Table 2

Summary of the SVM classifications containing the means, standard deviations, and 95% confidence intervals (CI) of the accuracy (acc), specificity (spec), sensitivity (sens) running 100 repetitions of 10-fold cross-validations with linear kernel.

Comparison

Classification

Permutation test


Acc

Spec

Sens

Acc

Spec

Sens


AD vs. control

93.3% ± 4.6

CI:92.4-94.2%

95.1% ± 5.4

CI:94.1-96.2%

91.5% ± 5.8

CI:90.4-92.7%

50.7% ± 12.5

CI:48.2-53.1%

50.7% ± 13.3

CI:48.1-53.3%

50.7% ± 14.1

CI:47.9-53.4%


MCI vs. control

84.2% ± 3.7

CI:83.4-84.9%

81.1% ± 5.6

CI:80.0-82.2%

87.7% ± 3.7

CI:87.0-88.5%

51.3% ± 11.4

CI:49.0-53.5%

52.0% ± 12.2

CI:50.0-54.4%

50.4% ± 13.5

CI:47.8-53.1%


PSY vs. control

97.1% ± 1.6

CI:96.8-97.4%

95.3% ± 1.7

CI:95.0-95.6%

99.0% ± 2.4

CI:98.5-99.4%

48.7% ± 10.6

CI:46.7-50.8%

48.5% ± 12.4

CI:46.0-50.9%

49.0% ± 12.1

CI:46.6-50.8%


Other ND vs. control

82.8% ± 5.0

CI:81.8-83.7%

84.0% ± 5.8

CI:83.0-85.2%

81.4% ± 6.7

CI:80.1-82.7%

50.3% ± 10.3

CI:48.3-52.3%

50.7% ± 11.7

CI:48.4-53.0%

50.0% ± 12.0

CI:47.6-52.3%


NEURO vs. control

86.1% ± 5.7

CI:85.0-87.2%

88.7% ± 6.8

CI:87.3-90.0%

83.6% ± 6.6

CI:82.3-84.9%

49.9% ± 10.8

CI:47.8-52.1%

50.1% ± 11.5

CI:47.9-52.3%

49.8% ± 13.3

CI:47.2-52.3%


AD vs. MCI

75.6% ± 7.8

CI:74.1-77.2%

76.7% ± 8.3

CI:75.1-78.4%

74.6% ± 9.7

CI:72.7-76.5%

50.6% ± 9.4

CI:48.7-52.4%

51.2% ± 10.4

CI:49.1-53.2%

49.9% ± 11.7

CI:47.7-52.2%


AD vs. PSY

77.8% ± 4.0

CI:77.0-78.5%

76.3% ± 4.8

CI:75.4-77.3%

79.2% ± 5.4

CI:78.1-80.2%

50.0% ± 8.0

CI:48.5-51.6%

49.1% ± 9.3

CI:47.3-50.9%

51.1% ± 10.3

CI:49.1-53.1


AD vs. other ND

73.8% ± 4.4

CI:72.9-74.7%

75.2% ± 4.7

CI:74.2-76.1%

72.4% ± 6.4

CI:71.2-73.7%

50.1% ± 7.3

CI:48.7-51.5%

49.2% ± 9.4

CI:47.4-51.1%

51.0% ± 8.5

CI:49.3-52.7%


The right part of the table contains the results for the permuted class labels. PSY = psychological disorders (DEP, BD, SCHIZ), other ND = other neurodegenerative disorders (PD, MS, MCI), NEURO = PSY + other ND

Leidinger et al. Genome Biology 2013 14:R78   doi:10.1186/gb-2013-14-7-r78

Open Data