Alert fatigue
When a tool fires so many low-value alerts that clinicians start tuning them all out, including the important ones. A common reason a technically accurate model fails in practice.
A deterioration model pages the ward nurse at a threshold that produces about 40 pages a day across the medical wards. A chart review finds roughly four of those patients went on to deteriorate, a positive predictive value of 10%, and within weeks nurses acknowledge pages from the station without going to the bedside, including for those four. Before go-live, ask for expected pages per nurse per shift at your chosen threshold, and tune the threshold, repeat-page suppression and routing until staff will still act on that number.
Terms like this come up in real clinical scenarios across the HelloAI courses: bite-sized modules with verifiable certificates. An account takes one minute, no password needed.
Sign in →