ABSTRACT
OBJECTIVE@#To explore the application and advantages of conditional inference forest in survival analysis.@*METHODS@#We used simulated experiment and actual data to compare the predictive performance of 4 models, including Coxproportional hazards model, accelerated failure time model, random survival forest model and conditional inference forest model based on their Brier scores.@*RESULTS@#Simulation experiment suggested that both of the two forest models had more accurate and robust predictive performance than the other two regression models. Conditional inference forest model was superior to the other models in analyzing time-to-event data with polytomous covariates, collinearity or interaction, especially for a large sample size and a high censoring rate. The results of actual data analysis demonstrated that conditional inference forest model had the best predictive performance among the 4 models.@*CONCLUSIONS@#Compared with the commonly used survival analysis methods, conditional inference forest model performs better especially when the data contain polytomous covariates with collinearity and interaction.
Subject(s)
Data Analysis , Proportional Hazards Models , Sample Size , Survival AnalysisABSTRACT
OBJECTIVE@#To explore the relationship between sample size in the groups and statistical power of ANOVA and Kruskal-Wallis test with an imbalanced design.@*METHODS@#The sample sizes of the two tests were estimated by SAS program with given parameter settings, and Monte Carlo simulation was used to examine the changes in power when the total sample size varied or remained fixed.@*RESULTS@#In ANOVA, when the total sample size was fixed, increasing the sample size in the group with a larger mean square error improved the statistical power, but an excessively large difference in the sample sizes between groups led to reduced power. When the total sample size was not fixed, a larger mean square error in the group with increased sample size was associated with a greater increase of the statistical power. In Kruskal-wallis test, when the total sample size was fixed, increasing the sample size in groups with large mean square errors increased the statistical power irrespective of the sample size difference between the groups; when total sample size was not fixed, a larger mean square error in the group with increased sample size resulted in an increased statistical power, and the increment was similar to that for a fixed total sample size.@*CONCLUSIONS@#The relationship between statistical power and sample size in groups is affected by the mean square error, and increasing the sample size in a group with a large mean square error increases the statistical power. In Kruskal-Wallis test, increasing the sample size in a group with a large mean square error is more cost- effective than increasing the total sample size to improve the statistical power.
Subject(s)
Computer Simulation , Models, Statistical , Monte Carlo Method , Sample SizeABSTRACT
OBJECTIVE@#We propose a strategy for identifying subgroups with the treatment effect from the survival data of a randomized clinical trial based on accelerated failure time (AFT) model.@*METHODS@#We applied adaptive elastic net to the AFT model (designated as the penalized model) and identified the candidate covariates based on covariate-treatment interactions. To classify the patient subgroups, we utilized a likelihood-based change-point algorithm to determine the threshold cutoff point. A two-stage adaptive design was adopted to verify if the treatment effect existed within the identified subgroups.@*RESULTS@#The penalized model with the main effect of the covariates considerably outperformed the univariate model without the main effect for the trial data with a small sample size, a high censoring rate, a small subgroup size, or a sample size that did not exceed the number of covariates; in other scenarios, the latter model showed better performances. Compared with the traditional design, the adaptive design improved the power for detecting the treatment effect where subgroup effect exists with a well-controlled type Ⅰ error.@*CONCLUSIONS@#The penalized AFT model with the main effect of the covariates has advantages in subgroup identification from the survival data of clinical trials. Compared with the traditional design, the two-stage adaptive design has better performance in evaluation of the treatment effect when a subgroup effect exists.
ABSTRACT
We propose a subgroup identification method based on the Logistic model for data from a two-arm clinical trial with dichotomous outcome variables.In this method, binary Logistic regression models are established for each group to calculate the outcome probabilities of each patient for comparison.According to the established rules, the patients are classified into their corresponding subgroups to establish a multinomial Logistic regression model.We simulated the false rate, correct judgment rate, coincidence rate and model correct judgment rate for different sample sizes and carried out an example analysis.The results of simulation showed that for different sample sizes, the false rates of this method were below 0.07 and the correct judgment rates were all above 0.75 with adequate coincidence rates and model correct judgment rates, demonstrating the effectiveness and reliability of the proposed method for subgroup identification.
