CN103970973A - Real-data-based simulative RCT (randomized controlled trial) analysis method - Google Patents
Real-data-based simulative RCT (randomized controlled trial) analysis method Download PDFInfo
- Publication number
- CN103970973A CN103970973A CN201310038845.5A CN201310038845A CN103970973A CN 103970973 A CN103970973 A CN 103970973A CN 201310038845 A CN201310038845 A CN 201310038845A CN 103970973 A CN103970973 A CN 103970973A
- Authority
- CN
- China
- Prior art keywords
- odds
- database
- intervention
- power
- analysis
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention belongs to the field of RCT (randomized controlled trial), and relates to a real-data-based simulative RCT analysis method. The analysis method is a data analysis method for comparative effectiveness researches, is used for analyzing result data generated by different interference, and comprises the steps of database creation, randomized grouping, statistical analysis, result obtaining and the like. The analysis method is used for performing comparative effectiveness analysis on clinically generated real data, confounding factors can be effectively controlled, an analysis result is high in extensionality, requirements on the sample size are low, and a plurality of result indexes can be simultaneously analyzed; the analysis method can be used for comparing the effectiveness and security of different clinical intervention methods or strategies, and new evidences for evidence-based medicine can be provided for clinical decision making.
Description
Technical field
The invention belongs to test data analyzer method field, relate to a kind of simulation RCT analytical approach based on True Data; The method is analyzed for the final result data (actual effect data) of randomized controlled trial comparative studies (Comparative Effectiveness Research, CER), for clinical decision provides circumstantial evidence.
Background technology
Evidence-based medicine EBM is to take evidence as basic medical science, and it is intended to utilize existing best evidence to help clinical practice, solves clinical problem.Classification according to GRADE in 2011 to evidence quality, based on high-quality randomization control test research (Randomized controlled trials, RCTs) RCT that meta analysis and system evaluation and strict design are implemented is considered to be the high-quality evidence of curative effect comparison, is the optimal path that obtains evidence; And the CER that utilizes actual effect data is the new field of the actual effect data analysis that utilizes clinical generation that produces in recent years, but the data analysing method of above-mentioned research is all based on classical epidemiology and statistical basic skills.Because RCT expense, ethics, research object are included the restriction of the factor such as comparatively strict in, the research quantity of RCT is few, actual effect is limited, extrapolation is limited, and this kind of evidence quantity is also limited.In addition, designing rigorous is that CER obtains one of evidence approach with implementing strict observational study, but because randomized grouping is not carried out in observational study, makes relatively becoming of different therapeutic modality results for the treatment of very complicated.In order to get rid of the impact of Confounding Factor, classic method is the effect of Confounding Factor of controlling by comparatively complicated multiplicity method, but relies on the observational study of limited quantity, judges that accurately the difference of curative effect of different therapeutic modalities is still very difficult.In addition, Rosenbaum and Rubin also solve unbalanced problem between the Confounding Factor group in observational study to a certain extent in the method for the Propensity Score (propensity score, PS) of nineteen eighty-three proposition; Yet due to the limitation of observational study self, its result of study is defined as inferior quality evidence by GRADE.Therefore, in clinical medicine practise, still have at present the decision problem of very many diagnosis and treatment to be badly in need of evidence-based medical, particularly, in recent years along with the appearance of a large amount of Novel medicines, new Clinics, make the RCTs quantity of carrying out at present far can not meet clinical demand.Conventionally before RCTs implements, clinical practice has produced the final result data of a large amount of diagnosis and treatment, how scientific and effective its patient is non-selective, utilize above-mentioned actual effect data, and generation can be to wait at present the problem that solves for the new evidence-based medical of clinical decision reference.
Summary of the invention
The object of the invention is to overcome defect and the deficiency of prior art, a kind of simulation RCT analytic approach based on True Data is provided; This analytic approach is for intervention effect comparative studies (Comparative Effectiveness Research, CER) data analysis, especially to accepting the final result data (actual effect data) of different dry prescheme generation, analyze, can be used for the comparison of the tactful effect & Safety of different dry prescheme clinically or implementation, for clinical decision provides new evidence-based medical.
