Table 6 Risk stratification outcomes and screening efficiency metrics.

From: Development and validation of a machine learning model for non-contact injury prediction based on lower limb strength asymmetry in professional football

Risk Category

Probability Range

Players n (%)

Injuries n (%)

Injury Rate

NNS*

Prevented/100†

Intervention

Low

< 15%

142 (45.5)

3 (3.4)

2.1%

42.3

2.4

Standard monitoring

Moderate

15–30%

98 (31.4)

19 (21.3)

19.4%

18.7

5.3

Preventive program

High

30–50%

54 (17.3)

42 (47.2)

77.8%

7.8

12.8

Modified training

Very High

> 50%

18 (5.8)

25 (28.1)

138.9%‡

4.2

23.8

Intensive intervention

Total

-

312 (100)

89 (100)

28.5%

12.4

8.1

-

Risk-Based Metrics

       

High + Very High

> 30%

72 (23.1)

67 (75.3)

93.1%

5.2

19.2

Combined approach

Sensitivity at 30% threshold

-

-

75.3%

-

-

-

-

Specificity at 30% threshold

-

-

84.3%

-

-

-

-

PPV at 30% threshold

-

-

67.4%

-

-

-

-

NPV at 30% threshold

-

-

88.9%

-

-

-

-

  1. *NNS: Number needed to screen to prevent one injury. †Injuries prevented per 100 players screened. ‡Rate > 100% indicates multiple injuries per player.