Table 5.

Multinomial logistic regression coefficients of the one standard error (1-SE) cross-validated Lasso model for a reduced set of INAS items^a

INAS item	XPA	XPC	XPD	XPE	XPG	XPV
LL hyperreflexia	−0.74	0.11	0.42	−0.38	1.17	−0.58
UL hyporeflexia	1.11	−0.52	0.65	−0.25	0.22	−1.21
LL hyporeflexia	−0.14	−1.94	1.76	−0.55	−0.33	1.21
Babinski sign	−0.34	−0.78	0.28	−0.24	1.43	−0.35
Gait spasticity	0.46	0.00	−0.14	−0.08	−0.07	−0.17
LL spasticity	0.14	0.02	0.00	−0.04	−0.05	−0.07
Distal UL paresis	0.01	−0.32	−0.29	−0.10	0.28	0.43
LL rigidity	0.14	−0.04	−0.08	−0.01	0.01	−0.02

INAS = Inventory of Non-Ataxia Signs; LL = lower limb; UL = upper limb.

^aCoefficients are presented as log(odds). A positive coefficient (>0) indicates that the presence of an item increases the odds of being classified in a complementation group. A negative coefficient (<0) indicates that the presence of an item decreases the odds of being classified in a complementation group. A coefficient close to the null value represents that the frequency of a sign in a complementation group is close to the average frequency of the other groups. Coefficients above the |±0.4| threshold are in bold.