CN105574351A - Medical data processing method - Google Patents

Medical data processing method Download PDF

Info

Publication number
CN105574351A
CN105574351A CN201511029760.6A CN201511029760A CN105574351A CN 105574351 A CN105574351 A CN 105574351A CN 201511029760 A CN201511029760 A CN 201511029760A CN 105574351 A CN105574351 A CN 105574351A
Authority
CN
China
Prior art keywords
patient
class
treatment
data
matrix
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
Application number
CN201511029760.6A
Other languages
Chinese (zh)
Other versions
CN105574351B (en
Inventor
黄亦谦
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.)
Beijing Kilo-Ampere Wise Man Information Technology Co Ltd
Original Assignee
Beijing Kilo-Ampere Wise Man 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 Beijing Kilo-Ampere Wise Man Information Technology Co Ltd filed Critical Beijing Kilo-Ampere Wise Man Information Technology Co Ltd
Priority to CN201511029760.6A priority Critical patent/CN105574351B/en
Publication of CN105574351A publication Critical patent/CN105574351A/en
Application granted granted Critical
Publication of CN105574351B publication Critical patent/CN105574351B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G16H50/70—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

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • Biomedical Technology (AREA)
  • Databases & Information Systems (AREA)
  • Pathology (AREA)
  • Epidemiology (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Medical Treatment And Welfare Office Work (AREA)

Abstract

The invention discloses a medical data processing method. The medical data processing method comprises the following steps: S1: dividing medical data in regard to a plurality of patients into patient feature data and treatment data, and respectively converting the feature data and the treatment data into a matrix through normalization; S2: finding out the patients with the similar treatment manner from the treatment data by utilizing hierarchical clustering; S3: obtaining a common treatment manner aiming at each type of patients; S4: obtaining common features aiming at each type of patients, and relating the treatment manner and the features corresponding to each type of patients aiming at each type of patients, so as to obtain the corresponding relation of the treatment manner and the features of the patients.

