Thresholds

A threshold is a value at which the interpretation or action associated with a model output changes.

Biological thresholds

Examples include \(R_0=1\) in simple epidemic theory, a minimum viable population, or a physiological concentration associated with a response.

Decision thresholds

If a model estimates event probability \(p\), an intervention rule may take the form

where \(p_c\) is chosen according to the costs and consequences of false alarms and missed events.

Uncertainty near a threshold

A point estimate just above or below a threshold can be misleading when uncertainty is substantial. The probability of crossing the threshold may be more informative.