Decision Curve Analysis

Preview in your theme

status="ready" · three biopsy strategies over a simulated validation cohort of 1,800 men, 407 with disease. The tinted band is the only stretch that matters — where a strategy clears both biopsy-all and biopsy-none by at least 0.002 net benefit. Point at the plot, or Tab to it and use the arrow keys: the legend names whichever line is highest at that threshold, and below 5% that is still "biopsy all". The PSA rule is a single cut-off, so its curve bends with the odds alone and sinks under the references long before the models do.

Biopsy decision · external validation cohort

prevalence 22.6% · 3 strategies · adds value 4.6% to 40%

net benefit at threshold 20%

Risk model + MRI0.102leads here
Clinical baseline0.079
PSA > 4 ng/mL rule0.075
Biopsy all0.033
Biopsy none0.000

Net benefit is true positives per patient, after charging every false positive at the odds of the threshold: at 20% the reader is accepting 4 unnecessary biopsy decisions to catch one more event. A strategy earns its keep only where it clears both biopsy all and biopsy none by at least 0.002 — one extra case per 500 patients — here, 4.6% to 40%. Point at the plot, or Tab to it and use the arrow keys, to read any threshold. Biopsy all leaves the frame at 33% and keeps falling; the axis clips it rather than flattening every other curve.

Decision curve: net benefit against threshold probability for 3 strategies, 1% to 40%. The references are derived from the prevalence, 22.6%: biopsy none is zero everywhere, biopsy all falls from 0.218 and crosses zero at 23%. A strategy beats both references from 4.6% to 40%. Exact values follow in the data table. Up to 14 thresholds, evenly spaced by rank.
Threshold	Risk model + MRI	Clinical baseline	PSA > 4 ng/mL rule	Biopsy all	Leads
1%	0.218	0.218	0.180	0.218	Biopsy all
4%	0.195	0.193	0.166	0.194	Biopsy all
6%	0.182	0.176	0.157	0.177	Risk model + MRI
10%	0.152	0.150	0.136	0.140	Risk model + MRI
12%	0.139	0.135	0.125	0.121	Risk model + MRI
16%	0.117	0.109	0.101	0.079	Risk model + MRI
18%	0.108	0.096	0.089	0.056	Risk model + MRI
22%	0.098	0.063	0.061	0.008	Risk model + MRI
24%	0.087	0.054	0.047	-0.018	Risk model + MRI
28%	0.065	0.039	0.015	-0.075	Risk model + MRI
30%	0.060	0.036	-0.003	-0.106	Risk model + MRI
34%	0.050	0.030	-0.041	-0.173	Risk model + MRI
36%	0.044	0.026	-0.061	-0.209	Risk model + MRI
40%	0.039	0.019	-0.107	-0.290	Risk model + MRI

status="empty" · a valid contract with no strategies at all, worded so it cannot be mistaken for a failed fetch — the prevalence is known, so both references are ready; the curves are not.

Biopsy decision · external validation cohort

Net benefit against threshold probability, read against biopsy all and biopsy none

No thresholds to plot

A decision curve needs net benefit measured across a sweep of threshold probabilities — not a single operating point, and not accuracy. The biopsy all and biopsy none references come from the prevalence alone, so they are ready as soon as one strategy is.

status="loading" · a skeleton holding the plot box, a rising model silhouette against the sloping treat-all reference and its zero line, aria-hidden, honouring prefers-reduced-motion, with no timer involved.

Biopsy decision · external validation cohort

Net benefit against threshold probability, read against biopsy all and biopsy none

status="error" · the fetch failed. Try again only appears when onRetry is passed, and clicking it flips the card to ready.

Biopsy decision · external validation cohort

Net benefit against threshold probability, read against biopsy all and biopsy none

Couldn't load this decision curve

The data source didn't respond.

Loading preview…

This component is part of Pro

The install command, prompt and concept map are included with a Pro license — one purchase unlocks every Pro component, forever.

Unlock with Pro