Skip to main content

Table 1 Dataset. For each feature, the type either static (S) or dynamic (D) is defined. For the continuous and ordinal features, percentage of native missing values and inter-quartile range (IQR) values at 25%, 50% and 75% are reported; for the categorical features, levels and corresponding percentage of instances are reported; for the NIV and PEG variables, we reported the total number of patients who were administered these interventions

From: Exploiting mutual information for the imputation of static and dynamic mixed-type clinical data with an adaptive k-nearest neighbours approach

Continuous features   Categorical features
Feature Type % NA IQR   Feature Type Levels %
BMI premorbid [kg/m2] S 2.08 23/25/28   sex S Female 47.6
BMI diagnosis [kg/m2] S 0.91 22/24/27     Male 52.4
FVC diagnosis [%] S 4.12 83/98/108     NA 0
age at onset [years] S 0 56/64/70   familiality S No 91.4
diagnostic delay [months] S 0 5/9/14     Yes 8.1
onset delta [months] S 0 -18/-11/-6     NA 0.5
      genetics S C9orf72 7.1
        FUS 0.3
        SOD1 1.4
        TARDBP 1.6
Ordinal features     wild type 83.6
Feature Type % NA IQR     NA 6.0
ALSFRS-R 1 D 0 2/3/4   FTD S No 53.0
ALSFRS-R 2 D 0 3/4/4     Yes 13.0
ALSFRS-R 3 D 0 2/3/4     NA 34.0
ALSFRS-R 4 D 0 2/3/4   onset site S Bulbar 34.4
ALSFRS-R 5 D 0 1/2/3     Limb 65.6
ALSFRS-R 6 D 0 1/2/3     NA 0
ALSFRS-R 7 D 0 1/3/3   NIV D No 59.6
ALSFRS-R 8 D 0 2/2/3     Yes 40.4
ALSFRS-R 9 D 0 0/1/3     NA 0
ALSFRS-R 10 D 0 3/4/4   PEG D No 31.9
ALSFRS-R 11 D 0 3/4/4     Yes 25.0
ALSFRS-R 12 D 0 4/4/4     NA 43.1