Table 2.

Prediction of variable using the RSF model

VariableaPredictive valuebVIMP
AKT1 rs24947402.71620.0066
BMI2.85660.0011
AKT1 rs24947442.93880.0090
AKT1 rs24987893.11140.0075
Dietary alcohol per day3.13860.0016
Age at menopause3.27400.0002
Depressive symptom3.41220.0004
AKT1 rs11302143.61160.0030
Waist-to-hip ratio3.8390−0.0005
IRS1 rs18012784.05620.0066
Percent calories from SFA per day4.05920.0011
Percent calories from fat per day4.1002−0.0007
Physical activity4.84080.0003
Age4.90820.0002
Age at menarche4.96620.0001
Family income5.41100.0000
Number of pregnancies5.8770−0.0001
E + P use7.81100.0008
Family history of breast cancer8.20200.0006
History of either hysterectomy or oophorectomy10.17060.0002
Smoking status10.20460.0000
E only use11.2790−0.0001
  • Abbreviations: BMI, body mass index; E, estrogen; P, progestin; SFA, saturated fatty acids; VIMP, variable of importance.

  • aVariables are ordered by predictive value.

  • bPredictive value of variable was assessed via minimal depth method in the nested RSF models. A lower value is likely to affect greatly prediction.