Table 1 Study population

From: AI-based differential diagnosis of dementia etiologies on multimodal data

Dataset (group)

Age (y), mean ± s.d.

Male, n (%)

Education (y), mean ± s.d.

Race (White, Black, Asian, American Indian, Pacific, multirace), n

CDR, mean ± s.d.

NACC

     

NC [n = 17,242]

71.25 ± 11.16

6,009, 34.85%

15.83 ± 2.98^

(13,266, 2541, 528, 109, 10, 575)^

0.05 ± 0.15

MCI [n = 7,582]

73.72 ± 9.81

3,615, 47.68%

15.16 ± 3.45^

(5,708, 1185, 231, 53, 5, 276)^

0.45 ± 0.18

AD [n = 16,131]

76.0 ± 10.31

7,234, 44.85%

14.52 ± 3.74^

(13,161, 1702, 354, 92, 10, 458)^

1.2 ± 0.73

LBD [n = 1,913]

75.01 ± 8.55

1,365, 71.35%

15.12 ± 3.63^

(1,659, 128, 39, 17, 0, 37)^

1.29 ± 0.78

VD [n = 1,919]

80.32 ± 8.76

947, 49.35%

14.15 ± 4.22^

(1,394, 332, 67, 2, 1, 68)^

1.22 ± 0.74

PRD [n = 114]

60.07 ± 10.36

62, 54.39%

14.8 ± 3.33^

(93, 5, 5, 0, 1, 1)^

1.95 ± 0.95

FTD [n = 2,898]

65.86 ± 9.36

1,603, 55.31%

15.45 ± 3.09^

(2,664, 69, 73, 4, 5, 39)^

1.2 ± 0.83

NPH [n = 138]

79.1 ± 9.24

69, 50.0%

15.0 ± 3.28^

(119, 10, 4, 0, 0, 4)^

1.18 ± 0.71

SEF [n = 808]

76.3 ± 11.15

413, 51.11%

14.6 ± 3.77^

(646, 95, 15, 5, 2, 31)^

1.11 ± 0.7

PSY [n = 2,700]

73.74 ± 10.78

1,102, 40.81%

14.13 ± 4.12^

(2,163, 238, 59, 14, 5, 87)^

1.1 ± 0.64

TBI [n = 265]

72.87 ± 11.23

192, 72.45%

14.42 ± 4.13^

(212, 27, 3, 2, 1, 11)^

1.11 ± 0.69

ODE [n = 1,234]

72.94 ± 12.14

654, 53.0%

14.5 ± 3.78^

(1,046, 93, 28, 5, 4, 36)^

1.2 ± 0.76

P value

< 1.0 × 10−200

<1.0 × 10−200

< 1.0 × 10−200

8.341 × 10−145

<1.0 × 10−200

NIFD

     

NC [n = 124]

63.21 ± 7.27

56, 45.16%

17.48 ± 1.87^

(89, 0, 0, 0, 0, 3)^

0.03 ± 0.12^

FTD [n = 129]

63.66 ± 7.33

75, 58.14%

16.18 ± 3.29^

(109, 1, 1, 0, 0, 4)^

0.82 ± 0.54^

P value

6.266 × 10−1

5.246 × 10−2

2.606 × 10−4

6.531 × 10−1

4.333 × 10−28

PPMI

     

NC [n = 171]

62.74 ± 10.12

109, 63.74%

15.82 ± 2.93

(163, 3, 2, 0, 0, 1)^

NA

MCI [n = 27]

68.04 ± 7.32

22, 81.48%

15.52 ± 3.08

(24, 1, 1, 0, 0, 1)

NA

P value

1.006 × 10−2

1.115 × 10−1

6.194 × 10−1

2.910 × 10−1

NA

AIBL

     

NC [n = 480]

72.45 ± 6.22

203, 42.29%

NA

NA

0.03 ± 0.12

MCI [n = 102]

74.73 ± 7.11

53, 51.96%

NA

NA

0.47 ± 0.14

AD [n = 79]

73.34 ± 7.77

33, 41.77%

NA

NA

0.93 ± 0.54

P value

5.521 × 10−3

1.887 × 10−1

NA

NA

4.542 × 10−158

OASIS

     

NC [n = 424]

71.34 ± 9.43

164, 38.68%

15.79 ± 2.62^

(53, 18, 1, 0, 0, 0)^

0.0 ± 0.02

MCI [n = 27]

