CN109801713A - A kind of health risk prediction technique based on schematic models - Google Patents
A kind of health risk prediction technique based on schematic models Download PDFInfo
- Publication number
- CN109801713A CN109801713A CN201910090787.8A CN201910090787A CN109801713A CN 109801713 A CN109801713 A CN 109801713A CN 201910090787 A CN201910090787 A CN 201910090787A CN 109801713 A CN109801713 A CN 109801713A
- Authority
- CN
- China
- Prior art keywords
- health
- node
- record
- foreign peoples
- health examination
- 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
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The health risk prediction technique based on schematic models that the invention discloses a kind of, comprising the following steps: the input form that the health examination of n S1, setting participants record；S2, schemed by foreign peoples of the health examination record building based on HER；S3, weight matrix is obtained from foreign peoples's figure by SHG-Health algorithm, and disease risks is predicted by weight matrix.The present invention predicts disease risks using semi-supervised learning (SSL) algorithm, and explores foreign peoples's figure based on HER, is classified in foreign peoples's figure to the case where gradually development with most of unlabelled data.It can be by inquiring participant piRecord, SHG-Health predict participant piWhether belong to high risk disease category or " unknown " classification, there is significant prediction effect on health examination data set and generated data collection, it is higher compared to other prediction model prediction accuracies, great contribution is made that disease risks prediction field.
Description
Technical field
The present invention relates to prediction model field more particularly to a kind of health risk prediction techniques based on schematic models.
Background technique
Semi-supervised learning (SSL) causes more and more in the health care application based on electric health record (EHRs)
Concern.For example, SSL method of one of the patent network based on figure, this method can understand patient risk's group, carry out patient
Risk stratification；A kind of SSL joint training method based on figure is predicted for breast cancer survival rate.When learner reaches common understanding
When, pseudo label is iteratively distributed to no label data by it, and includes pseudo label example, Zhi Daowu in the set for having label
Tag set stops reducing；A kind of two part figures of the Lung neoplasm image classification based on sequence a kind of are constructed based on sequence of events
The time diagram etc. of time phenotype；But no one of these methods consider " unknown " class, and they have for all
The predefined example of class is either still passed through what other mechanism were realized by expert.
General health inspection identifies risky participant to morning in the component part that many countries are health cares
It is critically important that phase early warning and prevention, which are intervened,.Although electric health record has attracted more and more researchs to focus in recent years
Come data mining and machine learning community, excavates general health and check that data are one and not yet sufficiently inquire into, in addition to minority
To healthy score taxonomy model in the propositions of chronic diseases early warning system such as research risk prediction and in the past work.However, these
Unlabelled data are not all accounted for.The basic challenge of study risk profile disaggregated model is that Unlabeled data constitutes
The major part of collected data.Particularly, unlabelled data describe health examination participant, their health status may
It is very different from health to grave illness, distinguishing their health status does not have the answer of standard.
Summary of the invention
Object of the present invention is in view of the above-mentioned problems, providing a kind of health risk prediction technique based on schematic models.
To achieve the goals above, the technical scheme is that
Method of the present invention by excavating health examination record (HERs), constructs a kind of semi-supervised learning based on chart
Algorithm is used for the model to risk profile, and this algorithm is referred to as SHG-Health (semi-supervised isomery health figure), explores one
A foreign peoples's figure based on HER, referred to as HeteroHER figure are divided with most of unlabelled data the case where gradually development
Class.Firstly, health examination record graphically illustrates, all relevant cases are linked together.Secondly, the multiclass of capture data item
Type relationship, and be naturally mapped in isomery figure.Third, characteristic by the label communication process on foreign peoples's figure with oneself
Type weighting.
The modeling method of the risk forecast model the following steps are included:
The Fitness Testing that n participants are arranged in step 1) records input form；
Step 2) constructs HeteroHER graph model；
Step 3) establishes risk forecast model.
Step 1) the concrete operations are as follows:
1.1) it setsIt is the n of participant iiThe set of a record, riIt is (xij, tij) a member
Element,It is in time tijA d dimensional vector, then S={ s1..., sl, sl+1..., snIt is n participants
Fitness Testing set of records ends.Tag set C={ 1 ..., C }, preceding l participant si(i≤l) is marked as yi∈ C, remaining u=
N-l participant sl+1..., sl+u(l < < u) will not be labeled.