Subject(s)
Humans , Clinical Trials as Topic , Computer Simulation , Logistic Models , Reproducibility of Results , Sample SizeABSTRACT
Objective To investigate the diagnostic value of six MRI characteristic features for diagnosing central neurocytoma (CN).Methods MRI data of 30 cases of CN and 68 cases of non-CN located in lateral ventricles were retrospectively analyzed.Six characteristic MRI features,including scalloping sign,broad-based attachment sign,soapbubble sign,peripheral cyst sign,fluid-fluid level sign and gemstone sign were scored based on a five-point scale.ROC curve was used to assess the diagnostic value of each MRI sign.Results The scalloping sign showed the highest area under the curve (AUC) value (0.82) among all 6 signs (all P＜0.05),followed by broad-based attachment,soap-bubble andperipheral cyst signs (AUC 0.73-0.75),higher than that of fluid-fluid level sign and gemstone sign (all P＜0.05).The scalloping sign exhibited the highest specificity (84.56 ％),followed by fluid-fluid level (77.94 ％),gemstone (74.26 ％) and peripheral cyst (70.34％) sign.The soap-bubble sign (83.89％) was the most sensitive sign,followed by broad-based attachment sign (76.11％) and peripheral cyst sign (75.00％).Conclusion The scalloping sign is the most valuable indicator for CN among six characteristic MRI features.
ABSTRACT
Objective To construct a new index(abbreviate FQ statistic) for testing the balance of covariates among 3 groups;to compare the power of hypothesis testing,standardized difference and FQ statistics to test the balance of covariates among 3 groups.Methods Using pooled variance to build FQ Statistic;Calculating propensity score for each individual by using ordinal logistic regression and multinomial logistic regression;Comparing the power of hypothesis testing,standardized difference and FQ statistics to test the balance of covariates among 3 groups by Monte Carlo simulation.Results The distribution of a covariate can be considered balanced among the 3 groups if FQ statisticsis less than 0.2.The power of hypothesis test is affected by sample size but FQ statistics and standardized difference.The power of FQ statistics and standardized difference to test the balance of covariates among 3 groups are higher than hypothesis testing,and both highly consistent.Conclusion FQ statistics and standardized differences are valid methods to test the balance of covariates among 3 groups.With more convenient calculating step than standardized difference,FQ statistic has more advantages in applications.
ABSTRACT
<p><b>OBJECTIVE</b>To realize propensity score matching in PS Matching module of SPSS and interpret the analysis results.</p><p><b>METHODS</b>The R software and plug-in that could link with the corresponding versions of SPSS and propensity score matching package were installed. A PS matching module was added in the SPSS interface, and its use was demonstrated with test data.</p><p><b>RESULTS</b>Score estimation and nearest neighbor matching was achieved with the PS matching module, and the results of qualitative and quantitative statistical description and evaluation were presented in the form of a graph matching.</p><p><b>CONCLUSION</b>Propensity score matching can be accomplished conveniently using SPSS software.</p>
Subject(s)
Propensity Score , SoftwareABSTRACT
Objective Medical statistics is an important tool in scientific research and practice.The article was to evaluate the current knowledge, learning needs and related influencing factors of medical statistics among medical doctoral students to provide references for the optimization in the teaching mode of medical statistics. Methods A questionnaire was used to investigate Grade 2014 doctoral students who took medical statistics course before and after the teaching in a medical university. Results The knowl-edge of doctoral students were at different levels and they had different learning requirements.More than 81.4% of the respondents have difficulty in the choice of scientific and reasonable design and the scientific design.Over 50%students had insufficient knowledge of advanced statistics and poor software application ability.They had the demands for more practical cases explanation in theory course, more software teaching hours and some short term specialized courses. Conclusion The university and teachers should carry out teaching reform according to the particularity of doctoral students and teach in small sections according to their knowledge levels of sta-tistics and learning requirements.It is suggested that the teaching should be combined with the students'own research projects in order to improve the teaching effect and improve the students'ability in sta-tistical design and solving the practical problem in scientific research statistics.
