Skip to main content
. 2025 May 8;46:101010. doi: 10.1016/j.bbih.2025.101010

Table 2.

Summary of the key predictor variables in the classification accuracy of future anxiety disorder by the random forest algorithm.


Variable
UK Biobank ID Direction of movement (in anxiety compared to control) Representative mean decrease Gini coefficient (relative importance)
1–5 years prospective anxiety All years prospective anxiety
Neuroticism score 20127 93.7 465.4
Number of self-reported non-cancer illnesses 135 31.4 122.2
Urea 30670 26.3 98.5
Total bilirubin 30840 24.7 112.8
GlycA 23480 24.7 96.6
Townsend deprivation index 22189 24.5 106.0
Alkaline phosphatase 30610 23.7 101.7
Histidine 23463 23.7 99.3
IGF-1 30770 23.6 96.7
Monocyte (%) 30190 23.2 96.5
C-reactive protein 30710 23.2 94.1
Red blood cell (erythrocyte) distribution width 30070 22.9 92.5
Urate 30880 22.6 100.3
Platelet count 30080 22.6 93.3
Creatinine (measured by enzymatic analysis) 30700 22.5 108.5
Body mass index (BMI) 21001 22.5 97.7
Platelet crit 30090 21.9 90.9
Haematocrit (%) 30030 21.8 95.1
Haemoglobin concentration 30020 21.7 99.4
Number of treatments/medications taken 137 21.5 111.4
Red blood cell (erythrocyte) count 30010 21.4 96.7
Creatinine 23478 21.3 104.4
Leucine 23466 21.2 90.7
Total concentration of branched-chain amino acids 23464 21.2 88.9
Overall health rating (%) 2178 10.7 46.4
Pain 6159 10.4 39.5
Long-standing illness, disability or infirmity 2188 9.9 45.5
Too sick for employment 6142 6.5 NA

Note: ↑ refers to an increase in anxious individuals compared to controls, ↓ refers to a decrease in anxious individuals compared to controls.