CN106446765A - Health state evaluation system based on multidimensional physiological big data depth learning - Google Patents
Health state evaluation system based on multidimensional physiological big data depth learning Download PDFInfo
- Publication number
- CN106446765A CN106446765A CN201610589952.0A CN201610589952A CN106446765A CN 106446765 A CN106446765 A CN 106446765A CN 201610589952 A CN201610589952 A CN 201610589952A CN 106446765 A CN106446765 A CN 106446765A
- Authority
- CN
- China
- Prior art keywords
- data
- physiological
- data set
- sample
- pond
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Abstract
The present invention discloses a health state evaluation system based on multidimensional physiological big data depth learning. The health state evaluation system utilizes an unsupervised convolutional neural network to extract the characteristics of the physiological time series data, and then utilizes the multivariate gauss distribution to carry out the anomaly detection on the characteristics. Results display that the system is an efficient anomaly detection system of the physiological signals which can learn the high-level characteristics of the signals from the original physiological signals and carry out the multivariate gauss distribution anomaly detection, users can discriminate some early stage diseases and adopt the corresponding prevention measures in advance to reduce the prevalence risk.
Description
Technical field
The present invention relates to medical treatment detection device.
Background technology
Recent years, our life and various smart machine contact more and more tightr, the life in modern times
Increasingly it be unable to do without these smart machines.Using these smart machines, present people can record various whenever and wherever possible
Physiological signal, as blood pressure, blood sugar, brain electricity, electrocardio, myoelectricity, body temperature, breathing physiological signal, by analyze these letter
Number, we will be seen that the information of some our body present situations.
But, the incidence of disease but more and more higher of various types of Chronic Non-Communicable Diseases.One of reason is medical resource
Relative shortage, the fast crowd of rhythm of life often just arrives examination in hospital after more serious problem in body and seeks medical advice, and leads to
The delay for the treatment of.
Content of the invention
It is an object of the invention to provide a kind of abnormality detection system not relying on pre-existing standard, can accurately detect
Potential health problem.
Employed technical scheme comprise that such for realizing the object of the invention, one kind is based on multidimensional physiology big data depth
Practise health status evaluation system it is characterised in that:Including background analysis system and detection means;
Described background analysis system is to original physiologic data sample CqWhen being analyzed, comprise the following steps:
1) obtain original physiologic data sample Cq, sample CqIn there is m kind physiological parameter, each physiological parameter is all in n0
The individual moment is collected;
2):By original physiologic data sample Cq, normalized, obtain Sq;
3) feature extraction
The physiological data sample S that selected physiological data is concentratedqInput variable, using the side setting convolutional layer sliding window
Method, obtain the data set H of Feature Mapping value compositionk;
Set the size of pond window, maximum pond data set Hk, obtain data set Ik:
By iteration, finally give characteristic data set IK;
4) with data set IKFor multivariate Gaussian distribution input, obtain Gaussian probability density distribution function p (X), set ε as
Threshold value;
The state model that secures good health is:
Described detection means includes data collection section data analysis part;Described data collection section collects tested
The physiological data of person is saved in matrix E, and passes to data analysis component;
In E, there is m kind physiological parameter, each physiological parameter is all in pn0The individual moment is collected;
Using with step 2) identical method, by data sample E, normalized obtains F;
The physiological data sample F input variable that selected physiological data is concentrated, the method using setting convolutional layer sliding window,
Obtain the data set PH of Feature Mapping value compositionk;
Set the size of pond window, maximum pond data set PHk, obtain data set PIk:
By iteration, finally give characteristic data set PIK;
By eigenmatrix PIKAs described health status modelPhysiological characteristic input matrix, defeated
Go out abnormal or normal.
The solution have the advantages that mathematical:Using multilayer convolutional network structure, extract phase from input signal
The feature representation closing, then delivers this to multivariate Gaussian abnormal distribution detection model to detect off-note.It is a kind of efficient
Can from the high-level feature of original physiological signal learning signal and multivariate Gaussian abnormal distribution detection physiological signal different
Often detecting system.
