1 FRESA.CAD Regresion Benchmark

1.1 BRCA Recurence RISK Data Set


lesionsSumDiffRed <- NULL
load("RadiomicsBRCA.RDATA")

RA_BRCA_FRESA <- FRESA.Model(formula = Risk ~ 1,data = lesionsSumDiffRed,repeats = 20)

1.2 Benchmark


cp <- CVRegBenchmark(theData = lesionsSumDiffRed, theOutcome = "Risk", reps = 200, fraction = 0.90, topincluded = 50 )


elapcol <- names(cp$times[[1]]) == "elapsed"
cputimes <- list(Fresa = mean(cp$times$fresatime[ elapcol ]),LASSO = mean(cp$times$LASSOtime[ elapcol ]),RF = mean(cp$times$RFtime[ elapcol ]),SVM = mean(cp$times$SVMtime[ elapcol ]))

featsize <- list(Fresa = mean(cp$featSize$FRESASize),LASSO = mean(cp$featSize$LASSOSize),Univ = mean(cp$featSize$UNIVSize))

1.2.1 Results

#The Times
pander::pander(cputimes)
  • Fresa: 1.25
  • LASSO: 0.2431
  • RF: 1.786
  • SVM: 0.2033
pander::pander(featsize)
  • Fresa: 7.315
  • LASSO: 15.06
  • Univ: 13.61


plotMAEEvolution(cp,30,main="Mean Absolute Error (MAE)", location="topright")




bp <- barPlotCiError(as.matrix(cp$CorTable),metricname="Pearson Correlation",thesets=thesets,themethod=theMethod,main="Pearson Correlation",offsets = c(0.5,0.05),args.legend = list(x = "bottomright"))


pander::pander(bp$barMatrix,caption = "Pearson Correlation",round = 3)
Pearson Correlation
  Default Regresion Method SVM Rendering Filtered
B:SWiMS 0.49 0.548
LASSO 0.372 0.45
RF 0.219 0.31
SVM 0.055 0.528 
pander::pander(bp$ciTable,caption = "Pearson Correlation with 95%CI",round = 3)
Pearson Correlation with 95%CI
  Pearson Correlation lower upper
Default Regresion Method 0.49 0.29 0.649
Default Regresion Method 0.372 0.152 0.557
Default Regresion Method 0.219 -0.015 0.43
Default Regresion Method 0.055 -0.181 0.284
SVM Rendering Filtered 0.548 0.36 0.693
SVM Rendering Filtered 0.45 0.242 0.618
SVM Rendering Filtered 0.31 0.083 0.507
SVM Rendering Filtered 0.528 0.336 0.677 


bp <- barPlotCiError(as.matrix(cp$RMSETable),metricname="RMSE",thesets=thesets,themethod=theMethod,main="RMSE",offsets = c(0.5,5),args.legend = list(x = "bottomright"))

pander::pander(bp$barMatrix,caption = "RMSE",round = 3)
RMSE
  Default Regresion Method SVM Rendering Filtered
B:SWiMS 48.42 46.58
LASSO 51.39 49.55
RF 54.05 53.01
SVM 56.13 47.32 
pander::pander(bp$ciTable,caption = "RMSE with 95%CI",round = 3)
RMSE with 95%CI
  RMSE lower upper
Default Regresion Method 48.42 41.6 57.94
Default Regresion Method 51.39 44.15 61.48
Default Regresion Method 54.05 46.44 64.68
Default Regresion Method 56.13 48.22 67.16
SVM Rendering Filtered 46.58 40.02 55.73
SVM Rendering Filtered 49.55 42.57 59.28
SVM Rendering Filtered 53.01 45.55 63.43
SVM Rendering Filtered 47.32 40.65 56.62 


bp <- barPlotCiError(as.matrix(cp$BiasTable),metricname="BIAS",thesets=thesets,themethod=theMethod,main="BIAS",offsets = c(0.5,0.5),args.legend = list(x = "bottomright"))

pander::pander(bp$barMatrix,caption = "BIAS",round = 3)
BIAS
  Default Regresion Method SVM Rendering Filtered
B:SWiMS 1.948 -0.157
LASSO 0.945 0.237
RF 0.47 -1.077
SVM -1.359 1.928 
pander::pander(bp$ciTable,caption = "BIAS with 95%CI",round = 3)
BIAS with 95%CI
  BIAS lower upper
Default Regresion Method 1.948 -9.504 13.4
Default Regresion Method 0.945 -11.22 13.11
Default Regresion Method 0.47 -12.32 13.26
Default Regresion Method -1.359 -14.64 11.92
SVM Rendering Filtered -0.157 -11.18 10.87
SVM Rendering Filtered 0.237 -11.49 11.96
SVM Rendering Filtered -1.077 -13.62 11.47
SVM Rendering Filtered 1.928 -9.263 13.12

