Table 4

Variable importance on random forest models and standard partial regression coefficient on multiple logistic regression models

RF model 1RF model 2RF model 3RF model 4RF model 5RF model 6Total
VariablesVI*VariablesVI*VariablesVI*VariablesVI*VariablesVI*VariablesVI*VariablesVI*
1HbA1c_dif100HbA1c_dif100HbA1c_dif100HbA1c_dif100HbA1c100HbA1c100HbA1c_dif100
2HbA1c66.8HbA1c41.6FBG53.3HbA1c96.6HbA1c_dif98.2FBG81.0HbA1c97.0
3RF_dif27.5MCV_dif28.7HbA1c50.8FBG87.2FBG81.9HbA1c_dif78.5FBG79.1
4A/G_dif27.0FBG28.4FBG_dif40.2FBG_dif65.0FBG_dif80.1FBG_dif62.7FBG_dif59.5
5MCV_dif25.2RF_dif25.7MCV_dif35.1ALP43.2ALP43.2Weight50.2Weight42.2
6P-FEV123.3MCHC_dif25.1CRP_dif34.1TC43.1Weight41.7ALP42.9ALP41.2
7P-FEV1_dif23.2FBG_dif25.0PLAT33.3weight43.0PLAT40.0TG42.4PLAT38.7
8GTP22.7P-FEV1_dif25.0FVC33.3CRP_dif42.5weight_dif38.4AST40.8CRP_dif38.1
9P-FVC_dif22.4P-FVC_dif24.6P-FEV132.6PLAT41.8CK_dif37.4weight_dif39.3ALP_dif38.0
10CRP_dif22.2CRP_dif24.4ALP_dif32.6ChE41.6ALP_dif37.2ALP_dif38.3TG37.7
MLR model 1MLR model 2MLR model 3MLR model 4MLR model 5MLR model 6Total
VariablesSRC*VariablesSRC*VariablesSRC*VariablesSRC*VariablesSRC*VariablesSRC*VariablesSRC*
1MCH100.0HbA1c_dif100.0MCH100.0MCH100.0MCH100.0Ht100.0MCH100.0
2MCV82.5HbA1c84.6MCHC53.2MCV61.9MCV64.7Hb80.0MCV72.6
3MCHC34.5FBG81.3MCV52.7MCHC46.2MCHC46.2Hb_dif32.9MCHC43.6
4HbA1c33.0TC75.7Ht46.5RCC30.0FEV133.6Ht_dif30.9Ht43.2
5Ht_dif32.9ALB_dif71.4RCC43.7Ht29.1MCH_dif33.0RCC29.6HbA1c_dif39.1
6TC30.7MCH67.3FVC42.8Hb_dif24.6FVC29.5MCHC23.6HbA1c33.6
7LDL-C26.2LDL-C63.3MCH_dif41.5RCC_dif20.5MCV_dif28.3MCV18.8FBG32.6
8HbA1c_dif24.7MCV58.4MCV_dif39.4MCV_dif14.8MCHC_dif25.9MCHC_dif14.9TC29.5
9FBG20.6HDL-C51.1HbA1c_dif37.3HbA1c_dif12.8Ht24.4HbA1c8.2RCC27.1
10HDL-C19.4A/G_dif43.6FBG31.9Cl10.6RCC23.1TC6.7LDL-C25.1
  • *Variable importance and standard partial regression coefficient are expressed as percentages. We set the degree of most influential variable in each model as 100% (SRC is converted to an absolute value because it can take a negative value).

  • A/G, albumin/globulin ratio; CRP, C reactive protein; dif, Change value from the previous year ; FBG, fasting blood glucose; FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity; GTP, glutamyl transpeptidase; Hb, haemoglobin; HbA1c, glycated haemoglobin A1c; HDL-C, high-density lipoprotein cholesterol; Ht, haematocrit; MCHC, mean corpuscular haemoglobin concentration; MLR, Multiple Logistic Regression; PLAT, platelet; RCC, red cell count; RF, Random Forest; SPRC, Standard Regression Coefficient; TC, total cholesterol; VI, Variable Importance.