Table 2 Coefficients (in original units) of the selected predictors of the most accurate models for each outcome.

From: Clinical predictive modelling of post-surgical recovery in individuals with cervical radiculopathy: a machine learning approach

Outcome—NDI

Outcome—EQ5D

Outcome—Neck pain

Outcome—Arm pain

Stepwise reg

Stepwise reg

Stepwise reg

LASSO

Predictor

Coef

Predictor

Coef

Predictor

Coef

Predictor

Coef

(Intercept)

19.650

(Intercept)

0.590

(Intercept)

24.710

(Intercept)

30.700

NDI

0.484

MSPQ

− 0.008

NDI

0.892

Vas_arm_worst

0.103

C6_touch_r.1

− 3.260

SES

0.002

C7_pin_r.1

− 14.340

NDI

0.245

C8_pin_r.1

− 3.450

AROM_F

− 0.004

Reflex_triceps_r.1

7.750

MSPQ

0.062

Reflex_ach_r.1

− 4.340

Sx.2

− 0.120

  

EQ5D

− 3.583

  

C7_pin_r.1

0.100

  

AROM_E

0.060

  

Strn_fingabd_r.1

0.110

  

AROM_RR

− 0.108

      

HRA_R

0.086

      

Handst_r

− 0.095

      

Romberg

0.040

      

Figure 8

0.100

      

CSQ_COP

1.155

      

C5_touch_r.1

− 0.860

      

C6_touch_r.1

− 13.040

      

C7_pin_r.1

− 0.200

  1. Reg regression, LASSO least absolute shrinkage and selection operator, Coef coefficient, NDI neck disability index, C6(C5)_touch_r.1 C6(C5) level light touch on right normal, C8(C7)_pin_r.1 C8(7) level pinprick on right normal, Reflex_ach (triceps)_r.1 Achilles (triceps brachii) muscle reflex on right normal, MSPQ modified somatic perception questionnaire, SES self efficacy scale, AROM_F(E/RR) cervical flexion(extension/right rotation) active range of motion, Sx.2 posterior cervical foraminotomy (PCF) with or without laminectomy, Strn_fingabd_r.1 strength of finger abductors on right normal, Vas_arm_worst worst arm pain intensity, EQ5D quality of life, HRA_R head reposition accuracy from right to neutral, Handst_r right hand grip strength, CSQ_COP coping strategies questionnaire, coping subscale.