Method of the present invention, related data (comprising essential characteristic data and final result data) for the object of accepting intervention to be measured and other interventions producing clinically, through simulation randomized grouping, the scheme set analysis strategy (Per-protocol Analysis strategy/PP strategy) that meets according to RCT after each grouping carries out intervention effect comparison, obtains the ratio (odds value) of the number percent of discrepant experiment frequency and the experiment frequency number percent of indifference and 95%CI thereof as the discrepant foundation of tool whether of effect and security between judgement implementation different dry prescheme or mode; When the upper limit of odds < 1 or 95%CI is less than 1, can think the fruit indifference of two kinds of intervention schemes or mode; When the lower limit of odds > 1 or 95%CI is greater than 1, can think that the effect of two kinds of intervention schemes or mode is variant; Odds=1 or 95%CI comprise 1, wouldn't provide clear and definite conclusion.In embodiments of the invention, carry out simulation randomized grouping 100 times, after each grouping, carry out intervention effect comparative analysis.
In the present invention, by 4 step explanation this analysis ratio juris, feasibility, stability and the result of determination of proof this analysis method and the consistance of truth; The final result variable and the difference of Confounding Factor between group that due to True Data, reflect intervention effect have no way of finding out about it, cannot judge effect quantity (Effect Size between the odds of this analysis method calculating and the distribution of 95%CI and real group, E/S) and the relation of power of a test (power) size, also cannot judge and use this analysis method judgement result of Different therapeutical effect and the consistance of truth, therefore, the present invention adopts the method for simulated data to verify;
Particularly, the simulation RCT analytical approach based on True Data of the present invention, is characterized in that, it comprises step:
(1) first, the variant database of effect that contains two kinds of means of intervention and final result variable (continuous variable is example) that utilizes computer simulation clinical intervention to produce: the body-mass index (BMI, kg/m2) of usining carries out Different therapeutical effect comparison as the final result variable of accepting after two kinds of different means of intervention; Experimental group in database (n1) and control group (n2) sample size are 50, and experimental group mean is 27.25kg/m2, and standard deviation is 0.85kg/m2, and control group mean is 30.26kg/m2, and standard deviation is 0.85kg/m2; Set up hypothesis: H
0for final result variable mean does not have difference between two groups; H
1, the mean of final result variable variant between two groups (two-sided test); With the random series that is numbered of research object, carry out randomized grouping again, then within error-free minute, enter the research object BMI of analysis according to PP strategy, carry out comparing between two groups; Repeat the comparison of 100 randomized groupings and BMI; Finally by calculating, in 100 BMI comparative analysis results, refuse null hypothesis experiment frequency and the ratio (odds) of not refusing null hypothesis experiment frequency, repeat this process and obtain odds=2.03 100 times, 95%CI is (1.20,3.09), the lower limit of odds > 1 and 95%CI is greater than 1, shows that the effect of these two kinds of means of intervention is variant; Result shows, the odds value obtaining and 95%CI thereof can reflect the real difference of intervention effect;
(2) whether the distribution of observation odds and 95%CI distributes consistent with E/S size and Power; According to power from 0.5 to 0.85, the simulation of the difference of corresponding E/S creates 8 of the simulated databases of 8 of the variant databases of intervention effect and intervention effect indifference and analyzes respectively with it, calculates odds and 95%CI thereof; Result demonstration odds and 95%CI thereof present good linear relationship with the variation of power, and when the discrepant database of intervention effect is analyzed, power is larger, and the odds and the 95%CI that calculate are larger; When the database of intervention effect indifference is analyzed, power is larger, and the odds and the 95%CI that calculate are less; No matter whether intervention effect there are differences, and when power is 0.5, the odds calculating approaches 1, and its 95%CI all comprises 1;
(3) between the distribution of observing Odds and 95%CI thereof and group, whether the pass of real difference and power ties up under different sample sizes (n=50,100,500 and 1000) stable; According under different sample sizes, the difference simulation of E/S and power creates a plurality of simulated databases and analyzes respectively, the odds that under different sample sizes, the variant database analysis of intervention effect obtains is all greater than to 1, odds value and 95%CI all increases with the increase of power; The odds value that under different sample sizes, intervention effect indifference database analysis obtains is all less than to 1, and odds value and 95%CI thereof all reduce with the increase of power; Result shows, adopts the Odds that this analysis methods analyst obtains and 95%CI carries out the result that between two groups, intervention effect difference is judged and truth consistance is higher and stability is better;
