Table 1 Demographics and descriptive statistics.

From: Detection of eye contact with deep neural networks is as accurate as human experts

 

TD young children

TD young children

ASD young children

ASD young children

ASD child/adolescent

 

Total

Validation

Total

Validation

 

N

55

9

66

9

15

Males

36 (65%)

5 (56%)

55 (83%)

5 (56%)

12 (80%)

Age (months)

27.45 (5.84)

28.78 (5.87)

44.00 (11.11)

46.67 (12.25)

95.56 (32.60)

Score

     

  CSS total

N/A

N/A

7.97 (2.19)

7.5 (1.41)

7.43 (1.74)

  CSS SA

N/A

N/A

7.49 (2.30)

7.5 (1.51)

7.64 (1.74)

  CSS RRB

N/A

N/A

8.19 (1.76)

7.25 (1.83)

6.79 (2.83)

  CBCL total T score

43.16 (9.02)

45.43 (10.24)

58.05 (15.35)

56.13 (13.28)

59.00 (8.88)

Race

     

  White/Caucasian (%)

60

89

61

33

60

  Black/African American (%)

22

11

4

11

0

  Asian/Pacific Islander (%)

0

0

15

33

7

  More than one race (%)

13

0

14

11

26

  Other/unknown (%)

5

0

6

11

7

Hispanic/Latino ethnicity

4 (7%)

0 (0%)

15 (23%)

3 (33%)

3 (20%)

Maternal education

     

  Some high school

1

1

0

0

0

  High school diploma/GED

3

0

3

0

2

  Some college

8

0

4

1

0

  College/technical degree

24

4

28

3

5

  Graduate school degree

19

4

30

5

8

  Unknown

0

0

1

0

0

  1. Age, CSS, and CBCL scores expressed as mean (standard deviation). Higher scores indicate more significant impairment/problem behavior. For young children samples, total sample and its validation sub-sample are both reported.
  2. CSS total: calibrated severity score from ADOS; CSS SA: ADOS calibrated severity score for social affect; CSS RRB: ADOS calibrated severity score for restricted and repetitive behaviors; CBCL total T score: score for internalizing and externalizing problem behavior.