. 2021 Jan 30;11(2):206. doi: 10.3390/diagnostics11020206

Table 2.

The table collects the studies that used different techniques than machine learning to predict diagnosis and prognosis.

Aim	Technique and Data	Results	Reference
ccRCC survival at 5 years	Clustering on cDNA microarray (29 patients)	40 genes correlated with survival (Kaplan–Meier, p < 0.0001) and histological grade.	Takahashi et al., 2001 [82]
metastatic ccRCC subgroups	Clustering on cDNA microarray (58 patients)	45 genes distinguishing groups for overall survival (p = 0.001)	Vasselli et al., 2003 [83]
ccRCC survival after surgery	Clustering and supervised PCA on cDNA microarray (177 patients)	259 genes correlated with survival (p < 0.001 by the log-rank test on test set)	Zhao et al., 2006 [84]
ccRCC survival after surgery		Top 4 genes correlated with survival (p = 0.02)	Zhao et al., 2006 [84]
ccRCC vs. normal	Clustering and PCA on cDNA microarray (16 patients)	21 genes over expressed in ccRCC compared to normal tissues	Skubitz et al., 2006 [57]
ccRCC subgroups	Clustering and PCA on cDNA microarray (16 patients)	Fewer genes distinguishing 2 ccRCC subgroups likely related to pathologic grade	Skubitz et al., 2006 [57]
ccRCC survival	PCA and clustering and logical analysis on cDNA microarray (48 patients)	110 genes associated with tumor stage (p = 0.009) and grade (p = 0.0007) and survival (median survival of 8.6 vs. 2.0 years, p = 0.002)	Brannon et al., 2010 [85]