1.2) it defines:In this wayIndicate the label of i type node.WithIndicate vector kth
Element.If xip(p node of type i) is marked with, xipWhen belonging to k classOtherwiseIf
xipIt is not labeled,
1.3) it enablesFor the soft label of calculating of m node type,Indicate the vector of certainty degree, xipBelong to any c+1 class.xipClass label byIt calculates.
Step 2) the concrete operations are as follows:
Graphical modeling allows to sparse data modeling.In order to capture the heterogeneity of naturally occurring in health examination project,
The present invention constructs the chart of an entitled HeteroHER, wherein including the polymorphic type node recorded based on health examination.
2.1) by all record value discretizations and it is converted into 0-1 binary representation first, lacks/exist as discrete value
Indicator vector.Particularly, for real value, such as age, it is first divided into fixed time interval.Then, by all ordinal numbers
Binary representation is converted to classification value.
2.2) node is inserted into: being obtained each element that binary form indicating value is 1 in step 2.1 and is modeled as
A node in HeteroHER figure, unlike, only abnormal results are modeled as detection project (including physics and the heart
Reason).
2.3) node type: each node category of test according to belonging to its original value is divided, and record class is connected to
The every other non-recorded type node of type node can be regarded as the attribute of these record type nodes.
2.4) link insertion: each attribute (non-recorded) type node is linked to a record type node, the node
It indicates to observe initial record.The weight of link is to assume to calculate based on one, and it is newer that this is assumed to be record, in risk
Prediction aspect is more important.It is as follows to define a simple function g ():
G (t)=(t-s+1) ÷ l
In formula, t is current record time, and l is time window of interest, at the beginning of s is time window.It can also make
With other functions, Gaussian Profile and chi square distribution is such as truncated.Length of window is the period for the record that model considers.Length of window
Only setting range.The contribution of model is recorded in link weight function control time t.
2.5) an isomery figure is exported, any type node i of two interconnections, one group of weight matrix of j are expressed as
Wij。
Step 3) the concrete operations are as follows:
Risk forecast model is made of SHG-Health algorithm and HeteroHER figure:
3.1) using health examination data (GHE) and the dead label related causes of description as SHG-Health algorithm
Input.
3.2) schemed to obtain weight matrix W by HeteroHERij。
3.3) pass throughCalculate the normalized weight of i, j=1 ..., m.
Wherein DijIt is ni×niDiagonal matrix, (p, p) element are dIj, pp, dIj, pp=∑qWIj, pqIt is WijThe sum of middle row p.
3.4) F is uniformly initialized in i type nodei, i=1 ..., m.
3.5) z=[z is set1..., zm]T, 0≤zj≤ 1 is the weight of j-th of class node.γijIt is defined as i class node and j
Class node intermediate form weight:
3.6) soft label of the i type node at (t+1) is determined by following two aspect: 1) is propagated in t moment by link
The calculating label score of adjacent node, 2) initial labels of .i type node.Diagonal matrixIt controls flat between both influences
Weighing apparatus.
3.7) pass through:
Update Fi。
WhereinIt is niRank diagonal matrix, wherein (p, p) element is
3.8) work as FiWhen convergence, F is returnedi。
Compared with prior art, the advantages and positive effects of the present invention are:
The health risk prediction technique based on schematic models that the invention proposes a kind of utilizes semi-supervised learning (SSL)
Algorithm predicts that disease risks, semi-supervised learning (SSL) algorithm is referred to as SHG-Health (semi-supervised isomery health
Figure), and foreign peoples's figure based on HER is explored, with most of unlabelled data to gradually developing in foreign peoples's figure
Situation is classified.Firstly, health examination record is indicated with chart, all relevant cases are linked together；Then, it catches
The polymorphic type relationship of detection data item is obtained, and is naturally mapped in isomery figure；Finally, characteristic passes through the mark on foreign peoples's figure
Communication process is signed to weight with the type of oneself.The present invention can be by inquiring participant piRecord, SHG-Health prediction ginseng
With person piWhether belong to high risk disease category or " unknown " classification, has on health examination data set and generated data collection
Significant prediction effect, compared to other prediction models, prediction accuracy is higher, is made that pole to disease risks prediction field
Big contribution.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention without any creative labor, may be used also for those of ordinary skill in the art
To obtain other drawings based on these drawings.