(4) prove the in the situation that of small sample (n=100), no matter be that two groups of sample sizes equate (n1=50, n2=50) database or sample size database (n1=30, n2=70) not etc., the harmony of Confounding Factor between two groups is all better; First database is carried out to randomized grouping 100 times, again to removing wrong minute sample according to PP strategy, then within error-free minute, entering the research object Confounding Factor of analysis, between two groups, whether there is difference and carry out test of hypothesis, result shows, the possibility of Confounding Factor indifference between group all surpasses 95% (possibility that is respectively sample size Confounding Factor Homogeneity between groups when equal is 97%, the possibility of sample size Confounding Factor equilibrium not etc. time be 99%).
Simulation RCT analytical approach based on True Data of the present invention is compared with the analytical approach of the high-level CER research of providing of prior art evidence-based medical, has the following advantages:
1) True Data that this method produces clinical intervention carries out intervention effect comparative analysis, but not to data or the data analysis of delivering;
2) effectively control Confounding Factor: although observational study can be by using the Confounding Factor between the balanced group of PS method to make the data of observational study reach the effect of " approaching randomization data ", but PS method can only be for observable variable, and the bias that potential unknown Confounding Factor is caused is helpless; Only have when all Confounding Factor are all Observable, PS method can obtain Treatment Effects without inclined to one side estimation; And this method is utilized RCT principle, existing non-data is clinically simulated to randomized grouping, utilize unknown and known Confounding Factor between the balanced group of method of randomized grouping, the impact of the Confounding Factor that may exist while getting rid of different interventions after each randomized grouping and effect comparison thereof; The calculating of statistic Odds and 95%CI thereof is that two groups of intervention effects based on after randomized grouping relatively carry out, the good control that the impact of Confounding Factor is obtained; In single RCT research and design and implementation process, the factors such as hiding and blind method imperfection due to Random assignment scheme, cause result of study can accurately not reflect the intervention effect of means of intervention to be studied, the principle that this method adopts approximate Meta to analyze, 100 intervention effect diversity ratios of integrated use, compared with the result of test of hypothesis, are usingd odds value and 95%CI thereof as assessing the foundation that between different means of intervention, whether intervention effect there are differences;
3) result has good epitaxy: in RCT research and design process, the selection of research object is had to the strict exclusion standard of including in; Therefore its representativeness is relatively poor, can not reflect the overall picture of disease, causes the epitaxy of result of study poor; Meta based on RCT analyzes, and its analysis result is also included in the impact of a plurality of independent RCT research epitaxy of research; And this method is to analyze based on existing non-data clinically, research object is all nonselective, and result of study has good epitaxy;
4) sample size requirements is low: because RCT studies the restriction that is subject to two class statistics mistakes, need to have enough sample sizes to reduce the possibility that I class mistake and II class mistake are made in test of hypothesis; As shown in Figure 1 and Figure 2, the trend that odds value under different sample sizes and 95%CI thereof change with power is in the method consistent, the odds value discrepant database analysis of intervention effect being obtained is all greater than 1, odds value and 95%CI all increases and constantly increases along with power is continuous; The odds value that intervention effect indifference database analysis is obtained is all less than 1, odds value and 95%CI all constantly increases with power and constantly reduces; Result shows, in the situation that sample size is less (minimum n=50) also available this analysis method whether the intervention effect of different means of intervention be there are differences and is analyzed;
5) can analyze a plurality of final result indexs: the main and Minor consequence index that need to clearly define research in the RCT research and design stage simultaneously, the calculating of sample size is also carried out based on main final result index, therefore, research conclusion can only show the intervention effect of means of intervention to be studied to certain specific final result index; And this method can may be studied by the final result index that intervention to be studied affects on a plurality of, thereby synthetically evaluate the intervention effect of means of intervention to be studied.