Brief description
The FB(flow block) of Fig. 1 present invention;
Fig. 2 deep neural network structure;
Fig. 3 autocoding structure chart;
The Gaussian Profile of Fig. 4 feature.
Specific embodiment
With reference to embodiment, the invention will be further described, but only should not be construed the above-mentioned subject area of the present invention
It is limited to following embodiments.Without departing from the idea case in the present invention described above, according to ordinary skill knowledge and used
With means, make various replacements and change, all should include within the scope of the present invention.
Embodiment 1:
A kind of health status evaluation system based on multidimensional physiology big data deep learning it is characterised in that:Including backstage
Analysis system and detection means;
Described background analysis system is to original physiologic data sample CqWhen being analyzed, comprise the following steps:
1) obtain original physiologic data sample Cq, q is physiology data sample numbering, q=1,2 ...,
Wherein: For a physiological parameter, sample CqIn, there is m kind
Physiological parameter (as blood pressure, heart rate, body weight, body temperature etc.), each physiological parameter is all collected in t, t=1,2 ... n0;
2):By original physiologic data sample Cq, normalized, obtainStructure
Build physiological data set { S1, S2 ... }
3) feature extraction
3-1) select the physiological data sample S that physiological data is concentratedqInput variable, is that input variable sets convolutional layer slip
Length L of windowkWith step-length Ak, extract each line parameter successively from input variable, using each described in sliding window traversal
Line parameter, is cut into some data slotsWherein, i isRound up the value of acquisition, k=1,2 ... K, and K is volume
The long-pending network number of plies;
3-2) with data slotAnd linear filterConvolution, adds a biasing bkInput as differentiable function
Variable, obtains the data set H of Feature Mapping value compositionk, described differentiable function be selected from sigmoid, thanh, ReLU or
softplus.
3-3) it is sized as GKPond window, maximum pond data set Hk, obtain data set Ik:
3-4) with IkTo update step 3-1) described in input variable, k is updated with k+1, reset convolutional layer slide
Length L of windowk+1With step-length Ak+1, resetting size is Gk+1Pond window, reset wave filterAgain
Set biasing bk+1;Repeat 3-2) arrive 3-3), finally give characteristic data set IK,
LKAnd AKIt is respectively kth and carry out step 3-2) arrive 3-3) when, the setting convolutional layer sliding window length of setting and step
Long;GKCarry out step 3-2 for kth) arrive 3-3) when, the pond window size of setting;
4) with data set IKFor multivariate Gaussian distribution input, obtain Gaussian probability density distribution function p (X), set ε as
Threshold value;
The state model that secures good health is:
Described detection means includes data collection section data analysis part;Described data collection section collects tested
The physiological data of person is saved in matrix E, and passes to data analysis component;
Wherein:
For a physiological parameter, in sample E, there is m
Plant physiological parameter, each physiological parameter is all collected in the pt moment, pt=1,2 ... pn0;
A) adopt and step 2) identical method, by data sample E, normalized obtains
B the physiological data sample F) selecting physiological data concentration is input variable, is that input variable sets convolutional layer slip
Length L of windowkWith step-length Ak, extract each line parameter successively from input variable, using each described in sliding window traversal
Line parameter, is cut into some data slotsWherein, u isRound up the value of acquisition, k=1,2 ... K, and K is
The convolutional network number of plies;
C) with data slotAnd linear filterConvolution, adds a biasing pbkAs sigmoid,
The input variable of the differentiable functions such as thanh, ReLU and softplus, obtains the data set PH of Feature Mapping value compositionk:
D) it is sized as GKPond window, maximum pond data set PHk, obtain data set PIk:
E) with PIkTo update step 3-1) described in input variable, k is updated with k+1, resets convolutional layer sliding window
Length L of mouthk+1With step-length Ak+1, resetting size is Gk+1Pond window, reset wave filterAgain set
Surely bias pbk+1;Repeat 3-2) arrive 3-3), finally give characteristic data set PIK, repeat 3-2) and arrive 3-3), finally give characteristic
According to collection PIK,
LKAnd AKIt is respectively kth and carry out step C) arrive D) when, the setting convolutional layer sliding window length of setting and step-length;
GKCarry out step C for kth) arrive D) when, the pond window size of setting;
By eigenmatrix PIKAs described health status modelPhysiological characteristic input matrix, defeated
Go out abnormal or normal.
