Machine learning techniques have been widely used in many scientific fields, but its use in medical literature is limited partly because of technical difficulties. k-nearest neighbors (kNN) is a simple method of machine learning. The article introduces some basic ideas underlying the kNN algorithm, and then focuses on how to perform kNN modeling with R. The dataset should be prepared before running the knn() function in R. After prediction of outcome with kNN algorithm, the diagnostic performance of the model should be checked. Average accuracy is the mostly widely used statistic to reflect the kNN algorithm. Factors such as k value, distance calculation and choice of appropriate predictors all have significant impact on the model performance.

Propensity score matching (PSM) is a popular method in clinical researches to create a balanced covariate distribution between treated and untreated groups. However, the balance diagnostics are often not appropriately conducted and reported in the literature and therefore the validity of the findings from the PSM analysis is not warranted. The special article aims to outline the methods used for assessing balance in covariates after PSM. Standardized mean difference (SMD) is the most commonly used statistic to examine the balance of covariate distribution between treatment groups. Because SMD is independent of the unit of measurement, it allows comparison between variables with different unit of measurement. SMD can be reported with plot. Variance is the second central moment and should also be compared in the matched sample. Finally, a correct specification of the propensity score model (e.g., linearity and additivity) should be re-assessed if there is evidence of imbalance between treated and untreated. R code for the implementation of balance diagnostics is provided and explained.

Time-varying covariance occurs when a covariate changes over time during the follow-up period. Such variable can be analyzed with the Cox regression model to estimate its effect on survival time. For this it is essential to organize the data in a counting process style. In situations when the proportional hazards assumption of the Cox regression model does not hold, we say that the effect of the covariate is time-varying. The proportional hazards assumption can be tested by examining the residuals of the model. The rejection of the null hypothesis induces the use of time varying coefficient to describe the data. The time varying coefficient can be described with a step function or a parametric time function. This article aims to illustrate how to carry out statistical analyses in the presence of time-varying covariates or coefficients with R.

Multiple imputation with multivariate imputation by chained equation (MICE) package

While purposeful selection is performed partly by software and partly by hand, the stepwise and best subset approaches are automatically performed by software. Two R functions stepAIC() and bestglm() are well designed for stepwise and best subset regression, respectively. The stepAIC() function begins with a full or null model, and methods for stepwise regression can be specified in the direction argument with character values “forward”, “backward” and “both”. The bestglm() function begins with a data frame containing explanatory variables and response variables. The response variable should be in the last column. Varieties of goodness-of-fit criteria can be specified in the IC argument. The Bayesian information criterion (BIC) usually results in more parsimonious model than the Akaike information criterion.

Logistic regression is one of the most commonly used models to account for confounders in medical literature. The article introduces how to perform purposeful selection model building strategy with R. I stress on the use of likelihood ratio test to see whether deleting a variable will have significant impact on model fit. A deleted variable should also be checked for whether it is an important adjustment of remaining covariates. Interaction should be checked to disentangle complex relationship between covariates and their synergistic effect on response variable. Model should be checked for the goodness-of-fit (GOF). In other words, how the fitted model reflects the real data. Hosmer-Lemeshow GOF test is the most widely used for logistic regression model.