Skip to main content

Table 2 LASSO-selection of best predictors

From: Serum metabolites predict response to angiotensin II receptor blockers in patients with diabetes mellitus

Metabolite Mean estimate Standard deviation 95 % CI Selection percentagea
Asymmetric dimethylarginine (ADMA) 4.4 6.5 0, 22 52.4
Asparagine (Asp) −3.0 7.4 −26, 0 26.2
Carnitine (C0) −2.4 6.8 −24, 0 20.0
Acylcarnitine (C12-DC) 11.3 24.6 0, 88 31.4
Linoleoylcarnitine (C18:2) −3.0 7.5 −27, 0 22.8
Acylcarnitine (C5:1-DC) 2.9 7.4 0, 25 23.4
Glutarylcarnitine (C5-DC/C6-OH) 3.3 7.5 0, 28 27.2
Acylcarnitine (C6:1) 3.7 9.9 0, 34 23.5
Acylcarnitine (C7-DC) 2.39 6.99 0, 25 20.6
Octanoylcarnitine (C8) −3.0 8.8 −31, 0 20.6
Citrulline (Cit) −2.0 4.9 −17, 0 27.5
Glutamine (Gln) 3.7 7.8 0, 28 31.8
Histidine (His) 7.5 14.0 0, 48 36.2
Lysophosphatidylcholines (lysoPC a C16:0) −10.4 18.4 −62, 0 36.9
Lysophosphatidylcholines (lysoPC a C16:1) −2.9 6.4 −22, 0 26.2
Phosphatidylcholines (PC aa C36:0) 4.5 7.0 0, 23 44.1
Phosphatidylcholines (PC aa C42:2) 9.7 13.8 0, 46 51.6
Symmetric dimethylarginine (SDMA) 0.3 1.0 0, 3 22.5
Spermine −3.9 7.0 −24, 0 36.2
Tryptophan (Trp) 6.2 12.2 0, 41 35.1
Valine (Val) 3.5 8. 7 0, 31 24.4
  1. Results from LASSO regression of 21 metabolites selected for the serum metabolites classifier, five fold cross-validation, and bootstrap resampling (N = 1000) in the discovery cohort of patients with type 2 diabetes with microalbuminuria (n = 49)
  2. aThe relative frequency of the marker being included in the model across 1000 bootstrap resamples