Skip to main content
. 2018 Dec 21;1(8):e186040. doi: 10.1001/jamanetworkopen.2018.6040

Table 3. Percentage of Correctly Reidentified Matches at Different Time Resolutions of Partial Aggregation of Physical Activity Data.

Machine Learning Algorithm Correctly Reidentified Matches, %
15 min 20 min 30 min 1 h 2 h 4 h 6 h 8 h 12 h 24 h
Adults in NHANES 2003-2004
Demographics onlya 81.1b
Physical activity
Linear SVMc 0 0 0 0.02 0 0.04 0 0.06 0.02 0.02
Random foresta 5.97 5.93 6.52 5.61 4.24 2.14 1.14 0.74 0.23 0.23
Demographics and physical activity
Linear SVMa 85.2 85.6 87.0 86.9 87.7 87.4 86.8 86.7 85.7 84.8
Random foresta 94.3 94.9 94.5 94.0 93.0 91.7 91.3 90.6 89.0 87.0
Adults in NHANES 2005-2006
Demographics onlya 80.3b
Physical activity
Linear SVMc 0 0.02 0.02 0 0.02 0 0.02 0.02 0.02 0.02
Random forest 6.40a 6.48a 6.55a 6.04a 4.55a 2.29a 1.09a 0.55a 0.23a 0.08b
Demographics and physical activity
Linear SVMa 84.5 84.8 84.7 85.1 86.3 86.1 85.9 85.5 85.0 83.1
Random foresta 93.5 93.8 93.2 92.8 91.9 91.0 90.5 89.4 87.9 85.8
Children in NHANES 2003-2004
Demographics onlya 61.6b
Physical activity
Linear SVMc 0.08 0 0.04 0.04 0 0.04 0 0.08 0.04 0.04
Random Forest 11.1a 10.5a 11.0a 7.83a 4.45a 1.98a 0.95a 0.58a 0.37a 0.04b
Demographics and physical activity
Linear SVMa 70.3 69.8 70.8 71.5 68.9 69.9 67.0 68.6 67.5 64.3
Random foresta 87.2 87.4 87.1 84.8 83.1 80.0 78.6 76.9 73.8 70.2
Children in NHANES 2005-2006
Demographics onlya 59.4b
Physical activity
Linear SVMc 0.08 0 0.04 0.08 0 0.08 0.04 0 0.08 0
Random foresta 10.6 11.5 10.0 7.17 4.25 1.81 1.02 0.91 0.32 0.32
Demographics and physical activity
Linear SVMa 67.2 67.2 67.0 67.2 66.0 67.0 66.4 67.0 64.7 62.4
Random foresta 84.8 85.5 84.4 82.6 80.3 78.2 75.9 74.5 70.9 67.3

Abbreviations: NHANES, National Health and Nutrition Examination Survey; SVM, support vector machine.

a

P < .001.

b

The demographics-only results are not subject to any time; thus, the same numbers are independent of the time columns

c

P > .99.