Skip to main content

Table 3 Relevance of the highlighted text fragments among relevant sentencesa

From: Creating efficiencies in the extraction of data from randomized trials: a prospective evaluation of a machine learning and text mining tool

Report section

Data element

Relevant sentences, n Totalb

Fragments, n totalc

Relevant fragments, n (%)d

Exact matches, n (%)d

Partial matches, n (%)d

Meta information

Funding source

79

157

124 (79.0)

24 (15.3)

100 (63.7)

Funding number

35

54

44 (81.5)

27 (50.0)

17 (31.5)

Registration number

63

104

104 (100.0)

103 (99.0)

1 (1.0)

Enrollment

Eligibility criteria

110

0

0 (0.0)

0 (0.0)

0 (0.0)

Sample size

125

125

110 (88.0)

92 (73.6)

18 (14.4)

Enrollment start date

55

51

51 (100.0)

43 (84.3)

8 (15.7)

Enrollment end date

56

50

47 (94.0)

45 (90.0)

2 (4.0)

Early stopping

7

3

3 (100.0)

1 (33.3)

2 (66.7)

Intervention

Experimental arm(s)

123

133

74 (55.6)

15 (11.3)

59 (44.4)

Control arm(s)

121

62

55 (88.7)

34 (54.8)

21 (33.9)

Route of administration

32

34

30 (88.2)

27 (79.4)

3 (8.8)

Dose

50

77

70 (90.9)

26 (33.8)

44 (57.1)

Frequency of administration

45

61

55 (90.1)

44 (72.1)

11 (18.0)

Duration of treatment

57

56

48 (85.7)

35 (62.5)

13 (23.2)

Outcome

Primary outcome(s)

95

78

74 (94.9)

55 (70.5)

19 (24.4)

Primary outcome time point

76

86

78 (90.7)

28 (32.6)

50 (58.1)

Secondary outcome(s)

75

53

51 (96.2)

16 (30.2)

35 (66.0)

Secondary outcome time point

43

53

51 (96.2)

20 (37.7)

31 (58.5)

Summary measure

Median (IQR), n

57 (54)

59 (33)

53 (47 to 74)

27 (15 to 41)

18 (4 to 34)

Median (IQR), %

-

-

90.4 (86.2 to 96.9)

52.4 (32.8 to 73.2)

28.0 (14.7 to 57.9)

  1. aExaCT does not provide fragments for publication information. Data are shown for the remaining 18 data elements. Values in italics typeface fall at or below the limit of the lowest quartile
  2. bAcross all 75 trials, the number of relevant sentences among the 5 sentences reported within the solution for each data element
  3. cContained within sentences considered to be relevant by the human reviewers (column 3)
  4. dRelevant fragments of those contained within sentences considered to be relevant by the human reviewers (denominator, column 4)