Description

Medical data disposal route
Technical field
The application relates to medical data mining field, relates more specifically to the relation excavation method of disease treatment mode and patient characteristics.
Background technology
In the process of find in disease, treating, doctor diagnoses accordingly according to the different characteristic of patient.Therefore, find that the relation between patient characteristics and therapeutic modality can select suitable medicine and therapeutic modality to have directive function to doctor.Patient information and therapeutic modality are carried out simple test of hypothesis by existing correlation analysis method usually successively, and lack the implementation of robotization.Therefore, to expect automatically to obtain medical treatment in batch patient the method for relation of mode and patient characteristics.
Summary of the invention
For solving the above-mentioned problems in the prior art, the embodiment of the application proposes a kind of medical data disposal route, comprise step S1: the medical data about multiple patient is divided into patient characteristics data and treatment data, and convert this characteristic and treatment data to matrix respectively by normalization; Step S2: utilize hierarchical clustering to find out the patient with similar therapeutic mode from treatment data; Step S3: obtain common therapeutic modality for each class patient; And step S4: obtain common feature for each class patient, and to each class patient, associate therapeutic modality corresponding to this type of patient and feature, the corresponding relation of the mode that obtains medical treatment and patient characteristics.
Accompanying drawing explanation
Fig. 1 shows the schematic flow diagram of the method for medical data process according to the embodiment of the present invention.
Embodiment
Below in conjunction with accompanying drawing, embodiments of the present invention are described in detail.
Fig. 1 shows the schematic flow diagram of medical data disposal route according to the embodiment of the present invention.With reference to figure 1, in embodiments of the present invention, provide medical data disposal route, the method can comprise:
Step S1: the medical data about multiple patient collected is divided into patient characteristics data and treatment data, and convert this characteristic and treatment data to matrix respectively by normalization.In this application, described medical data mainly can be divided into two classes.The first kind is patient characteristics data, such as, before patient basis, treatment vital sign etc. before Biochemical Information, treatment before routine urinalysis information, treatment before clinical test information, treatment.Equations of The Second Kind is treatment data, such as medication information, therapeutic modality etc.Such as, feature " sex " can transfer-1,1 to according to man, female to data normalization; Feature " positive " can transfer 0,1 to according to value; Medication information can transfer 1,0 etc. according to whether to this medicine.
The input of this step for the patient data collected, can comprise patient characteristics data and treatment data.Output can be patient characteristics matrix and treatment matrix, and patient characteristics matrix behavioural characteristic, is classified as patient, is worth for the value of raw readings after transforming.Treatment matrix behaviour therapy information, is classified as patient, is worth for the value of raw readings after transforming.
In embodiments of the present invention, step S1 can comprise:
Step S1-1: medical data is divided into patient characteristics data and treatment data.Patient characteristics data can be patient information before treatment, such as, before patient basis, treatment vital sign etc. before Biochemical Information, treatment before routine urinalysis information, treatment before clinical test information, treatment.Treatment data can be such as medication information, therapeutic modality etc.
Step S1-2: by patient characteristics Data Discretization, quantize.Patient characteristics data can be divided into discrete type, continuous type.For the factor of discrete type value, can be such as discrete value 1,2 by numbers translate ...And for the factor of continuous type value, retain numerical value.
Step S1-3: by discretize and the patient characteristics data normalization after quantizing.By (1) normalization according to the following formula of patient characteristics data, thus obtain patient characteristics matrix X.In formula (1), x is patient characteristics data, x irepresent the characteristic of i-th patient, x maxrepresent the maximal value of patient characteristics data, x minrepresent the minimum value of patient characteristics data.
x i = x i - x m i n x max - x m i n - 1 Formula (1)
This patient characteristics matrix X is such as shown below, and behavial factor, is classified as patient, and each value is a numerical value.In this formula, f representation feature, total n feature, p patient.X ijrepresent the value of i-th feature of a jth patient.
X = f 1 f 2 M f n x 11 x 12 L x 1 p x 21 x 22 L x 2 p M M O M x n 1 x n 2 L x n p
Step S1-4: by treatment Data Discretization, quantize.Treatment data can be divided into discrete type, continuous type.For the factor of discrete type value, can be such as discrete value 1,2 by numbers translate ...For the factor of continuous type value, retain numerical value.
Step S1-5: by discretize and the treatment data normalization after quantizing.Data (3) normalization according to the following formula will be treated, thus obtain Case treatment matrix Y.In formula (3), y is treatment data, y irepresent the treatment data of i-th patient, y maxrepresent the maximal value of Case treatment data, y minrepresent the minimum value of Case treatment data.