Fig. 1 is the health examination record schematic diagram that participant takes one's test three discontinuous times；
Fig. 2 is the health inspection inspection record foreign peoples figure extracted from Fig. 1；
Fig. 3 is model structure of the invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment, any modification, equivalent replacement, improvement and so on should all be included in the protection scope of the present invention.
The health risk prediction technique based on schematic models that the invention proposes a kind of utilizes semi-supervised learning (SSL)
Algorithm predicts that disease risks, semi-supervised learning (SSL) algorithm is referred to as SHG-Health (semi-supervised isomery health
Figure), and foreign peoples's figure based on HER is explored, referred to as HeteroHER figure, with most of unlabelled data in foreign peoples
Classify in figure to the case where gradually development.Its forecasting problem for mainly solving Unlabeled data；The risk forecast model
Modeling method the following steps are included:
Step 1, the Fitness Testing that n participants are arranged records input form,
Step 2, HeteroHER graph model is constructed,
Step 3, risk forecast model is established.
The step 1 includes
The Fitness Testing record input for defining n participants is set: S={ s1..., sl, sl+1..., sn, whereinIt is the n of participant iiThe set of a record, riIt is (xij, tij) an element,
It is in time tijA d dimensional vector.Tag set C={ 1 ..., C }, preceding l participant si(i≤l) is marked as yi∈
C, remaining u=n-l participant sl+1..., sl+u(l < < u) will not be labeled.
The step 1 further includes
Definition:In this wayIndicate the label of i type node.WithIndicate vector yipThe
K element.If xip(p node of type i) is marked with, xipWhen belonging to k classOtherwiseSuch as
Fruit xipIt is not labeled,
The step 1 further includes
It enablesFor the soft label of calculating of m node type,Indicate the vector of certainty degree, xipBelong to any c+1 class.xipClass label byIt calculates.
The step 2 includes
By all record value discretizations and it is converted into 0-1 binary representation first ,/existing index is lacked as discrete value
Vector.Particularly, for real value, such as age, it is first divided into fixed time interval.Then, all ordinal sums are classified
Value is converted to binary representation.
Each element that binary form indicating value is 1 will be obtained and be modeled as a node in HeteroHER figure, it is different
, only abnormal results are modeled as detection project (including physics and psychology).As shown in Fig. 1, this is participant p1Three
One example of the health examination record that a discontinuous time takes one's test, it is abnormal including different classes of test item
Result queue is black.Configuration file C includes demographic information and the habit of all patients.
Each node category of test according to belonging to its original value is divided, be connected to record type node it is all its
His non-recorded type node can be regarded as the attribute of these record type nodes.
Each attribute (non-recorded) type node is linked to a record type node, which indicates to observe initially
Record.The weight of link is to assume to calculate based on one, and it is newer that this is assumed to be record, heavier in terms of risk profile
It wants.It is as follows to define a simple function g ():
G (t)=(t-s+1) ÷ l
In formula, t is current record time, and l is time window of interest, at the beginning of s is time window.It can also make
With other functions, Gaussian Profile or chi square distribution is such as truncated.Length of window is the period for the record that model considers.Length of window
Only setting range.The contribution of model is recorded in link weight function control time t.
An isomery figure is exported, any type node i of two interconnections, one group of weight matrix W of j are expressed asij.Such as
Shown in Fig. 2, the Fitness Testing to extract from Fig. 1 example records isomery figure, if a3The result of (Section 3 of A class) is in r11
(p1First record) in abnormal, then r11And a3There is link.Link indicates weight with g (t),Window width is equal to 6 years.Right figure star chart-pattern in Fig. 2 is this
The type grade mode of one figure of sample.
The step 3 includes
Using health examination data (GHE) and the dead label related causes of description as the input of SHG-Health algorithm.