Accompanying drawing explanation
Fig. 1 has shown the trend that odds value that intervention effect indifference database analysis is obtained and 95%CI thereof change with power, and result shows, in different sample sizes, the odds value of intervention effect indifference database and 95%CI thereof are all along with power increase and reduce.
Fig. 2 has shown the trend that odds value that the variant database analysis of intervention effect is obtained and 95%CI thereof change with power, wherein, n represents sample size size, and power represents power of a test, Lower represents the lower limit of odds value 95%CI, and Upper represents the upper limit of odds value 95%CI;
Result demonstration, in different sample sizes, the odds value of the variant database of intervention effect and 95%CI thereof are all along with power increases and increases.
Embodiment
Embodiment 1
Utilize the programming tool of Statall.0 statistical analysis software specifically to implement, application Statall.0 software programming forms complete program, adopts simulated data to analyze accordingly.During practical application, do not comprise the process that creates simulated database.This analysis method can directly be carried out for different true clinical datas the analyses such as balance between the two groups analysis of repetition randomized grouping, Odds and 95%CI calculating, Confounding Factor, finally provide Odds and 95%CI, and the result of Confounding Factor balance between the two groups comparison (unbalanced probability between Confounding Factor group) is as the foundation of the different intervention effects of judgement; Concrete steps are as follows:
1, creation database
Utilize PS (power and sample size program) software to calculate under different sample sizes (n=50,100,500, and 1000), the size (table 1) of the corresponding E/S of Power.According to the difference of power, E/S under different sample sizes, utilize computer simulation to produce the database that comprises Confounding Factor variable, final result variable and different means of intervention, under different sample sizes, all contain 16 databases, each 8 of the databases that comprises the variant and intervention effect indifference of intervention effect; The height (cm) of take is example as the Confounding Factor that may exist in experimentation, evaluates Confounding Factor whether balanced comparable between group; The body-mass index (BMI, kg/m2) of usining carries out intervention effect diversity ratio as the final result variable of accepting after two kinds of different means of intervention; Take below sample size as 100, power is 0.5 o'clock, the variant database of intervention effect be created as routine presentation database constructive process, the establishment of other databases can create according to this process, only need to revise corresponding with it correlation parameter during establishment; In actual application, skip this step, from randomized grouping.