Embodiment 2:
A kind of health status evaluation system based on multidimensional physiology big data deep learning it is characterised in that:Including backstage
Analysis system and detection means;
Described background analysis system is to original physiologic data sample CqWhen being analyzed, comprise the following steps:
1) obtain original physiologic data sample Cq, q is physiology data sample numbering, q=1,2 ...,
Wherein: For a physiological parameter, sample CqIn, there is m kind
Physiological parameter, each physiological parameter is all collected in t, t=1,2 ... n0;
2):By original physiologic data sample Cq, normalized, obtainStructure
Build physiological data set { S1, S2 ... }
3) (in physiological signal abnormality detection, network structure designs as shown in Fig. 2 mainly including characterology for feature extraction
Practise and abnormality detection two parts)
3-1) select the physiological data sample S that physiological data is concentratedqInput variable, is that input variable sets convolutional layer slip
Length L of windowkWith step-length Ak, extract each line parameter successively from input variable, using each described in sliding window traversal
Line parameter, is cut into some data slotsWherein, i isRound up the value of acquisition, k=1,2 ... K, and K is volume
The long-pending network number of plies;
3-2) with data slotAnd linear filterConvolution, adds a biasing bkInput as differentiable function
Variable, obtains the data set H of Feature Mapping value compositionk, described differentiable function be selected from sigmoid, thanh, ReLU or
softplus.
3-3) it is sized as GKPond window, maximum pond data set Hk, obtain data set Ik:
Layer modal wave filter size in pond is 2 × 2, wave filter along the x-axis of image and y-axis, with certain step-length
Slide and travel through whole picture, and in this patent, because signal is one-dimensional clock signal, so wave filter is designed as 2 ×
1, travel through this clock signal along time-axis direction.
3-4) with IkTo update step 3-1) described in input variable, k is updated with k+1, reset convolutional layer slide
Length L of windowk+1With step-length Ak+1, resetting size is Gk+1Pond window, reset wave filterAgain
Set biasing bk+1;Repeat 3-2) arrive 3-3), finally give characteristic data set IK,
LKAnd AKIt is respectively kth and carry out step 3-2) arrive 3-3) when, the setting convolutional layer sliding window length of setting and step
Long;GKCarry out step 3-2 for kth) arrive 3-3) when, the pond window size of setting;
Autocoder passes through to determine the parameter of coding function, and an input space is converted into a new distribution and expression
Process be referred to as encode.As shown in figure 3, autocoder passes through to reduce reconstructed error, determine the parameter of decoding functions, study why
Sample is referred to as decoding by exporting the process reconstructing input signal space, and wherein the coding parameter of autocoder is also used to reconstruct
Input signal.
In this patent, autocoding is used for training non-supervisory convolutional neural networks parameter, from original input signal learning
The character representation of signal.And the feature obtaining from input signal learning can be looked for this as the input of grader in turn
To some features of input signal or find mapping relations between input and target.
4) with data set IKFor multivariate Gaussian distribution input, obtain Gaussian probability density distribution function p (X), set ε as
Threshold value;
The state model that secures good health is:
In embodiment, the feature of obtained eight physiological signals of input meets Gaussian Profile (Fig. 4).If primary signal
The ratio of exceptional value is 1%, at this moment threshold epsilon=0.2, it is possible to obtain one group of abnormal data;If primary signal exceptional value
Ratio is 5%, at this moment threshold epsilon=0.3, it is possible to obtain another group of abnormal data.In embodiment, threshold epsilon can be set to
0.2nd, 0.23,0.25 and 0.26, the ratio respectively obtaining abnormal data is 1%, 2%, 3% and 5%, and ratio is regarded as exception
The serious degree of data.
