Table 8 Real-data scenario (MIMIC) for prediction of death within 48 h after ICU admission using baseline demographics, neurological assessments, vitals, and laboratory measurements within the first 2 h of admission: 6890 individuals, 17 continuous variables with a missing rate of up to 84%

From: Secure distributed multiple imputation enables missing data inference for private data proprietors

 

Final analysis

Performance

Technology

AUC

Accuracy

Discr.

Time (s)

Net (MB)

Python

0.888

0.786

6

1.701

N/A

PyMICE

0.892

0.783

N/A

3.907

N/A

SMC-MICE

0.886

0.780

5

285.450

2647

MHE-MICE

0.886

0.780

5

6440

88,920

  1. The bolded values denote the best result in each column.
Back to article page