Skip to main content

Table 4 Associations of health endpoints (blood pressure, hypertension, and mortality) with ERS’s from different statistical approaches. For the comparison purpose, associations with blood lead and blood cadmium are presented

From: Construction of environmental risk score beyond standard linear models using machine learning methods: application to metal mixtures, oxidative stress and cardiovascular disease in NHANES

   AENET-M AENET-I BART SL Blood Lead Blood Cadmium
SBP β
(95% CI)
0.03
(−0.45, 0.51)
0.69
(0.20, 1.18)
0.52
(0.07, 0.97)
1.03
(0.57, 1.48)
0.78
(0.22, 1.34)
0.28
(−0.25, 0.82)
DBP β
(95% CI)
1.12
(0.73, 1.50)
1.50
(1.11, 1.88)
0.90
(0.55, 1.24)
1.61
(1.28, 1.95)
0.97
(0.60, 1.34)
0.29
(−0.12, 0.71)
Hypertension OR
(95% CI)
1.11
(1.02, 1.22)
1.26
(1.15, 1.38)
1.17
(1.09, 1.25)
1.30
(1.20, 1.40)
1.08
(0.99, 1.18)
1.06
(0.98, 1.16)
Total mortality HR
(95% CI)
1.07
(0.89, 1.30)
1.07
(0.92, 1.24)
1.07
(0.96, 1.19)
1.15
(0.98, 1.35)
1.08
(0.86, 1.36)
1.37
(1.10, 1.70)
CVD mortality HR
(95% CI)
0.99
(0.69, 1.41)
1.09
(0.78, 1.54)
1.07
(0.76, 2.51)
0.98
(0.71, 1.36)
0.92
(0.60, 1.41)
1.24
(0.80, 1.92)
Cancer mortality HR
(95% CI)
1.50
(1.05, 2.15)
1.24
(0.93, 1.63)
1.23
(0.99, 1.54)
1.23
(0.94, 1.60)
1.41
(1.01, 1.96)
1.50
(1.07, 2.10)
  1. AENET-M adaptive elastic net for main effects, AENET-I adaptive elastic net for main effects and pairwise interactions, BART Bayesian Additive Regression Tree, SL Super Learner
  2. Effect estimates (β, odds ratio (OR), and hazard ratio (HR)) are based on a standardized increment which is equivalent to one standard deviation increase in each ERS. All models were adjusted for age (except mortality outcomes), sex, race/ethnicity, body mass index, smoking status, education