Skip to main content

Table 5 Classification characteristics

From: Internal validation of modified Mirels’ scoring system for pathologic femur fractures

 

Internal validation dataset

Original MSTS dataset

Patient image source

Radiograph

CT Scan

Number of cases

7 Fracture | 47 No Fracture

8 Fracture | 40 No Fracture

Fracture prevalence (%)

12.9

16.7

 

OM Scoring

MM Scoring

OM Scoring

MM Scoring

ROC AUC

0.754 (0.639, 0.869)

0.878 (0.791, 0.965)

0.853 (0.777, 0.929)

0.941 (0.891, 0.991)

True positive rate (%)

85.7 (48.7,97.4)

100 (64.5, 100)

100 (67.5, 100)

87.5 (67.6, 100)

False positive rate (%)

55.4 (41.3, 68.6)

29.8 (18.7, 44.0)

55.0 (39.1, 69.3)

37.5 (22.2, 50.5)

PPV (%)

18.7 (8.9, 35.3)

33.3 (17.2, 54.6)

26.7 (14.2, 44.4)

34.7 (19.7, 57.0)

NPV (%)

95.4 (78.2, 99.2)

100 (89.5, 100)

100 (82.4, 100)

100 (87.1, 100)

Misclassification (%)

50.0

25.9

45.8

31.2

  1. Legend: Classification characteristics of impending pathological fracture using Mirels’ score of 9 or greater (9 +) as the cut-off threshold for current internal validation dataset and initial MSTS study test dataset. Results are shown for Original Mirels’ (OM) scoring and Modified Mirels’ (MM) scoring. Mean and (95% Confidence Interval) values are shown for each parameter