Table 4.

Prediction for chemoresistance in the proof-of-principle test set

Sample ID	Feature selection P < 0.05				Feature selection P < 0.01
	Predictive Index Percentile1	Prediction For Pretreatment Sample2	Overall Prediction3		Predictive Index Percentile	Prediction for Pretreatment Sample	Overall Prediction
1 pre	39%	low	incorrect		35%	low	incorrect
1 post	22%				24%
2 pre	48%	low	correct		49%	low	correct
2 post	79%				77%
3 pre	55%	low	correct		46%	low	correct
3 post	73%				77%
4 pre	13%	low	correct		24%	low	correct
4 post	72%				73%
5 pre	31%	low	correct		33%	low	correct
5 post	48%				46%
6 pre	13%	low	correct		27%	low	correct
6 post	16%				66%
7 pre	7%	low	correct		12%	low	correct
7 post	66%				57%
8 pre	52%	low	incorrect		50%	low	incorrect
8 post	17%				24%

¹The predictive index was computed for each sample by this supervised principal component method, where a high value of the predictive index corresponds to a rapid progression after chemotherapy (i.e., short TTP). If the predictive index of a sample in the test set corresponded to the median predictive index of the training set, the sample was assigned a 50% predictive index.

²The risk was predicted low, if predictive index percentile of the pretreatment sample was less than 67%

³The prediction was considered correct if post-treatment samples were assigned a higher predictive index than pre-treatment samples.