Back

Act or Defer: Error-Controlled Decision Policies for Medical Foundation Models

2026-02-26 health informatics Title + abstract only
View on medRxiv
Show abstract

Clinical deployment of foundation models requires decision policies that operate under explicit error budgets, such as a cap on false-positive clinical calls. Strong average accuracy alone does not guarantee safety: errors can concentrate among patients selected for action, leading to harm and inefficient use of healthcare resources. Here we introduce SO_SCPLOWTRATC_SCPLOWCP, a stratified conformal framework that turns foundation model predictions into decision-ready outputs through error-contro...

Predicted journal destinations

1
npj Digital Medicine
85 training papers
Top 0.5% 14.7%
2
Journal of the American Medical Informatics Association
53 training papers
Top 1% 11.9%
3
Nature Communications
483 training papers
Top 7% 11.9%
4
Scientific Reports
701 training papers
Top 27% 7.8%
5
JAMIA Open
35 training papers
Top 1% 6.4%
6
Journal of Biomedical Informatics
37 training papers
Top 1% 6.4%
7
PLOS Digital Health
88 training papers
Top 4% 6.0%
8
BMC Medical Informatics and Decision Making
36 training papers
Top 3% 5.0%
9
Journal of Medical Internet Research
81 training papers
Top 6% 2.8%
10
Nature Medicine
88 training papers
Top 2% 1.9%
11
Computers in Biology and Medicine
39 training papers
Top 4% 1.9%
12
PLOS Computational Biology
141 training papers
Top 7% 1.9%
13
PLOS ONE
1737 training papers
Top 95% 1.6%
14
JMIR Medical Informatics
16 training papers
Top 4% 1.5%
15
BMC Medical Research Methodology
41 training papers
Top 4% 1.3%
16
Communications Medicine
63 training papers
Top 2% 1.3%
17
International Journal of Medical Informatics
25 training papers
Top 5% 1.3%
18
Cancers
57 training papers
Top 7% 1.0%
19
Bioinformatics
24 training papers
Top 2% 1.0%
20
Patterns
15 training papers
Top 2% 1.0%
21
Frontiers in Digital Health
18 training papers
Top 4% 0.7%
22
International Journal of Molecular Sciences
39 training papers
Top 7% 0.7%
23
JAMA Network Open
125 training papers
Top 30% 0.7%
24
Statistics in Medicine
17 training papers
Top 0.4% 0.7%