Described detection means includes data collection section data analysis part;Described data collection section collects tested
The physiological data of person is saved in matrix E, and passes to data analysis component;
Wherein:
For a physiological parameter, in sample E, there is m
Plant physiological parameter, each physiological parameter is all collected in the pt moment, pt=1,2 ... pn0;
A) adopt and step 2) identical method, by data sample E, normalized obtains
B the physiological data sample F) selecting physiological data concentration is input variable, is that input variable sets convolutional layer slip
Length L of windowkWith step-length Ak, extract each line parameter successively from input variable, using each described in sliding window traversal
Line parameter, is cut into some data slotsWherein, u isRound up the value of acquisition, k=1,2 ... K, K
For the convolutional network number of plies;
C) with data slotAnd linear filterConvolution, adds a biasing pbkAs sigmoid,
The input variable of the differentiable functions such as thanh, ReLU and softplus, obtains the data set PH of Feature Mapping value compositionk:
D) it is sized as GKPond window, maximum pond data set PHk, obtain data set PIk:
E) with PIkTo update step 3-1) described in input variable, k is updated with k+1, resets convolutional layer sliding window
Length L of mouthk+1With step-length Ak+1, resetting size is Gk+1Pond window, reset wave filterAgain set
Surely bias pbk+1;Repeat 3-2) arrive 3-3), finally give characteristic data set PIK, repeat 3-2) and arrive 3-3), finally give characteristic
According to collection PIK,
LKAnd AKIt is respectively kth and carry out step C) arrive D) when, the setting convolutional layer sliding window length of setting and step-length;
GKCarry out step C for kth) arrive D) when, the pond window size of setting;
By eigenmatrix PIKAs described health status modelPhysiological characteristic input matrix, defeated
Go out abnormal or normal.
Claims (1)
1. a kind of health status evaluation system based on multidimensional physiology big data deep learning it is characterised in that:Divide including backstage
Analysis system and detection means;
Described background analysis system is to original physiologic data sample CqWhen being analyzed, comprise the following steps:
1) obtain described original physiologic data sample Cq, sample CqIn there is m kind physiological parameter, each physiological parameter is all in n0
The individual moment is collected;
2):By original physiologic data sample Cq, normalized, obtain Sq;
3) feature extraction
The physiological data sample S that selected physiological data is concentratedqInput variable, the method using setting convolutional layer sliding window, obtain
The data set H constituting to Feature Mapping valuek;
Set the size of pond window, maximum pond data set Hk, obtain data set Ik:
By iteration, finally give characteristic data set IK;
4) with data set IKFor the input of multivariate Gaussian distribution, obtain Gaussian probability density distribution function p (X), set ε as threshold value;
The state model that secures good health is:
Described detection means includes data collection section data analysis part;Described data collection section collects measured's
Physiological data is saved in matrix E, and passes to data analysis component;
In E, there is m kind physiological parameter, each physiological parameter is all in pn0The individual moment is collected;
Using with step 2) identical method, by data sample E, normalized obtains F;
The physiological data sample F input variable that selected physiological data is concentrated, the method using setting convolutional layer sliding window, obtain
The data set PH that Feature Mapping value is constitutedk;
Set the size of pond window, maximum pond data set PHk, obtain data set PIk:
By iteration, finally give characteristic data set PIK;
By eigenmatrix PIKAs described health status modelPhysiological characteristic input matrix, output different
