In-silico performance, validation, and modeling of the Nanostring Banff Human Organ transplant gene panel using archival data from human kidney transplants

Table 3 Additional models for confusion matrices/discrepancies

A. Average discrepancies between Diagnoses All PC
Model	Target	Data	% Discrepancies
Multinomial logistic regression	Diagnoses	ALL PC	37.3
Boot strap forest	Diagnoses	ALL PC	24.1
Partial least squares	Diagnoses	ALL PC	37.3

B. Average discrepancies between Clusters All PC
Model	Target	Data	% Discrepancies
Multinomial logistic regression	Clusters	ALL PC	10.0
Boot strap forest	Clusters	ALL PC	12.2
Partial least squares	Clusters	ALL PC	15.1

C. Average discrepancies CPPC, UPC, & PBPC
Model	Target	Data	% Discrepancies
Multinomial logistic regression	Diagnoses	CPPC	36.9
Boot strap forest	Diagnoses	CPPC	25.6
Partial least squares	Diagnoses	CPPC	27.2
Multinomial logistic regression	Diagnoses	UPC	24.4
Boot strap forest	Diagnoses	UPC	24.9
Partial least squares	Diagnoses	UPC	27.0
Multinomial logistic regression	Diagnoses	PBPC	35.2
Boot strap forest	Diagnoses	PBPC	27.5
Partial least squares	Diagnoses	PBPC	29.2

D. Average per sample discrepancies between Diagnoses All PC
Model	Target	Data	% Discrepancies
Multinomial logistic regression vs Boot strap forest	Diagnoses	ALL PC	33.1
Multinomial logistic regression vs Partial least squares	Diagnoses	ALL PC	32.7
Boot strap forest vs Partial least squares	Diagnoses	ALL PC	33.1

E. Average per sample discrepancies between Clusters All PC
Model	Target	Data	% Discrepancies
Multinomial logistic regression vs Boot strap forest	Clusters	ALL PC	12.4
Multinomial logistic regression vs Partial least squares	Clusters	ALL PC	13.5
Boot strap forest vs Partial least squares	Clusters	ALL PC	14.4

F. Additional models for diagnoses all PC, CPPC, PBPC, & UPC*
Model	Target	Data	Mean accuracy	Average precision/recall	Percent errors
Extreme Gradient Boosting	Diagnoses	ALL PC	0.74	0.79	36.5
Linear Discriminant Analysis	Diagnoses	CPPC	0.75	0.78	32.6
Extra Tree Classifier	Diagnoses	PBPC	0.75	0.76	35.3
Linear discriminant analysis	Diagnoses	UPC	0.76	0.70	33.9

G. Additional models for clusters all PC, CPPC, PBPC, & UPC*
Model	Target	Data	Mean accuracy	Average precision/recall	Percent errors
Multinomial logistic regression	Clusters	ALL PC	0.93	0.95	6.9
Extra tree classifier	Clusters	CPPC	0.90	0.93	6.9
Extreme gradient boosting	Clusters	PBPC	0.92	0.94	6.2
Multinomial logistic regression	Clusters	UPC	0.92	0.94	6.9

ISSN: 1755-8794