The feature of table 1 simulated database: effect value and test of hypothesis power of a test between sample size, group
Tab1Characteristics of simulation datasets：sample size、effect size and power
Notes Significant：The effectiveness of two different treatments have significant differences；Nosignificant：The effectiveness of two different treatments have no significant differences。
(1) create BMI data
Set sample size and be 100, mean and standard deviation be respectively 27.5kg/m
2and 6kg/m
2, utilize invnorm (uniform ()) * σ+μ order to create and meet the control group BMI data of normal distribution and calculate standard deviation; Obtaining on the basis of control group BMI data and standard deviation (sd) thereof, according to sample size in table 1 be 100, power is 0.5 o'clock, the E/S size that the variant database of curative effect is corresponding, creates the experimental group BMI data that meet normal distribution;
(2) produce height data
Set sample size and be 100, mean and standard deviation be 174.5cm and 4cm respectively, utilize invnorm (uniform ()) * σ+μ order to create respectively and meet the control group of normal distribution and the height data of experimental group; Though control group and experimental group height data all adopt identical parameter to create, but due to unassigned species subnumber, the experimental group therefore producing and the height data of control group unequal;
(3) generating database
Utilize the orders such as stack, merge, merge height and BMI data formation database;
2, randomized grouping
The database, the recording mechanism of usining that adopt simple randomization packet mode to produce simulation are carried out randomized grouping as random sequence number;
3, statistical analysis
Data analysis adopts Statal1.0 statistical software, uses PP strategy to retain after randomized grouping with it at the consistent sample of former database Intervention group of living in, removes the sample of wrong minute, and continuous data is with mean ± standard deviation
represent; Between two groups, relatively adopt two independent sample t checks, take α=0.05 as inspection level, P < 0.05 has statistical significance for difference;
(1) compare intervention effect difference
For comparing two kinds of its intervention effects of means of intervention, whether there are differences, first simulated database is carried out to randomized grouping 100 times, and the BMI entering for error-free minute after each randomized grouping in the sample of analysis is carried out to test of hypothesis, then calculate respectively the frequency of refusing null hypothesis in 100 BMI test of hypothesis results and not refusing null hypothesis;
(2) Odds value and 95%CI
According to the result of 100 BMI test of hypothesis, calculate refusal null hypothesis frequency and the ratio (odds value) of not refusing null hypothesis frequency, and the computation process by repetition odds value 100 times, obtain distribution and the 95%CI thereof of odds value;
(3) between comparative group, Confounding Factor is harmonious
While equating for comparing two groups of sample sizes, after randomized grouping, whether Confounding Factor is balanced comparable between group, first two groups of sample sizes are equated and be 50 simulated database to carry out randomized grouping 100 times, and the height entering for error-free minute after each randomized grouping in the sample of analysis is carried out to test of hypothesis; Then calculate respectively the frequency of refusing null hypothesis in 100 height test of hypothesis results and not refusing null hypothesis;
Meanwhile, the harmony of Confounding Factor between group after randomized grouping when verifying that above-mentioned two groups of sample sizes are unequal, first according to experimental group sample size of above-mentioned database initialize process simulation is 30, control group sample size is 70 database, then this simulated database is carried out to randomized grouping 100 times, and the height entering for error-free minute after each randomized grouping in the sample of analysis is carried out to test of hypothesis; Finally calculate respectively the frequency of refusing null hypothesis in 100 height test of hypothesis results and not refusing null hypothesis;
4, result
According to the relevant parameter creation database in table 1, table 2,3,4,5 is described respectively the essential characteristic that sample size is 50,100,500,1000 o'clock databases; In all database, height is equal indifference (P > 0.05) between group; In the variant database of intervention effect, BMI there are differences (P < 0.05) between group; In intervention effect indifference database, BMI is indifference (P > 0.05) between group.
The odds value and the 95%CI thereof that whole simulated database analyses are obtained according to this analysis method are as shown in table 6, and the odds value that the variant database analysis of intervention effect is obtained is all greater than 1, and the odds value that the database analysis of intervention effect indifference is obtained is all less than 1.