Often or normal.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610589952.0A CN106446765B (en) | 2016-07-26 | 2016-07-26 | A kind of health status evaluation system based on multidimensional physiology big data deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610589952.0A CN106446765B (en) | 2016-07-26 | 2016-07-26 | A kind of health status evaluation system based on multidimensional physiology big data deep learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106446765A true CN106446765A (en) | 2017-02-22 |
CN106446765B CN106446765B (en) | 2019-09-03 |
Family
ID=58184654
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610589952.0A Active CN106446765B (en) | 2016-07-26 | 2016-07-26 | A kind of health status evaluation system based on multidimensional physiology big data deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106446765B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107374632A (en) * | 2017-07-21 | 2017-11-24 | 青岛康庆和医药科技有限责任公司 | Breath sound monitoring device and its application method in a kind of surgical operation |
CN107491638A (en) * | 2017-07-28 | 2017-12-19 | 深圳和而泰智能控制股份有限公司 | A kind of ICU user's prognosis method and terminal device based on deep learning model |
CN109085918A (en) * | 2018-06-28 | 2018-12-25 | 天津大学 | Acupuncture needling manipulation training method based on myoelectricity |
CN109614576A (en) * | 2018-12-11 | 2019-04-12 | 福建工程学院 | Transformer exception detection method based on Multi-dimensional Gaussian distribution and trend segmentation |
CN110334869A (en) * | 2019-08-15 | 2019-10-15 | 重庆大学 | A kind of mangrove forest ecological health forecast training method based on dynamic colony optimization algorithm |
CN111898194A (en) * | 2020-05-25 | 2020-11-06 | 北京空间飞行器总体设计部 | Evaluation and prediction method for health tolerance of individual spacecraft in-orbit space radiation environment |
CN113807527A (en) * | 2020-06-11 | 2021-12-17 | 华硕电脑股份有限公司 | Signal detection method and electronic device using same |
CN116564527A (en) * | 2023-07-11 | 2023-08-08 | 南京裕隆生物医学发展有限公司 | Respiratory health analysis method based on physiological feature deep learning |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104182772A (en) * | 2014-08-19 | 2014-12-03 | 大连理工大学 | Gesture recognition method based on deep learning |
CN104346625A (en) * | 2013-08-09 | 2015-02-11 | 株式会社东芝 | Medical image processing device, medical image processing method and classifier training method |
CN104750880A (en) * | 2015-04-28 | 2015-07-01 | 苏州大学 | Big data-based early warning method and big data-based early warning method for human body cold resistance |
-
2016
- 2016-07-26 CN CN201610589952.0A patent/CN106446765B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104346625A (en) * | 2013-08-09 | 2015-02-11 | 株式会社东芝 | Medical image processing device, medical image processing method and classifier training method |
CN104182772A (en) * | 2014-08-19 | 2014-12-03 | 大连理工大学 | Gesture recognition method based on deep learning |
CN104750880A (en) * | 2015-04-28 | 2015-07-01 | 苏州大学 | Big data-based early warning method and big data-based early warning method for human body cold resistance |
Non-Patent Citations (3)
Title |
---|
MARTIN LANGKVIST等: "A review of unsupervised feature learning and deep learning for time-series modeling", 《PATTERN RECOGNITION LETTERS》 * |
于冰洁等: "基于高斯过程模型的异常检测算法", 《计算机工程与设计》 * |
雷亚国等: "基于深度学习理论的机械装备大数据健康监测方法", 《机械工程学报》 * |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107374632A (en) * | 2017-07-21 | 2017-11-24 | 青岛康庆和医药科技有限责任公司 | Breath sound monitoring device and its application method in a kind of surgical operation |
CN107374632B (en) * | 2017-07-21 | 2021-12-24 | 浙江通用海特医疗科技有限公司 | Respiratory sound monitoring device in surgical operation and use method thereof |
CN107491638A (en) * | 2017-07-28 | 2017-12-19 | 深圳和而泰智能控制股份有限公司 | A kind of ICU user's prognosis method and terminal device based on deep learning model |
CN109085918B (en) * | 2018-06-28 | 2020-05-12 | 天津大学 | Myoelectricity-based acupuncture needle manipulation training method |
CN109085918A (en) * | 2018-06-28 | 2018-12-25 | 天津大学 | Acupuncture needling manipulation training method based on myoelectricity |
