Machine learning model for identifying important clinical features for predicting remission in patients with rheumatoid arthritis treated with biologics

Table 3 Rankings of the important features for the prediction of remission in each biologic

Clinical feature	Average ranking	Adalimumab	Etanercept	Infliximab	Golimumab	Abatacept	Tocilizumab
ESR	4.0	− 0.041	− 0.155	− 0.514	− 0.227	− 0.356	− 0.104
Hemoglobin	5.3	+ 0.134	+ 0.071	+ 0.401	0	+ 0.656	+ 0.061
Age	6.7	− 0.250	− 0.040	− 0.230	− 0.033	^a0.241	− 0.097
DAS28-ESR	7.8	− 0.059	− 0.125	− 0.116	− 0.065	+ 0.114	− 0.100
Rheumatoid factor	8.0	+ 0.029	− 0.234	^a0.176	0	0.273^a	− 0.075
Anti-CCP antibody	8.3	− 0.052	− 0.029	− 0.285	− 0.070	− 0.347	+ 0.029
CRP	8.7	0	+ 0.102	+ 0.117	0	+ 0.350	+ 0.135
Disease duration	11.5	− 0.081	0	0.297^a	− 0.024	− 0.804	0
Methotrexate dose	11.7	0	0	0.046^a	0	+ 0.668	+ 0.049
Platelet	11.8	0	− 0.132	0	− 0.026	− 0.448	− 0.074
Cholesterol	12.5	− 0.052	− 0.042	− 0.254	0	0	− 0.080
ALT	15.0	− 0.081	0	+ 0.245	0	0	− 0.079
BUN	15.0	0	+ 0.032	0	0	− 0.403	+ 0.134
Triglyceride	16.0	− 0.036	− 0.023	0	0	0	− 0.067
ANA	16.0	0	0	− 0.060	0	− 0.430	0

^aA nonlinear relationship such as quadratic effect or mixed effect between drugs and variables. The average ranking was obtained by averaging the rankings of the 6 bDMARDs
ESR erythrocyte sedimentation rate, DAS28-ESR disease activity scores in 28 joints using the erythrocyte sedimentation rate, CRP C-reactive protein, ALT alanine aminotransferase, BUN blood urea nitrogen, ANA anti-nuclear antibody

ISSN: 1478-6362