Confounder
In causal inference, a variable that affects both the intervention and the outcome, and that may cause an observed association to not reflect a true causal relationship.
Timeline (last 90 days)
A simple 'pre-implementation vs post-implementation' comparison cannot reliably isolate the contribution of the AI itself, because outcome changes may stem from various confounders such as changes in the number of healthcare staff, workflow redesign, or changes in patient visit volume.
Pitfalls that causal evaluation of medical AI needs to guard against include: confounders, selection effects, implementation fidelity, unintended consequences, and differences between patient populations.
A confounding variable is a third-party variable that simultaneously affects both the independent and dependent variables, leading to the erroneous conclusion that a causal relationship exists between the two variables.
All Facts (3)
A simple 'pre-implementation vs post-implementation' comparison cannot reliably isolate the contribution of the AI itself, because outcome changes may stem from various confounders such as changes in the number of healthcare staff, workflow redesign, or changes in patient visit volume.
50%UnverifiedPitfalls that causal evaluation of medical AI needs to guard against include: confounders, selection effects, implementation fidelity, unintended consequences, and differences between patient populations.
50%UnverifiedA confounding variable is a third-party variable that simultaneously affects both the independent and dependent variables, leading to the erroneous conclusion that a causal relationship exists between the two variables.
50%