y i = y i - y m i n y max - y m i n - 1 Formula (3)
Case treatment matrix Y is such as shown below, and behaviour therapy information, is classified as patient, and each value is a numerical value.In formula, t represents treatment information, total m feature, p patient.Y ijrepresent the value of i-th the treatment information of a jth patient.
Y = t 1 t 2 M t m y 11 y 12 L y 1 p y 21 y 22 L y 2 p M M O M y m 1 y m 2 L y m p
Step S2: therapeutic modality cluster: utilize hierarchical clustering to find out the patient with similar therapeutic mode from treatment data.The input of this step can be Case treatment matrix Y, and output can be patient classification.Here therapeutic modality cluster can refer to and utilize hierarchical clustering to find out the patient with similar therapeutic mode from treatment data.The form for the treatment of data can for suffering from the disease the treatment information of people and correspondence thereof.For each patient, these treatment information can form vector.Hierarchical clustering is carried out to treatment information dimension, selects classification N, patient is divided into N number of classification.Each class patient has similar therapeutic modality.
Specifically, step S2 can comprise:
Step S2-1: calculate Case treatment mode vector Euclidean distance between two according to following formula (4).If P i=(y 1i,y 2iy mi) be therapeutic modality vector corresponding to patient i, in formula (4), D ijrepresent the Euclidean distance of vector, y airepresent the value of a the treatment information of i-th patient, y ajrepresent the value of a the treatment information of a jth patient, total m feature, p patient.
D i j = Σ ( y a i - y a j ) 2 , i , j = 1 , 2 , L , p , a = 1 , 2 , L , m Formula (4)
Step S2-2: p patient is divided into p class, namely each class is only containing a patient, according to the following formula (5) compute classes spacing.In formula (5), P irepresent the therapeutic modality vector that i-th patient is corresponding, P jrepresent the corresponding therapeutic modality vector of a jth patient, DC representation class spacing, C represents classification, DC rsrepresent classification C rwith classification C sbetween class distance.
DC r s = min P i ∈ C r , P j ∈ C s D i j Formula (5)
Step S2-3: merge two apart from minimum classification.
Step S2-4: repeat step S2-3, until last patient is merged, forms hierarchical clustering result.
Step S2-5: select classification number N, patient is divided into N class (N is positive integer).
Step S3: obtain way of concurrent therapy: obtain common therapeutic modality for each class patient.The input of this step can be patient classification, treatment matrix Y, and output can be therapeutic modality corresponding to patient classification.This acquisition way of concurrent therapy can refer to finds common therapeutic modality to each class patient.To each class of multiclass (such as N class) patient, using current research class as positive sample, from remaining N-1 class, select the negative sample of equal samples amount randomly.To each therapeutic modality, t test of hypothesis is utilized to obtain significant therapeutic modality.Through this step, obtain the way of concurrent therapy that each class patient has.
Particularly, this step S3 can comprise:
Step S3-1: using the current research class of N class patient as positive sample, selects the negative sample of equal samples amount randomly from remaining N-1 class.
Step S3-2: the row that the patient contained with positive sample packages in treatment matrix Y is associated, deletes all the other row in treatment matrix Y, to form matrix A; The row be associated with the patient that negative sample comprises in treatment matrix Y, delete all the other row in treatment matrix Y, to form matrix B.
Step S3-3: for each therapeutic modality, carries out t test of hypothesis to matrix A and B.Select the way of concurrent therapy of therapeutic modality as current research class patient of statistical significance <0.01.
Step S3-4: to each class of N class patient, repeats step S3-1 to step S3-3, obtains the therapeutic modality that every class patient is corresponding.
Step S4: obtain common feature for each class patient.The input of this step can be patient classification, eigenmatrix X, and output can be patient classification's characteristic of correspondence.Acquisition common trait refers to finds common feature to each class patient.Such as to each class of N class patient, using current research class as positive sample, the negative sample of random equal samples amount of selecting from remaining N-1 class.To each patient characteristics, t test of hypothesis is utilized to obtain significant feature.Through this step, obtain the common trait that each class patient has, in conjunction with the way of concurrent therapy that each class patient has, obtain significant therapeutic modality and patient characteristics corresponding relation.
Particularly, this step S4 can comprise:
Step S4-1: using the current research class of N class patient as positive sample, selects the negative sample of equal samples amount randomly from remaining N-1 class.
Step S4-2: the row that the patient contained with positive sample packages in keeping characteristics matrix X is associated, deletes all the other row in eigenmatrix X, with formed matrix A '; The row be associated with the patient that negative sample comprises in keeping characteristics matrix Y, delete all the other row in eigenmatrix Y, with formed matrix B '.
Step S4-3: to each feature, to matrix A ' and B' carry out t test of hypothesis.Select the common trait of feature as current research class patient of statistical significance <0.01.
