From: Subtyping irritable bowel syndrome using cluster analysis: a systematic review
Study, year [reference] | Preprocessing methods | Determination of clusters number | Clustering algorithms | Validation | Interpretation of results |
---|---|---|---|---|---|
Black et al. 2021 [59] | T-test and partially overlapping z-test | Bayesian information criterion of the log-likelihood (BIC(LL)) | Latent class analysis | tenfold cross-validation | χ2 test, ANOVA |
Han et al. 2019 [60] | EFA using the principal component method | likelihood‐based criteria and model entropy | Latent class analysis | NR | Χ2 test, ANOVA R2 and Cohen’s d |
Lackner et al. 2013 [61] | Simple linear transformation, CFA based on maximum likelihood methods, EFA based on principal components extraction | Successive solutions that increment the value of k by 1 | K-means cluster analysis | Exploratory cluster analysis | Squared semi-partial correlation |
Nevé et al. 2013 [63] | PFA | A natural break in distance jumps | Ascending hierarchical cluster analysis | Silhouette coefficient | ANOVA, correlation matrix, and scatterplots |
Eslick et al. 2004 [42] | Factor analysis, PCA, varimax rotation | Euclidian distance | K-means cluster analysis | NR | Describing a cluster profile that comprised the mean score per factor per cluster and, no cluster could be made up of less than 5% of the entire sample |
Guthrie et al. 2003 [51] | NR | Euclidian distance | K-means cluster analysis | Bonferroni corrected pairwise comparisons | ANOVA or × 2 tests, Pearson correlation coefficients |
Ragnarsson et al. 1999 [52] | Transformed, to have a mean of 0 and a standard deviation of 1 | Euclidian distance | K-means cluster analysis | NR | Mann–Whitney and Kruskal– Wallis tests, ANOVA |
Howard Mertz et al. 1995 [57] | unpaired Student's t-test, standardized | Pseudo f statistic | K-means cluster analysis | NR | Pearson r correlation coefficient, χ2 analysis |
Ragnarsson et al. 1999 [53] | Transformed, to have a mean of 0 and standard deviation of 1 | Squared Euclidian distance | K-means cluster analysis | NR | Mann–Whitney, Kruskal–Wallis tests, ANOVA, Wilcoxon signed-rank test |
Bouchoucha et al. 1999 [54] | Normalized by subtracting the pressure of the first measured point from the measured values in each experiment | NR | K-mean cluster analysis | NR | Mann–Whitney test, Wilcoxon signed rank test., correlation coefficient used spearman r |
Bouchoucha et al. 2006 [55] | NR | Pseudo-f statistic | K-means cluster analysis | NR | Chi-square test, GLM, ANOVA |
Bennet et al. 2018 [62] | NR | NR | Hierarchical cluster analysis | Q2 | Mann–Whitney u test, Kruskal Wallis followed by Dunn’s test, hoteling’s T2, OPLS-DA, PCA, Non-parametric spearman’s rank coefficient |
Johanna Sundin et al. 2019 [56] | NR | NR | Hierarchical cluster analysis | Cross-validation by Q2 parameter | OPLS‐ DA, Mann‐Whitney, and Kruskal‐Wallis followed by Dunn's test, Hoteling’s T2 |
Ian et al. 2014 [58] | Normalized by scaling to an intensity of 1 to control for differing numbers of reads | NR | Unsupervised hierarchical cluster analysis | NR | Pearson correlation coefficient test, Kruskal Wallis and Mann Whitney test, Blast method, PMANOVA |