1.3 Features

pander::pander(summary(RA_BRCA_FRESA$BSWiMS.model,caption="Recurency Risk model",round = 3))

  • coefficients:

    Table continues below
      Estimate lower mean upper
    D_CC_LH_3_ICDF_0.99 -0.4619 -0.4714 -0.4619 -0.4523
    S_CC_FRACTAL_ICDF_0.95 0.2233 0.2196 0.2233 0.2271
    D_CC_HL_4_ICDF_0.999 -0.05269 -0.05418 -0.05269 -0.0512
    D_CC_LH_3_ICDF_0.999 -0.04899 -0.05046 -0.04899 -0.04753
    S_CC_FRACTAL_Mean 0.02206 0.01801 0.02206 0.02612
    S_CC_FRACTAL_ICDF_0.75 0.0183 0.01525 0.0183 0.02136
    S_CC_LH_2_Mean -0.2247 -0.3738 -0.2247 -0.07557
    S_CC_FRACTAL_ICDF_0.25 0.003036 0.001275 0.003036 0.004796
    S_CC_FRACTAL_ICDF_0.05 0.001385 0.0005282 0.001385 0.002241
    S_CC_LH_2_z_Mean -8.144 -13.7 -8.144 -2.586
    Table continues below
      u.MSE r.MSE model.MSE NeRI F.pvalue
    D_CC_LH_3_ICDF_0.99 2443 2043 1632 0.07465 6.916e-05
    S_CC_FRACTAL_ICDF_0.95 2422 2162 1655 0.4225 1.363e-05
    D_CC_HL_4_ICDF_0.999 2750 1995 1632 0.07324 0.0001689
    D_CC_LH_3_ICDF_0.999 2503 2495 2128 0.1393 0.0008095
    S_CC_FRACTAL_Mean 2473 2685 2215 0.2254 0.0002312
    S_CC_FRACTAL_ICDF_0.75 2483 2729 2278 0.1744 0.0003814
    S_CC_LH_2_Mean 2586 2776 2400 0.1479 0.001501
    S_CC_FRACTAL_ICDF_0.25 2535 2965 2473 0.1925 0.0003579
    S_CC_FRACTAL_ICDF_0.05 2575 2873 2443 0.1362 0.000755
    S_CC_LH_2_z_Mean 2629 2911 2535 0.1338 0.00194
      t.pvalue Sign.pvalue Wilcox.pvalue
    D_CC_LH_3_ICDF_0.99 0.01884 0.3023 0.0487
    S_CC_FRACTAL_ICDF_0.95 0.001452 0.0002114 0.001148
    D_CC_HL_4_ICDF_0.999 0.006883 0.308 0.009542
    D_CC_LH_3_ICDF_0.999 0.02556 0.1338 0.04206
    S_CC_FRACTAL_Mean 0.02121 0.02836 0.01173
    S_CC_FRACTAL_ICDF_0.75 0.0309 0.08703 0.01957
    S_CC_LH_2_Mean 0.02613 0.1247 0.03159
    S_CC_FRACTAL_ICDF_0.25 0.0239 0.06265 0.01467
    S_CC_FRACTAL_ICDF_0.05 0.02807 0.134 0.05397
    S_CC_LH_2_z_Mean 0.03496 0.1559 0.05288
  • MSE: 1616
  • R2: 0.4791

  • bootstrap:

gain <- length(RA_BRCA_FRESA$BSWiMS.models$formula.list)/20
gplots::heatmap.2(gain*RA_BRCA_FRESA$BSWiMS.models$bagging$formulaNetwork,trace="none",mar=c(10,10),main="B:SWiMS Formula Network")