Step S4-4: to each class of N class patient, repeats step S4-1 to step S4-3, obtains every class patient characteristic of correspondence.
Step S4-5: to each class of N class patient, associates therapeutic modality corresponding to this type of patient and feature, the corresponding relation of the mode that obtains medical treatment and patient characteristics.
Alternatively, step S4 can comprise step S4-6: adopt therapeutic modality and patient characteristics that Computer display is relevant.
Step S5: adjustment parameter is to obtain relevant therapeutic modality and patient characteristics.Through step S1 to S4, if the therapeutic modality do not satisfied condition and patient characteristics, this step can be adopted obtain.Here parameter adjustment refers to if the therapeutic modality do not satisfied condition and patient characteristics, obtains relevant therapeutic modality and patient characteristics by the double counting of adjustment classification number.
Particularly, step S5 can comprise:
Step S5-1: to certain class in N class patient, if step S3-3 does not obtain way of concurrent therapy corresponding to this type of patient, then can improve statistical significance threshold value, such as, bring up to 0.05.
Step S5-2: to certain class in N class patient, if step S4-3 does not obtain common trait corresponding to this type of patient, then can improve statistical significance threshold value, such as, bring up to 0.05.
Step S5-3: the value reducing or raise classification number in step S2-5, adjusts the class number of patient, repeats step S3 and/or S4, different rent fine granularities obtains the therapeutic modality of being correlated with and patient characteristics.
The t test of hypothesis mentioned in above-mentioned steps S3 and step S4 can be the method for inspection known in affiliated field, hereafter simply introduces the basic thought of t test of hypothesis.
If overall X and Y is independent, σ 1represent the standard deviation of X, μ 1represent the average of X, S 1represent the sample variance of X, represent the sample average of X, σ 2represent the standard deviation of Y, μ 2represent the average of X, S 2represent the sample variance of Y, represent the sample average of Y, p represents statistical significance.If the variance of two normal populations is equal, namely consider two-sided hypothesis test problem:
H 0:μ 1＝μ 2 H 1:μ 1≠μ 2
At this moment at μ 1=μ 2under can obtain:
T = ( X &OverBar; - Y &OverBar; ) - ( &mu; 1 - &mu; 2 ) ( 1 n 1 + 1 n 2 ) S 2 : t ( n 1 + n 2 - 2 )
Wherein S 2 = ( n 1 - 1 ) S 1 2 - ( n 2 - 1 ) S 2 2 ( n 1 - 1 ) + ( n 2 - 1 ) .
The region of rejection obtaining above-mentioned Hypothesis Testing Problem is thus
{|T|＞t 1-p/2(n 1+n 2-2)}
It will be appreciated by those skilled in the art that above-mentioned method can be implemented by the mode of the combination of software, hardware and software and hardware.
In the process of find in disease, treating, doctor can diagnose accordingly according to the different characteristic of patient.Find that the relation between patient characteristics and therapeutic modality can instruct doctor to select suitable medicine and therapeutic modality.Embodiments of the present invention compared with prior art have one of at least following advantage.
1. existing correlation analysis method lacks the implementation of robotization usually.A kind of relation excavation method of factor before embodiments of the present invention propose disease treatment mode and treat, automatically can find out the patient with similar therapeutic mode in large-scale patient, and detect the common trait that similar therapeutic mode patient has, thus the relation of obtain medical treatment mode and patient characteristics.
2. patient information and therapeutic modality are carried out test of hypothesis by existing correlation analysis method usually successively, and there is similarity between patient in reality.The step S2 of embodiments of the present invention, from the angle of entirety, utilizes the thought of cluster to find out the patient with similar therapeutic mode from treatment data, patient is divided into different classifications and studies.
3. usually lack finer analysis after existing correlation analysis method classification.The step S3 of embodiments of the present invention obtains common therapeutic modality for each class patient, and step S4 obtains common feature for each class patient.Step S3 and S4, on the basis of overall angle, has carried out finer analysis to each class.
4. usually to perform step comparatively single for existing correlation analysis method, and what-if result and expection are not inconsistent, and lacks feedback procedure.Step S5 in embodiments of the present invention proposes the feedback procedure of regulating parameter, obtains the result tallied with the actual situation.
It will be understood by those skilled in the art that the step of the medical data disposal route of above-described embodiments of the present invention is not must perform according to the order of shown step.As long as embodiments of the present invention can be implemented, step can be performed by other orders, or at least some step synchronously can be carried out.Such as, any one step of step S3 and step S4 can first be performed, or is performed simultaneously.Such as, any one combination in the combination of the combination of step S1-2, S1-3 and step S1-4, S1-5 can first be performed, or is performed simultaneously.
Although the application is the mode by describing embodiment describe the present invention, it will be appreciated by those skilled in the art that these concrete embodiments are schematic and nonrestrictive.Various amendment, distortion and replacement can be carried out to embodiment those skilled in the art are by understanding the design of the application at the embodiment reading the application.