As shown in figure 3, being the SHG-Health algorithm that is proposed to the model of risk profile.
The normalized weight of i, j=1 ..., m are calculated by following formula:
Wherein DiJ is ni×niDiagonal matrix, (p, p) element are dIj, pp, dIj, pp=∑qWIj, pqIt is WijThe sum of middle row p.
F is uniformly initialized in i type nodei, i=1 ..., m.
If z=[z1..., zm]T, 0≤zj≤ 1 is the weight of j-th of class node.
γijIt is defined as i class node and j class node intermediate form weight:
Soft label of the i type node at (t+1) is determined by following two aspect: 1) passes through the adjacent of link propagation in t moment
The calculating label score of node, 2) initial labels of .i type node.Diagonal matrixControl the balance between both influences.In parameter alphaip(p=1 ..., ni), in xipα is defined as when labeledip=αl, otherwise αip=αu。αlAnd αuMore big then Y
The influence of initial labels is smaller.Particularly, work as αuWhen being set as very close 1 value, it means that Unlabeled data it is initial
Label does not almost work in study.
F is updated by following formulai:
WhereinIt is niRank diagonal matrix, (p, p) element are
Work as FiWhen convergence, F is returnedi。
In order to illustrate technical effect of the invention, implementation verifying is carried out to the present invention by experiment.
Experiment, which has been used, two is combined into data set based on what real data set distribution generated, is equilibrium data collection and not respectively
Flag data collection.Select 10 ICD disease categories as high-risk disease category.First three bit digital of ICD 10 is for defining disease
Classification, and it is mapped to the accordingly point 3 ICD9 codes being previously recorded for 2009.Model four kinds of node types: record, physics are surveyed
Examination, psychological assessment and configuration file.By inquiring the record of participant, SHG-Health can predict piWhether high risk is belonged to
Disease category or " unknown " classification.
Experiment flow:
Experiment one:
It lists the elder's health and checks determinant attribute, such as table 1:
Table 1
The 10 kinds of disease categories divided by participation number (P) and record (R) and unmarked case are listed, such as table 2:
Table 2
Experiment two:
It executes following algorithm: (1) using the SHG-Health (algorithm 1) of g (t) structure figures；(2) using Chi squares of truncation
The SHG-Health (algorithm 2) of weighting function structure figures；(3) using the SHG-Health (algorithm of truncation Gaussian function structure figures
3)；(4) support vector machines (algorithm 4)；(5) nearest neighbor classifier (algorithm 5)；(7) GNetMine (algorithm 6)；(7) based on figure
General semi-supervised learning (algorithm 7).
This experimental design two stages assessment strategy assesses the SHG-Health of proposition.
In the first stage, the ability i.e. energy of testing algorithm prediction mortality risk of algorithm identification high risk case history is had evaluated
Power.Due to not having brass tacks for passive or health case, this experiment has used the method for laying particular emphasis on and actively predicting, i.e., quasi-
True property recalls/sensitivity and F- score.Harmonic-mean between F- score counting accuracy and full degree of looking into.
In second stage, the correct disease category of assessment prediction recalls measure using macroscopical accuracy and macroscopic view.Macroscopic view is average
Take the average value of the calculated accuracy rate of each class or recall rate.It is assumed that all classifications are all of equal importance, therefore few
The performance of several classes of types can reflect on macroscopical average mark.
Experimental result and analysis
Table 3 shows two-value prediction result.
Table 3
In the first stage of two stages assessment, algorithm 2 obtains 99.33% accuracy rate, and SHG-Health algorithm obtains
43.93% recall rate and 60.32% F scoring.The recall rate of algorithm 7 is 100%, but its accuracy rate is extremely low, is
5.21%.Algorithm 6 is partial to " unknown " class completely, therefore has zero accuracy rate, recall rate and F points.Institute is methodical to recall point
The fact that number is lower than 50% also indicates that, from it is big, noisy, and do not capture positive case in markd case and be difficult.
Table 4 further has evaluated condition performance of the algorithm in multicategory classification.