75.04 ± 7.25

14, 51.85%

15.19 ± 2.76

(4, 1, 0, 0, 0, 0)^

0.52 ± 0.09

AD [n = 32]

77.44 ± 7.42

20, 62.5%

15.19 ± 2.8

(8, 1, 0, 0, 0, 0)^

0.86 ± 0.44

LBD [n = 4]

74.75 ± 5.67

4, 100.0%

16.0 ± 2.83

NA

1.0 ± 0.0

FTD [n = 4]

64.25 ± 8.61

3, 75.0%

16.5 ± 2.96

(4, 0, 0, 0, 0, 0)

1.25 ± 0.75

P value

7.789 × 10−4

3.239 × 10−3

5.507 × 10−1

8.735 × 10−1

2.855 × 10−169

LBDSU

     

NC [n = 134]

68.77 ± 7.62

61, 45.52%

17.27 ± 2.47^

NA

NA

MCI [n = 35]

70.16 ± 8.41

26, 74.29%

16.6 ± 2.58

NA

NA

LBD [n = 13]

73.42 ± 7.81

8, 61.54%

16.77 ± 2.15

NA

NA

P value

1.033 × 10−1

7.863 × 10−3

3.243 × 10−1

NA

NA

4RTNI

     

NC [n = 12]

68.08 ± 4.92

5, 41.67%

15.45 ± 2.57^

(12, 0, 0, 0, 0, 0)

0.0 ± 0.0

MCI [n = 31]

67.61 ± 7.0

11, 35.48%

16.68 ± 4.02

(25, 1, 2, 0, 1, 1)^

0.55 ± 0.15

FTD [n = 37]

69.14 ± 7.43

20, 54.05%

16.46 ± 4.21

(31, 1, 0, 0, 1, 2)^

1.27 ± 0.55

P value

6.691 × 10−1

2.992 × 10−1

6.843 × 10−1

7.620 × 10−1

5.700 × 10−16

ADNI

     

NC [n = 868]

72.7 ± 6.57

383, 44.12%

16.51 ± 2.52

(730, 92, 28, 2, 0, 12)^

0.0 ± 0.04^

MCI [n = 1119]

72.77 ± 7.65

648, 57.91%

15.97 ± 2.75

(1,023, 56, 17, 2, 2, 13)^

0.5 ± 0.06

AD [n = 417]

74.99 ± 7.78

232, 55.64%

15.25 ± 2.92

(383, 20, 10, 0, 0, 4)

0.77 ± 0.27

P value

8.911 × 10−8

3.090 × 10-09

2.869 × 10−14

2.828 × 10−5

<1.0 × 10−200

FHS

    

*

NC [n = 394]

74.9 ± 10.22^

206, 52.28%

NA

(394, 0, 0, 0, 0, 0)

0.0 ± 0.0

MCI [n = 434]

79.92 ± 8.8^

203, 46.77%

NA

(434, 0, 0, 0, 0, 0)

0.49 ± 0.07

AD [n = 687]

82.99 ± 7.87^

211, 30.71%

NA

(687, 0, 0, 0, 0, 0)

2.04 ± 0.88

LBD [n = 73]

79.34 ± 9.37^

46, 63.01%

NA

(73, 0, 0, 0, 0, 0)

1.84 ± 0.84

VD [n = 113]

81.74 ± 7.3^

48, 42.48%

NA

(113, 0, 0, 0, 0, 0)

1.85 ± 0.8

FTD [n = 8]

85.67 ± 5.91^

4, 50.0%

NA

(8, 0, 0, 0, 0, 0)

2.0 ± 0.87

P value

1.316 × 10−31

7.905 × 10−14

NA

1.0

<1.0 × 10−200

  1. Nine independent datasets were used for this study, including ADNI, NACC, NIFD, PPMI, OASIS, LBDSU, 4RTNI and FHS. Data from NACC, NIFD, PPMI, OASIS, LBDSU and 4RTNI were used for model training. Data from ADNI, FHS and a held-out set from NACC were used for model testing. The P value for each dataset indicates the statistical significance of intergroup differences per column. We used one-way analysis of variance (ANOVA) and two-sided χ2 tests for continuous and categorical variables, respectively. Please refer to Glossary 1 for more information on the acronyms. NA, not available. Due to the absence of CDR scores in the FHS dataset, we used the following definition: 0.0, NC; 0.5, cognitive impairment; 1.0, mild dementia; 2.0, moderate dementia; 3.0, severe dementia. The symbol ^ indicates that data was not available for some subjects.