16 simulated database final result variablees of table 2 sample size 50 and Confounding Factor variable essential characteristic
Tab2Characteristics of outcome variables and confounder variables in sixteensimulation datasets of50samples
16 simulated database final result variablees of table 3 sample size 100 and Confounding Factor variable essential characteristic
Tab3Characteristics of outcome variables and confounder variables in sixteensimulation datasets of100samples
16 simulated database final result variablees of table 4 sample size 500 and Confounding Factor variable essential characteristic
Tab4Characteristics of outcome variables and confounder variables in sixteensimulation datasets of500samples
16 simulated database final result variablees of table 5 sample size 1000 and Confounding Factor variable essential characteristic
Tab5Characteristics of outcome variables and confounder variables in sixteensimulation datasets of1000samples
Odds value and 95%CI under the different sample sizes of table 6
Tab6Odds and95％CI in difierent samples groups
Notes Significant differences：The effectiveness of two different treatments have significant differences or not；Y：The effectiveness of two different treatments have significant differences；N：The effectiveness of two different treatments have no significant differences。
(significant difference is described: whether the effect value between two different disposal groups has significant difference; Y: the effect value between two different disposal groups has notable difference; N: the effect value no significant difference between two different disposal groups.）
Claims (7)
1. the simulation RCT analytical approach based on True Data, is characterized in that, it comprises step,
(1) the variant database of intervention effect that contains two kinds of means of intervention and final result variable that first, utilizes computer simulation clinical intervention to produce: the body-mass index of usining carries out intervention effect diversity ratio as the final result variable of accepting after two kinds of different means of intervention; With the random series that is numbered of research object, carry out randomized grouping again, then within error-free minute, enter the research object BMI of analysis according to PP strategy, carry out comparing between two groups; Repeat the comparison of randomized grouping and BMI; Finally by calculating in BMI comparative analysis result, refuse null hypothesis experiment frequency and the ratio odds that does not refuse null hypothesis experiment frequency, repeat this process and obtain odds value and 95%CI;
(2) whether the distribution of observation odds and 95%CI distributes consistent with E/S size and Power; Odds and 95%CI present linear relationship with the variation of power;
(3) between the distribution of observing Odds and 95%CI thereof and group, whether the pass of real difference and power ties up under different sample sizes stable;
(4), the in the situation that of small sample, comprise database that sample size is equal and sample size database not etc., relatively the harmony of Confounding Factor between two groups.
2. by method claimed in claim 1, it is characterized in that, in described step (1), in database, experimental group n1 and control group n2 sample size are 50, wherein, and H
0for final result variable mean does not have difference, H between two groups
1, the mean of final result variable variant between two groups (two-sided test).
3. by method claimed in claim 1, it is characterized in that, in described step (1), described randomized grouping repeats 100 times.
4. by method claimed in claim 1, it is characterized in that, in described step (2), according to power value and with it the difference simulation of corresponding E/S create 8 of the simulated databases of 8 of the variant databases of intervention effect and intervention effect indifference and analyze respectively, calculate odds and 95%CI thereof.
5. by method claimed in claim 1, it is characterized in that, in described step (2), when the discrepant database of intervention effect is analyzed, power is larger, and the odds and the 95%CI that calculate are larger; When the database of intervention effect indifference is analyzed, power is larger, and the odds and the 95%CI that calculate are less; No matter whether intervention effect there are differences, and when power is 0.5, the odds calculating approaches 1, and its 95%CI all comprises 1.
6. by method claimed in claim 1, it is characterized in that, in described step (3), according under different sample sizes, the difference simulation of E/S and power creates a plurality of simulated databases and analyzes respectively, the odds that under different sample sizes, the variant database analysis of intervention effect obtains is all greater than to 1, odds value and 95%CI all increases with the increase of power; The odds value that under different sample sizes, intervention effect indifference database analysis obtains is all less than to 1, and odds value and 95%CI thereof all reduce with the increase of power.
7. by method claimed in claim 1, it is characterized in that, in described step (4), first database is carried out to randomized grouping 100 times, again to remove wrong minute sample according to PP strategy, then within error-free minute, entering the research object Confounding Factor of analysis, between two groups, whether there is difference and carry out test of hypothesis.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310038845.5A CN103970973B (en) | 2013-01-31 | 2013-01-31 | A kind of simulation RCT analysis methods based on True Data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310038845.5A CN103970973B (en) | 2013-01-31 | 2013-01-31 | A kind of simulation RCT analysis methods based on True Data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103970973A true CN103970973A (en) | 2014-08-06 |
| CN103970973B CN103970973B (en) | 2017-07-07 |
Family
ID=51240463
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310038845.5A Expired - Fee Related CN103970973B (en) | 2013-01-31 | 2013-01-31 | A kind of simulation RCT analysis methods based on True Data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103970973B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108242268A (en) * | 2017-07-18 | 2018-07-03 | 嘉兴太美医疗科技有限公司 | A kind of clinical investigation subject randomized grouping and the wrong identification and correcting method of therapy distribution |
| WO2021012074A1 (en) * | 2019-07-19 | 2021-01-28 | Ebay Inc. | Sample delta monitoring |
| CN112907364A (en) * | 2021-04-01 | 2021-06-04 | 重庆度小满优扬科技有限公司 | Assessment method and device of wind control analysis model |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101334770A (en) * | 2007-06-28 | 2008-12-31 | 科凌力医学软件（深圳）有限公司 | Evidence-based medical data base establishment method and related evidence-based medical disease diagnostic method |
-
2013
- 2013-01-31 CN CN201310038845.5A patent/CN103970973B/en not_active Expired - Fee Related
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108242268A (en) * | 2017-07-18 | 2018-07-03 | 嘉兴太美医疗科技有限公司 | A kind of clinical investigation subject randomized grouping and the wrong identification and correcting method of therapy distribution |
| WO2021012074A1 (en) * | 2019-07-19 | 2021-01-28 | Ebay Inc. | Sample delta monitoring |
| CN112907364A (en) * | 2021-04-01 | 2021-06-04 | 重庆度小满优扬科技有限公司 | Assessment method and device of wind control analysis model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103970973B (en) | 2017-07-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Lou et al. | AI on drugs: Can artificial intelligence accelerate drug development? Evidence from a large-scale examination of bio-pharma firms | |
| Hameduh et al. | Homology modeling in the time of collective and artificial intelligence | |
| Hajiramezanali et al. | Bayesian multi-domain learning for cancer subtype discovery from next-generation sequencing count data | |
| Zhang | Big data and clinical research: perspective from a clinician | |
| Kennedy et al. | Why data coherence and quality is critical for understanding interareal cortical networks | |
| Yang et al. | Unsupervised binning of environmental genomic fragments based on an error robust selection of l-mers | |
| CN103970973A (en) | Real-data-based simulative RCT (randomized controlled trial) analysis method | |
| Yan et al. | Radiomics analysis using stability selection supervised component analysis for right-censored survival data | |
| Zhou et al. | Audit to Forget: A Unified Method to Revoke Patients' Private Data in Intelligent Healthcare | |
| Dunn | Pragmatic trials of complex psychosocial interventions: methodological challenges | |
| Bastianini et al. | Recent developments in automatic scoring of rodent sleep | |
| Shiying et al. | Research on syndrome classification prediction model of tibetan medicine diagnosis and treatment based on data mining | |
| CN104239504A (en) | Data processing method for establishing of doctor competency model | |
| Li et al. | The research and development thinking on the status of artificial intelligence in traditional Chinese medicine | |
| Pak et al. | Developing disease risk prediction model based on environmental factors | |
| CN109887604B (en) | Quantitative determination system for similarity of traditional Chinese medicine disease names | |
| Guyon et al. | Design of the first neuronal connectomics challenge: From imaging to connectivity | |
| Park | Challenges and opportunities in biomarker-driven trials: adaptive randomization | |
| Wright et al. | Exploiting the speed-accuracy trade-off. | |
| Gupta et al. | Genome Sequence Identification using Deep Learning for Lung Cancer Diagnosis | |
| CN117393171B (en) | Method and system for constructing prediction model of LARS development track after rectal cancer operation | |
| CN113707247B (en) | Method for quantitatively predicting cartilage repair rate after mesenchymal stem cell therapy based on machine learning of blockchain | |
| Anand et al. | Building an intelligent integrated method of gene selection for facioscapulohumeral muscular dystrophy diagnosis | |
| Harvey et al. | Computational modelling of lung injury: is there potential for benefit? | |
| Owen et al. | Gameplay as assessment: Analyzing event-stream player data and learning using GBA (a Game-Based Assessment Model) |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170707 Termination date: 20210131 |