253.7A CN106250712A (en) 2016-09-28 2016-09-28 A kind of ureteral calculus Forecasting Methodology based on increment type neural network model and prognoses system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610861253.7A CN106250712A (en) 2016-09-28 2016-09-28 A kind of ureteral calculus Forecasting Methodology based on increment type neural network model and prognoses system

Publications (1)

Publication Number Publication Date
CN106250712A true CN106250712A (en) 2016-12-21

Family

ID=57611850

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610861253.7A Pending CN106250712A (en) 2016-09-28 2016-09-28 A kind of ureteral calculus Forecasting Methodology based on increment type neural network model and prognoses system

Country Status (1)

Country Link
CN (1) CN106250712A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108197737A (en) * 2017-12-29 2018-06-22 山大地纬软件股份有限公司 A kind of method and system for establishing medical insurance hospitalization cost prediction model
CN109685796A (en) * 2018-12-26 2019-04-26 上海联影智能医疗科技有限公司 Medical image processing method, device, equipment and storage medium
CN111096735A (en) * 2018-10-26 2020-05-05 深圳市理邦精密仪器股份有限公司 Electrocardiogram analysis system capable of being updated iteratively
CN112802544A (en) * 2021-03-12 2021-05-14 杭州度安医学检验实验室有限公司 System for predicting urinary calculus based on machine learning and metabonomics

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1477581A (en) * 2003-07-01 2004-02-25 南京大学 Predictive modelling method application to computer-aided medical diagnosis
CN1768696A (en) * 2005-09-22 2006-05-10 上海交通大学 Network based diabetes pedigree control system
CN102647292A (en) * 2012-03-20 2012-08-22 北京大学 Intrusion detecting method based on semi-supervised neural network
CN202855026U (en) * 2012-10-17 2013-04-03 山东中创软件工程股份有限公司 Internet of things monitoring information system
CN105023220A (en) * 2015-08-14 2015-11-04 北京大学第三医院 Health guidance system for polycystic ovarian syndrome
CN105118010A (en) * 2015-09-30 2015-12-02 成都信汇聚源科技有限公司 Chronic disease management method with functions of real-time data processing and real-time information sharing and life style intervention information
CN105912829A (en) * 2015-11-12 2016-08-31 上海童舒房科技股份有限公司 Bad habit correcting system based on cloud

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1477581A (en) * 2003-07-01 2004-02-25 南京大学 Predictive modelling method application to computer-aided medical diagnosis
CN1768696A (en) * 2005-09-22 2006-05-10 上海交通大学 Network based diabetes pedigree control system
CN102647292A (en) * 2012-03-20 2012-08-22 北京大学 Intrusion detecting method based on semi-supervised neural network
CN202855026U (en) * 2012-10-17 2013-04-03 山东中创软件工程股份有限公司 Internet of things monitoring information system
CN105023220A (en) * 2015-08-14 2015-11-04 北京大学第三医院 Health guidance system for polycystic ovarian syndrome
CN105118010A (en) * 2015-09-30 2015-12-02 成都信汇聚源科技有限公司 Chronic disease management method with functions of real-time data processing and real-time information sharing and life style intervention information
CN105912829A (en) * 2015-11-12 2016-08-31 上海童舒房科技股份有限公司 Bad habit correcting system based on cloud

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
MEHMED KANTARDZIC: "《数据挖掘：概念、模型、方法和算法》", 31 January 2013 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108197737A (en) * 2017-12-29 2018-06-22 山大地纬软件股份有限公司 A kind of method and system for establishing medical insurance hospitalization cost prediction model
CN111096735A (en) * 2018-10-26 2020-05-05 深圳市理邦精密仪器股份有限公司 Electrocardiogram analysis system capable of being updated iteratively
CN109685796A (en) * 2018-12-26 2019-04-26 上海联影智能医疗科技有限公司 Medical image processing method, device, equipment and storage medium
CN109685796B (en) * 2018-12-26 2021-05-18 上海联影智能医疗科技有限公司 Medical image processing method, apparatus, device and storage medium
CN112802544A (en) * 2021-03-12 2021-05-14 杭州度安医学检验实验室有限公司 System for predicting urinary calculus based on machine learning and metabonomics

Similar Documents

Publication Publication Date Title
CN106446560A (en) Hyperlipidemia prediction method and prediction system based on incremental neural network model
CN106407699A (en) Coronary heart disease prediction method and prediction system based on incremental neural network model
CN106355035A (en) Pneumonia prediction method and prediction system based on incremental neural network model
CN106446552A (en) Prediction method and prediction system for sleep disorder based on incremental neural network model
CN106250712A (en) A kind of ureteral calculus Forecasting Methodology based on increment type neural network model and prognoses system
CN106384013A (en) Incremental neural network model-based type-II diabetes prediction method and prediction system
CN106650206A (en) Prediction method of high blood pressure based on incremental neural network model and prediction system
CN106295238A (en) A kind of hypertensive nephropathy Forecasting Methodology based on increment type neural network model and prognoses system
CN106384012A (en) Incremental neural network model-based allergic dermatitis prediction method and prediction system
CN106355034A (en) Sub-health prediction method and prediction system based on incremental neural network model
CN106250715A (en) A kind of chronic pharyngolaryngitis Forecasting Methodology based on increment type neural network model and prognoses system
CN106202986A (en) A kind of tonsillitis Forecasting Methodology based on increment type neural network model and prognoses system
CN106384008A (en) Incremental neural network model-based allergic rhinitis prediction method and prediction system
CN106407694A (en) Neurasthenia prediction method and prediction system based on incremental neural network model
CN106407693A (en) Hepatitis B prediction method and prediction system based on incremental neural network model
CN106384005A (en) Incremental neural network model-based depression prediction method and prediction system
CN106250714A (en) A kind of renal calculus Forecasting Methodology based on increment type neural network model and prognoses system
CN106295239A (en) A kind of fatty liver Forecasting Methodology based on increment type neural network model and prognoses system
CN106339605A (en) Colonitis prediction method and colonitis prediction system based on incremental nerve network model
CN106407695A (en) Anxiety disorder prediction method and prediction system based on incremental neural network model
CN106446553A (en) Stomach illness prediction method and prediction system based on incremental neural network model
CN106372442A (en) Dental ulcer prediction method and system based on incremental neural network model
CN106446551A (en) Incremental neural network model based chronic gastroenteritis prediction method and system
CN106446550A (en) Cold prediction method and system based on incremental neutral network model
CN106384009A (en) Incremental neural network model-based HIV prediction method and prediction system

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20161221