E current study protocol. All patients have signed written informed consents : Différence entre versions
m |
m |
||
| Ligne 1 : | Ligne 1 : | ||
| − | + | NLR ratio was calculated as NLR after initial treatment divided by NLR before initial treatment. 2.1. Statistical analysis For categorical data, the Fisher exact test or Pearson x2 test were used. Accordingly, the Mann hitney U test or independent sample t test were employed. To develop the predictive model, stepwise logistic regression was used, where the final diagnosis was set as the dependent variable and the following characteristics as independent variables: patient age, gender, primary tumor size, tumor location, tumor grade, histological classification, monocyte ratio, and NLR ratio. The final model was established through eliminating variables by backward selection, where the [http://www.share-dollar.com/comment/html/?19045.html Yond the present capacity (over 30,000 coding sequences] selective criterion was statistically significant level of 0.05. If using a relevantly more liberal P value of 0.10, similar results would be observed. After tested all potential clinical interactions, since no statistically significant results were found, all of them were eliminated in the final model. Furthermore, all predictors entered in the final model were reported their odds ratios (ORs) and 95 confidence intervals (CIs). The final model could be applied to compute the estimated probabilities of metastases for study individuals. To construct the receiver-operating characteristic curve, the predicted probabilities and definitive diagnoses of metastases were used. Then, in order to describe the accuracy of the model, the AUCs and their 95 CI were reported. The cross-validation procedure was selected in model validation procedure, whose advantages were allowing to access the entire dataset for model validation. P values | |
Version du 27 février 2018 à 09:46
NLR ratio was calculated as NLR after initial treatment divided by NLR before initial treatment. 2.1. Statistical analysis For categorical data, the Fisher exact test or Pearson x2 test were used. Accordingly, the Mann hitney U test or independent sample t test were employed. To develop the predictive model, stepwise logistic regression was used, where the final diagnosis was set as the dependent variable and the following characteristics as independent variables: patient age, gender, primary tumor size, tumor location, tumor grade, histological classification, monocyte ratio, and NLR ratio. The final model was established through eliminating variables by backward selection, where the Yond the present capacity (over 30,000 coding sequences selective criterion was statistically significant level of 0.05. If using a relevantly more liberal P value of 0.10, similar results would be observed. After tested all potential clinical interactions, since no statistically significant results were found, all of them were eliminated in the final model. Furthermore, all predictors entered in the final model were reported their odds ratios (ORs) and 95 confidence intervals (CIs). The final model could be applied to compute the estimated probabilities of metastases for study individuals. To construct the receiver-operating characteristic curve, the predicted probabilities and definitive diagnoses of metastases were used. Then, in order to describe the accuracy of the model, the AUCs and their 95 CI were reported. The cross-validation procedure was selected in model validation procedure, whose advantages were allowing to access the entire dataset for model validation. P values
