Table 3 Clinical expert’s rating in gold-label training and testing sets.

From: Development of a screening algorithm for borderline personality disorder using electronic health records

Rating category

Gold-label training set no. (%)

Gold-label testing set no. (%)

A: Not likely BoPD - primarily physical condition

37 (16%)

45 (20%)

B: Not likely BoPD - severe psychotic/substance abuse

28 (12%)

33 (14%)

C: Unsure - more information needed to make judgement

24 (11%)

19 (8%)

D: Possible BoPD

73 (32%)

70 (31%)

E: Most likely BoPD

66 (29%)

61 (27%)

- E1(a subset of E): Classic BoPD

12 (18%)

9 (15%)

Total

228 (100%)

228 (100%)