Claims (9)

1. a medical data disposal route, the method comprises:
Step S1: the medical data about multiple patient is divided into patient characteristics data and treatment data, and convert this characteristic and treatment data to matrix respectively by normalization;
Step S2: utilize hierarchical clustering to find out the patient with similar therapeutic mode from treatment data;
Step S3: obtain common therapeutic modality for each class patient; And
Step S4: obtain common feature for each class patient, and to each class patient, associate therapeutic modality corresponding to this type of patient and feature, the corresponding relation of the mode that obtains medical treatment and patient characteristics.
2. method according to claim 1, wherein, described step S1 comprises:
Step S1-1: medical data is divided into patient characteristics data and treatment data;
Step S1-2: by patient characteristics Data Discretization, quantize;
Step S1-3: by discretize and the patient characteristics data after quantizing according to formula (1) normalization, thus obtain patient characteristics matrix X,
x i = x i - x m i n x max - x m i n - 1 Formula (1)
Wherein, x irepresent the characteristic of i-th patient, x maxrepresent the maximal value of patient characteristics data, x minrepresent the minimum value of patient characteristics data;
Step S1-4: by treatment Data Discretization, quantize;
Step S1-5: by discretize and the treatment data after quantizing according to formula (3) normalization, thus obtain Case treatment matrix Y,
y i = y i - y m i n y max - y m i n - 1 Formula (3)
Wherein, y irepresent the treatment data of i-th patient, y maxrepresent the maximal value of Case treatment data, y minrepresent the minimum value of Case treatment data.
3. method according to claim 1, wherein, described step S2 comprises:
Step S2-1: calculate Case treatment mode vector Euclidean distance between two according to formula (4);
If P i=(y 1i, y 2iy mi) be therapeutic modality vector corresponding to patient i, in formula (4), D ijrepresent the Euclidean distance of vector, y airepresent the value of a the treatment information of i-th patient, y ajrepresent the value of a the treatment information of a jth patient,
D i j = &Sigma; ( y a i - y a j ) 2 , i , j = 1 , 2 , L , p , a = 1 , 2 , L , m Formula (4)
Step S2-2: p patient is divided into p class, namely each class is only containing a patient, according to the following formula (5) compute classes spacing, and in formula (5), DC representation class spacing, C represents classification, DC rsrepresent classification C rwith classification C sbetween class distance,
DC r s = min P i &Element; C r , P j &Element; C s D i j - - - ( 5 )
Step S2-3: merge two apart from minimum classification;
Step S2-4: repeat step S2-3, until last patient is merged, to form hierarchical clustering result;
Step S2-5: select classification number N, patient is divided into N class.
4. method according to claim 1, wherein, described step S3 comprises:
Step S3-1: using the current research class of N class patient as positive sample, selects the negative sample of equal samples amount randomly from remaining N-1 class;
Step S3-2: the row that the patient contained with positive sample packages in treatment matrix Y is associated, deletes all the other row in treatment matrix Y, to form matrix A; The row be associated with the patient that negative sample comprises in treatment matrix Y, delete all the other row in treatment matrix Y, to form matrix B;
Step S3-3: for each therapeutic modality, carries out t test of hypothesis to matrix A and B, selects the way of concurrent therapy of therapeutic modality as current research class patient of statistical significance <0.01;
Step S3-4: to each class of N class patient, repeating said steps S3-1, to described step S3-3, obtain the therapeutic modality that every class patient is corresponding.
5. method according to claim 1, wherein, described step S4 comprises:
Step S4-1: using the current research class of N class patient as positive sample, selects the negative sample of equal samples amount randomly from remaining N-1 class;
Step S4-2: the row that the patient contained with positive sample packages in keeping characteristics matrix X is associated, deletes all the other row in eigenmatrix X, with formed matrix A '; The row be associated with the patient that negative sample comprises in keeping characteristics matrix Y, delete all the other row in eigenmatrix Y, with formed matrix B ';
Step S4-3: to each feature, to matrix A ' and B' carry out t test of hypothesis.Select the common trait of feature as current research class patient of statistical significance <0.01;
Step S4-4: to each class of N class patient, repeating said steps S4-1, to step S4-3, obtain every class patient characteristic of correspondence;
Step S4-5: to each class of N class patient, associates therapeutic modality corresponding to this type of patient and feature, the corresponding relation of the mode that obtains medical treatment and patient characteristics.
6. the method according to claim 4 or 5, the method also comprises:
Step S5: adjustment parameter is to obtain relevant therapeutic modality and patient characteristics.
7. method according to claim 6, wherein, described step S5 comprises:
To the class in N class patient, if described step S3-3 does not obtain way of concurrent therapy corresponding to this type of patient, then improve statistical significance threshold value, then repeating said steps S3.
8. method according to claim 6, wherein, described step S5 comprises:
To the class in N class patient, if described step S4-3 does not obtain common trait corresponding to this type of patient, then statistical significance threshold value can be improved, then repeating said steps S4.
9. method according to claim 6, wherein, described step S5 comprises:
Reduce in described step S2-5 or raise class number, repeating said steps S3 and/or described step S4.
CN201511029760.6A 2015-12-31 2015-12-31 Medical data processing method Active CN105574351B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201511029760.6A CN105574351B (en) 2015-12-31 2015-12-31 Medical data processing method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201511029760.6A CN105574351B (en) 2015-12-31 2015-12-31 Medical data processing method

