Skip to main content

Table 2 nRMSD scores for the continuous features in the training set. The best performances are highlighted in bold

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

Features Imputation methods
  Amelia II MICE missForest k-RN wk-NN k-RN wk-NN MI
     k=10 k=10 k=20 k=20
BMI premorbid 0.1012 0.0960 0.1323 0.1634 0.1286 0.1617 0.0731
BMI diagnosis 0.1560 0.1069 0.1476 0.1750 0.1457 0.1687 0.0965
FVC diagnosis 0.2466 0.2463 0.2534 0.1970 0.1876 0.1953 0.1839
age at onset 0.2355 0.2362 0.2393 0.1855 0.1748 0.1820 0.1735
diagnostic delay 0.1150 0.1218 0.1316 0.1484 0.1282 0.1495 0.0850
onset delta 0.1362 0.1362 0.1665 0.1848 0.1584 0.1778 0.1049
Average 0.1651 0.1572 0.1784 0.1757 0.1539 0.1725 0.1195