Table 1 Cooney’s criteria for evaluating clinically useful risk assessment algorithms

1

Appropriateness of statistical methods used to derive the function.

• Representativeness of the algorithm’s derivation sample, optimal statistical power and methods, and clarity of end point predicted by the function.

2

Performance of the function: internal and external validity.

• Discrimination, calibration, and sensitivity of the algorithm(s) in the derivation and external datasets.

3

Usability of the algorithm.

• Impact of an algorithm’s format on its use and uptake in clinical settings.

4

Inclusion of appropriate risk factors.

• Inclusion of major risk factors known to be prevalent in the target population.

5

Measurable health gains associated with the use of the algorithm(s).

• Tangible clinical benefits associated with use of the algorithm(s).