Table 4
In second stage, condition performance of the algorithm in multicategory classification is further had evaluated.The display of table 4, the SHG- of this paper
Health, especially algorithm 2 are better than every other algorithm, and macro accuracy rate is 90.58%, and macro recall rate is 90.73%.
In conclusion the present invention proposes a kind of health risk prediction technique based on schematic models, and in real world
The validity of the model is demonstrated in HERs data set.According to experimental result, this method can be effective by selection weighting function
Solve the forecasting problem of Unlabeled data.
Claims (4)
1. a kind of health risk prediction technique based on schematic models, it is characterised in that: the following steps are included:
The input form that the health examination of n S1, setting participants record；
S2, schemed by foreign peoples of the health examination record building based on HER；
S3, weight matrix is obtained from foreign peoples's figure by SHG-Health algorithm, and disease risks is carried out by weight matrix
Prediction.
2. the health risk prediction technique based on schematic models as described in claim 1, it is characterised in that: in the step S1
Be arranged n participants health examination record input form the following steps are included:
S11, it setsIt is the n of participant iiThe set of a health examination record, riIt is (xij, tij) one
A element,It is in time tijA d dimensional vector, S={ s1..., Sl, Sl+1..., snIt is n participants
Health examination set of records ends, C={ 1 ..., c } be tag set, preceding l participant si(i≤l) is marked as yi∈ C,
Remaining mono- l of u=n (l < < u) position participant sl+1..., sl+uIt will not be labeled；
S12, it setsThenIndicate the label of i type node, ifFor vector yipKth
A element；Work as xipWhen labeled, xipWhen belonging to k classxipWhen being not belonging to k classWork as xipNot by
When label,
S13, it setsFor the soft label of calculating of m node type,Indicate the vector of certainty degree.
3. the health risk prediction technique based on schematic models as claimed in claim 2, it is characterised in that: in the step S2
By health examination record building based on HER foreign peoples figure the following steps are included:
S21,0-1 binary representation is converted for all health examination record according to step S12；
S22, a node in foreign peoples's figure is set by the element that the binary form indicating value obtained in step S21 is 1；
S23, each node inspection classification according to belonging to its original value is divided；
S24, link is inserted into the node of each classification, obtains foreign peoples's figure.
4. the health risk prediction technique based on schematic models as claimed in claim 3, it is characterised in that: in the step S3
Disease risks are predicted by weight matrix the following steps are included:
S31, pass through formulaCalculate the normalized weight of i, j=1 ..., m；
Wherein, WijFor weight matrix, DijIt is ni×niDiagonal matrix, (p, p) element are dIj, pp, dIj, pp=∑qWIj, pqIt is WijMiddle row
The sum of p；
S32, type i node to FiCarry out unification；
S33, z=[z is set1..., zm]T,0≤zj≤ 1 is the weight of j-th of class node, γijFor type i-node and type j
Node intermediate form weight, then:
S34, pass through formulaTo FiIt is updated, wherein It is ni×niAngular moment battle array, (p, p) element are dIj, pp,Pass through FiIt obtains
Disease risks prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910090787.8A CN109801713A (en) | 2019-01-30 | 2019-01-30 | A kind of health risk prediction technique based on schematic models |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910090787.8A CN109801713A (en) | 2019-01-30 | 2019-01-30 | A kind of health risk prediction technique based on schematic models |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109801713A true CN109801713A (en) | 2019-05-24 |
Family
ID=66559273
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910090787.8A Pending CN109801713A (en) | 2019-01-30 | 2019-01-30 | A kind of health risk prediction technique based on schematic models |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109801713A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110298669B (en) * | 2019-06-27 | 2023-09-12 | 上海麦克风文化传媒有限公司 | iOS charging risk control system based on graphic structure |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107658029A (en) * | 2017-09-29 | 2018-02-02 | 杭州电子科技大学 | A kind of brand-new distribution and privatization miRNA diseases contact Forecasting Methodology |
| CN107861481A (en) * | 2017-10-31 | 2018-03-30 | 东北大学 | The method for diagnosing faults excavated based on the semi-supervised new class of industrial big data figure |