CN109614576B (en) * | 2018-12-11 | 2022-08-30 | 福建工程学院 | Transformer anomaly detection method based on multi-dimensional Gaussian distribution and trend segmentation |
CN109614576A (en) * | 2018-12-11 | 2019-04-12 | 福建工程学院 | Transformer exception detection method based on Multi-dimensional Gaussian distribution and trend segmentation |
CN110334869A (en) * | 2019-08-15 | 2019-10-15 | 重庆大学 | A kind of mangrove forest ecological health forecast training method based on dynamic colony optimization algorithm |
CN111898194A (en) * | 2020-05-25 | 2020-11-06 | 北京空间飞行器总体设计部 | Evaluation and prediction method for health tolerance of individual spacecraft in-orbit space radiation environment |
CN111898194B (en) * | 2020-05-25 | 2023-08-22 | 北京空间飞行器总体设计部 | Spacecraft individual on-orbit space radiation environment tolerance health evaluation and prediction method |
CN113807527A (en) * | 2020-06-11 | 2021-12-17 | 华硕电脑股份有限公司 | Signal detection method and electronic device using same |
CN116564527A (en) * | 2023-07-11 | 2023-08-08 | 南京裕隆生物医学发展有限公司 | Respiratory health analysis method based on physiological feature deep learning |
CN116564527B (en) * | 2023-07-11 | 2023-09-15 | 南京裕隆生物医学发展有限公司 | Respiratory health analysis method based on physiological feature deep learning |
Also Published As
Publication number | Publication date |
---|---|
CN106446765B (en) | 2019-09-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106446765A (en) | Health state evaluation system based on multidimensional physiological big data depth learning | |
CN103034837B (en) | Characteristic parameter is associated with pulse condition key element | |
CN107832737A (en) | Electrocardiogram interference identification method based on artificial intelligence | |
CN116705337B (en) | Health data acquisition and intelligent analysis method | |
CN106821376A (en) | A kind of epileptic attack early warning system and method based on deep learning algorithm | |
CN109567793B (en) | Arrhythmia classification-oriented ECG signal processing method | |
CN106419893A (en) | Method and device for detecting sleep state | |
CN106503799A (en) | Deep learning model and the application in brain status monitoring based on multiple dimensioned network | |
CN105595990A (en) | Intelligent terminal device for evaluating and distinguishing quality of electrocardiosignal | |
CN108511055B (en) | Ventricular premature beat recognition system and method based on classifier fusion and diagnosis rules | |
CN108597601A (en) | Diagnosis of chronic obstructive pulmonary disease auxiliary system based on support vector machines and method | |
CN110020714B (en) | Model training and data analysis method, device, equipment and storage medium | |
CN108256453A (en) | A kind of method based on one-dimensional ECG signal extraction two dimension CNN features | |
CN109645989A (en) | A kind of depth of anesthesia estimation method and system | |
CN113768519B (en) | Method for analyzing consciousness level of patient based on deep learning and resting state electroencephalogram data | |
CN107137093A (en) | A kind of continuous blood sugar monitoring device for including abnormal plasma glucose probability alarm | |
CN114358194A (en) | Gesture tracking based detection method for abnormal limb behaviors of autism spectrum disorder | |
CN107242876A (en) | A kind of computer vision methods for state of mind auxiliary diagnosis | |
CN110916672A (en) | Old people daily activity monitoring method based on one-dimensional convolutional neural network | |
CN113576472A (en) | Blood oxygen signal segmentation method based on full convolution neural network | |
CN117133464B (en) | Intelligent monitoring system and monitoring method for health of old people | |
CN113611414A (en) | Dynamic monitoring and performance evaluation method and system for acute altitude disease | |
CN107595305A (en) | Anxiety state detection method and device | |
CN107837083A (en) | J ripple automatic testing methods based on least square method supporting vector machine | |
CN111466877B (en) | LSTM network-based oxygen reduction state prediction method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |