Skip to main content

Table 3 Predictive models for the discharge clinical outcomes, including coefficients of each predictor and model goodness-of-fit (R2, \( {R}_{adj}^2 \), MAE, and MAEn)

From: Inpatient stroke rehabilitation: prediction of clinical outcomes using a machine-learning approach

  Predictive equation for clinical outcomes at discharge Study patients (N = 50) New patients
(LOOCV)
R2\( \left({\boldsymbol{R}}_{\boldsymbol{adj}}^{\mathbf{2}}\right) \) MAE (MAEn) MAE (MAEn)
FIM 60.14 + 2.23 TMWT + 0.35 FIM + 0.5 BBS − 0.24 age − 0.02 TSA + 0.8 EDU − 1.71 LI − 0.05 BMI − 10.6 HM − 3.76 SI 0.76
(0.70)
7.6
(0.10)
10.2
(0.13)
TMWT (m/s) −0.16 + 0.7 TMWT + 0.01 FIM − 0.003 TSA + 0.02 EDU + 0.44 HIS − 0.15 LI 0.70
(0.66)
0.26
(0.13)
0.3
(0.15)
SMWT (m) 190.83 + 101.72 TMWT + 1.03 BBS + 0.54 SMWT − 2.21 age 0.70
(0.67)
73.2
(0.13)
80.8
(0.14)
BBS 13.27 + 10.1 TMWT + 0.33 FIM + 0.21 BBS − 0.24 age − 0.08 TSA + 0.42 EDU − 5.57 WHT − 1.96 LI 0.77
(0.73)
6.4
(0.11)
7.4
(0.14)
  1. MAE Mean Absolute Error, TMWT, FIM, BBS, SMWT clinical test scores at admission, TSA time from stroke to admission, EDU education in years, BMI Body Mass Index. The following variables are binary and receive the value of 1 or 0: HM hemorrhagic stroke, HIS Hispanic, WHT White, SI speech impairment, LI language impairment. The “New patients” MAE is the averaged error across all left-out subjects during the LOOCV procedure