Skip to main content

Table 7 Two-fold relevancy prediction in terms of WSS@95% on DERP systematic review collections

From: Improving reference prioritisation with PICO recognition

 

[84]

[119]

[95]

LR

PICO

Δ

ACE Inhibitors

52.3

73.3

*80.1

78.5

77.6

-0.9

ADHD

62.2

52.6

*79.3

75.5

74.5

-0.9

Antihistamines

14.9

*23.6

13.7

4.9

5.0

0.1

Atypical Antipsychotics

20.6

17.0

*25.1

19.9

20.9

1.0

Beta Blockers

36.7

46.5

42.8

*55.5

54.1

-1.4

Calcium Channel Blockers

23.4

43.0

*44.8

38.8

39.3

0.6

Estrogens

37.5

41.4

*47.1

41.0

43.7

2.7

NSAIDS

52.8

67.2

*73.0

65.3

66.5

1.2

Opioids

55.4

36.4

*82.6

53.3

57.0

3.7

Oral Hypoglycemic

8.5

*13.6

11.7

7.1

8.9

1.8

Proton Pump Inhibitors

22.9

32.8

*37.8

32.6

31.0

-1.6

Skeletal Muscle Relaxants

26.5

37.4

*55.6

40.1

45.3

5.3

Statins

31.5

*49.1

43.6

42.2

44.3

2.1

Triptans

27.4

34.6

41.2

40.6

*51.2

10.5

Urinary Incontinence

29.6

43.2

*53.0

52.4

52.4

0.0

Average

33.5

40.8

48.8

43.2

44.8

1.6

  1. Δindicates the change between adding the PICO features to the baseline logistic regression classifier (LR)
  2. *indicate best performance per review