Publications (2)

Publication Number Publication Date
CN105574351A true CN105574351A (en) 2016-05-11
CN105574351B CN105574351B (en) 2017-02-15

Family

ID=55884476

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201511029760.6A Active CN105574351B (en) 2015-12-31 2015-12-31 Medical data processing method

Country Status (1)

Country Link
CN (1) CN105574351B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106650299A (en) * 2017-01-18 2017-05-10 浙江大学 Quick calculating method for patient similarity analysis
CN106951710A (en) * 2017-03-22 2017-07-14 华东师范大学 CAP data systems and method based on privilege information Learning support vector machine
CN108109700A (en) * 2017-12-19 2018-06-01 中国科学院深圳先进技术研究院 A kind of chronic disease Drug efficacy evaluation method and apparatus
CN108346474A (en) * 2018-03-14 2018-07-31 湖南省蓝蜻蜓网络科技有限公司 The electronic health record feature selection approach of distribution within class and distribution between class based on word

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1704961A (en) * 2004-05-26 2005-12-07 冲博子 Analysis method for diagnosis and treatment behavior and administration
CN101334843A (en) * 2007-06-29 2008-12-31 中国科学院自动化研究所 Pattern recognition characteristic extraction method and apparatus
CN103745227A (en) * 2013-12-31 2014-04-23 沈阳航空航天大学 Method for identifying benign and malignant lung nodules based on multi-dimensional information
CN104915560A (en) * 2015-06-11 2015-09-16 万达信息股份有限公司 Method for disease diagnosis and treatment scheme based on generalized neural network clustering
CN104951649A (en) * 2015-05-27 2015-09-30 华南农业大学 HBV classifying method based on Gaussian blur integrals