ABSTRACT
<p><b>OBJECTIVE</b>To study the incidence of implantation metastasis of breast cancer in vacuum-assisted breast biopsy (VABB) needle tract in Chinese patients and evaluate the effect of neoadjuvant chemotherapy on needle tract metastasis following VABB.</p><p><b>METHODS</b>The breast cancer patients with established diagnosis by VABB were divided into two groups to receive open surgery or neoadjuvant chemotherapy prior to open surgery. The incidence of needle tract metastasis, disease-free survival (DFS) and overall survival (OS) were compared between the two groups.</p><p><b>RESULTS</b>A total of 214 patients were enrolled, among whom 94 directly underwent surgeries and 120 had neoadjuvant chemotherapy before surgery. The two groups showed no significant differences in the incidence of needle tract metastasis (3.2% vs 0.8%, P=0.206), DFS (P=0.221), or OS (P=0.531).</p><p><b>CONCLUSION</b>The incidence of needle tract metastasis is low after VABB, and neoadjuvant chemotherapy does not increase this risk.</p>
Subject(s)
Female , Humans , Biopsy, Needle , Methods , Breast , Breast Neoplasms , Pathology , Disease-Free Survival , Incidence , Needles , Neoadjuvant Therapy , Neoplasms, Second Primary , Drug Therapy , VacuumABSTRACT
<p><b>OBJECTIVE</b>To investigate the impact of pre-operative uric acid on acute kidney injury (AKI) after cardiac surgery in elderly patients.</p><p><b>METHODS</b>Clinical data were collected from 936 elderly patients (age ≥ 60 years) undergoing cardiac surgery with cardiopulmonary bypass in Guangdong General Hospital between January 2005 and May 2011. The baseline serum creatinine was defined as the latest serum creatinine before surgery, and AKI was diagnosed according to RIFLE criteria. Patients were divided into three groups according to the sex-specific cutoff values of serum uric acid tertiles (group A: ≤ 384.65 µmol/L in men, and ≤ 354.00 µmol/L in women; group B:384.66-476.99 µmol/L in men and 354.01-437.96 µmol/L in women; group C: ≥ 477.00 µmol/L in men and ≥ 437.97 µmol/L in women). Multivariate logistic regression analysis was used to analyze the independent risk factors for AKI.</p><p><b>RESULTS</b>Among 936 elderly patients, 576 cases (61.5%) developed AKI. Mean uric acid concentration was higher in AKI patients than in Non-AKI patients ( (436.6 ± 119.1) µmol/L vs. (398.0 ± 107.2) µmol/L, P < 0.001). The incidence of AKI was 56.1% (175/312) in group A, 56.3% (175/311) in group B, 72.2% (226/313) in group C (P < 0.001). Multiple logistic regression analysis showed that, after adjusted for age, gender, co-morbidities(hypertension, diabetes mellitus, cerebrovascular disease, chronic obstructive pulmonary disease), previous cardiac surgery, eGFR<60 ml×min(-1) ×1.73 m(-2), heart function ≥ 3 (NYHA), positive urine protein, combination of coronary artery bypass grafting and valvular surgery, cardiopulmonary bypass operation time, aortic cross-clamping time, pre-operative angiotensin converting enzyme inhibitor or angiotensin II receptor blockers and lipid-lowering drugs use, early postoperative angiotensin converting enzyme inhibitor or angiotensin II receptor blockers, diuretics and digoxin use, post-operation central venous pressure, risk of post operative AKI was significantly higher in group C than in group A (OR:1.897, 95%CI: 1.270-2.833, P = 0.002).</p><p><b>CONCLUSION</b>Pre-operative elevated uric acid is an independent risk factor of AKI after cardiac surgery in elderly patients.</p>
Subject(s)
Aged , Female , Humans , Male , Middle Aged , Acute Kidney Injury , Angiotensin Receptor Antagonists , Angiotensin-Converting Enzyme Inhibitors , Cardiac Surgical Procedures , Cardiopulmonary Bypass , Coronary Artery Bypass , Incidence , Kidney Function Tests , Predictive Value of Tests , Risk Factors , Uric Acid , BloodABSTRACT
<p><b>OBJECTIVE</b>To analyze the risk factors of nonspecific low back pain in community populations.</p><p><b>METHODS</b>Two community populations were investigated using questionnaires in this case-control study. The questionnaire was designed to collect data including age, gender, body weight, marriage, education, income, occupation, labor intensity, smoking, alcohol drinking and social mental status. The subjects with low back pain constituted the case group and those without low back pain served as the control group, and the data was analyzed by a Logistic regression model.</p><p><b>RESULTS</b>A total of 1747 community residents participated in this survey, among whom 398 subjects had low back pain and 1126 subjects without low back pain were selected as the control group. Of all the latent risk factors of low back pain in Logistic regression model, gender was the most relevant factor (OR=3.5522) followed by education (OR=1.958), labor intensity (OR=1.956), marital status (OR=1.612), vibration source exposure (OR=1.491), BMI (OR=1.127) and age (OR=1.060).</p><p><b>CONCLUSION</b>Gender, education, labor intensity, marriage, vibration source exposure and BMI are risk factors of nonspecific low back pain in community populations, and exercises and mental status can be protective factors against low back pain.</p>
Subject(s)
Humans , Case-Control Studies , Logistic Models , Low Back Pain , Epidemiology , Risk Factors , Surveys and QuestionnairesABSTRACT
<p><b>OBJECTIVE</b>To analyze binary classification repeated measurement data with generalized estimating equations (GEE) and generalized linear mixed models (GLMMs) using SPSS19.0.</p><p><b>METHODS</b>GEE and GLMMs models were tested using binary classification repeated measurement data sample using SPSS19.0.</p><p><b>RESULTS AND CONCLUSION</b>Compared with SAS, SPSS19.0 allowed convenient analysis of categorical repeated measurement data using GEE and GLMMs.</p>