| CN107918379A (en) * | 2017-11-29 | 2018-04-17 | 东北大学 | Based on the industrial big data incipient fault detection method for scheming semi-supervised cost-sensitive |
| CN108428478A (en) * | 2018-02-27 | 2018-08-21 | 东北师范大学 | The thyroid cancer Risk Forecast Method excavated based on heterogeneous medical data |
| WO2018220600A1 (en) * | 2017-06-02 | 2018-12-06 | University Of Florida Research Foundation, Inc. | Method and apparatus for prediction of complications after surgery |
-
2019
- 2019-01-30 CN CN201910090787.8A patent/CN109801713A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018220600A1 (en) * | 2017-06-02 | 2018-12-06 | University Of Florida Research Foundation, Inc. | Method and apparatus for prediction of complications after surgery |
| CN107658029A (en) * | 2017-09-29 | 2018-02-02 | 杭州电子科技大学 | A kind of brand-new distribution and privatization miRNA diseases contact Forecasting Methodology |
| CN107861481A (en) * | 2017-10-31 | 2018-03-30 | 东北大学 | The method for diagnosing faults excavated based on the semi-supervised new class of industrial big data figure |
| CN107918379A (en) * | 2017-11-29 | 2018-04-17 | 东北大学 | Based on the industrial big data incipient fault detection method for scheming semi-supervised cost-sensitive |
| CN108428478A (en) * | 2018-02-27 | 2018-08-21 | 东北师范大学 | The thyroid cancer Risk Forecast Method excavated based on heterogeneous medical data |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110298669B (en) * | 2019-06-27 | 2023-09-12 | 上海麦克风文化传媒有限公司 | iOS charging risk control system based on graphic structure |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Fong et al. | Composite Monte Carlo decision making under high uncertainty of novel coronavirus epidemic using hybridized deep learning and fuzzy rule induction | |
| CN104572583B (en) | Method and system for data densification | |
| CN109636061A (en) | Training method, device, equipment and the storage medium of medical insurance Fraud Prediction network | |
| US20130253942A1 (en) | Methods and Apparatus for Smart Healthcare Decision Analytics and Support | |
| Alipour-Vaezi et al. | Prioritizing and queueing the emergency departments’ patients using a novel data-driven decision-making methodology, a real case study | |
| CN112037925B (en) | LSTM algorithm-based early warning method for new major infectious diseases | |
| CN101334770A (en) | Evidence-based medical data base establishment method and related evidence-based medical disease diagnostic method | |
| Shenas et al. | Identifying high-cost patients using data mining techniques and a small set of non-trivial attributes | |
| CN107194508A (en) | A kind of method based on time series dynamic prediction middle or short term region Pediatric Clinic person-time | |
| Liu et al. | Hybrid intelligent algorithm and its application in geological hazard risk assessment | |
| JP2022530868A (en) | Target object attribute prediction method based on machine learning, related equipment and computer programs | |
| CN112580902B (en) | Object data processing method and device, computer equipment and storage medium | |
| CN107292103A (en) | A kind of prognostic chart picture generation method and device | |
| Nagaraj et al. | University Recommender System based on Student Profile using Feature Weighted Algorithm and KNN | |
| Rabbi et al. | An Approximation For Monitoring The Efficiency Of Cooperative Across Diverse Network Aspects | |
| Hijry et al. | Predicting patient waiting time in the queue system using deep learning algorithms in the emergency room | |
| Markazi-Moghaddam et al. | Evaluating patient flow in the operating theater: An exploratory data analysis of length of stay components | |
| CN109192312A (en) | A kind of heart failure patient adverse events intelligent management system and method | |
| CN116883768A (en) | Lung nodule intelligent grading method and system based on multi-modal feature fusion | |
| CN109801713A (en) | A kind of health risk prediction technique based on schematic models | |
| CN113616209B (en) | Method for screening schizophrenic patients based on space-time attention mechanism | |
| CN117012380A (en) | Event estimation method, device, equipment, storage medium and program product | |
| Johnson et al. | District-level estimates of institutional births in Ghana: Application of small area estimation technique using census and DHS data | |
| CN114783586A (en) | Intelligent medical management method, system, terminal and medium based on 5G technology | |
| Colbaugh et al. | Analyzing social media content for security informatics |
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: 20190524 |