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1704961A (en) * 2004-05-26 2005-12-07 冲博子 Analysis method for diagnosis and treatment behavior and administration
CN101334843A (en) * 2007-06-29 2008-12-31 中国科学院自动化研究所 Pattern recognition characteristic extraction method and apparatus
CN103745227A (en) * 2013-12-31 2014-04-23 沈阳航空航天大学 Method for identifying benign and malignant lung nodules based on multi-dimensional information
CN104951649A (en) * 2015-05-27 2015-09-30 华南农业大学 HBV classifying method based on Gaussian blur integrals
CN104915560A (en) * 2015-06-11 2015-09-16 万达信息股份有限公司 Method for disease diagnosis and treatment scheme based on generalized neural network clustering

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106650299A (en) * 2017-01-18 2017-05-10 浙江大学 Quick calculating method for patient similarity analysis
CN106650299B (en) * 2017-01-18 2019-01-25 浙江大学 A kind of quick calculation method of patient's similarity analysis
CN106951710A (en) * 2017-03-22 2017-07-14 华东师范大学 CAP data systems and method based on privilege information Learning support vector machine
CN108109700A (en) * 2017-12-19 2018-06-01 中国科学院深圳先进技术研究院 A kind of chronic disease Drug efficacy evaluation method and apparatus
CN108109700B (en) * 2017-12-19 2021-05-25 中国科学院深圳先进技术研究院 Method and device for evaluating curative effect of chronic disease
CN108346474A (en) * 2018-03-14 2018-07-31 湖南省蓝蜻蜓网络科技有限公司 The electronic health record feature selection approach of distribution within class and distribution between class based on word

Also Published As

Publication number Publication date
CN105574351B (en) 2017-02-15

Similar Documents

Publication Publication Date Title
Nguena Nguefack et al. Trajectory modelling techniques useful to epidemiological research: a comparative narrative review of approaches
US20220375611A1 (en) Determination of health sciences recommendations
Perros et al. SPARTan: Scalable PARAFAC2 for large & sparse data
WO2017152802A1 (en) Intelligent system and method for converting textual medical report into structured data
Xiong et al. Online single-cell data integration through projecting heterogeneous datasets into a common cell-embedding space
Gupta et al. Performance analysis of various data mining classification techniques on healthcare data
CN105574351A (en) Medical data processing method
Shashikumar et al. Multiscale network representation of physiological time series for early prediction of sepsis
Yin et al. Identifying sepsis subphenotypes via time-aware multi-modal auto-encoder
Siuly et al. An advanced analysis system for identifying alcoholic brain state through EEG signals
Li et al. Multi-modal multi-instance learning using weakly correlated histopathological images and tabular clinical information
Poon et al. A novel approach in discovering significant interactions from TCM patient prescription data
CN106066936A (en) A kind of method and system improving adverse effect precision of prediction
Sun et al. Multitask balanced and recalibrated network for medical code prediction
CN109815478A (en) Medicine entity recognition method and system based on convolutional neural networks
Islam et al. Cartography of genomic interactions enables deep analysis of single-cell expression data
Seiler et al. Uncertainty quantification in multivariate mixed models for mass cytometry data
Cao et al. Identifying hiv-induced subgraph patterns in brain networks with side information
Kaur et al. Analysis of big data cloud computing environment on healthcare organizations by implementing Hadoop clusters
Dadok et al. A probabilistic framework for a physiological representation of dynamically evolving sleep state
CN109859813A (en) A kind of entity modification word recognition method and device
Jeong Parallel processing scheme for minimizing computational and communication cost of bioinformatics data
Qiao et al. A personalized low-rank subspace clustering method based on locality and similarity constraints for scRNA-seq data analysis
Yu et al. GSEnet: feature extraction of gene expression data and its application to Leukemia classification
Miller et al. Detection of relationships among multi-modal brain imaging meta-features via information flow

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant