In the context of predictive modeling, how does the sensitivity of a predictor relate to its specificity?

Sensitivity measures how often a predictor correctly flags positive cases, while specificity measures how often it correctly identifies negatives. When you adjust a model’s threshold to catch more positives, you usually increase sensitivity but lower specificity, and vice versa, because the same decision boundary is used for both types of errors. Thus, sensitivity and specificity are generally inversely related: improving one tends to hurt the other. For example, a spam filter set to catch every spam email will catch almost all spam (high sensitivity) but also mistakenly mark many legitimate emails as spam (low specificity). Understanding this trade‑off helps choose a threshold that balances false positives and false negatives for the problem at hand.