Back

Model uncertainty quantification: A post hoc calibration approach for heart disease prediction

2025-09-30 cardiovascular medicine Title + abstract only
View on medRxiv
Show abstract

We investigate whether post-hoc calibration improves the clinical trustworthiness of heart-disease predictions beyond conventional accuracy metrics. Using a structured clinical dataset (1,025 records; 85/15 train-test split), we benchmarked six classifiers logistic regression, SVM, k-nearest neighbors, naive Bayes, random forest, and XGBoost on accuracy, ROC-AUC, precision, recall, and F1, and then evaluated probability quality before and after Platt (sigmoid) and isotonic calibration using Brie...

Predicted journal destinations

1
European Heart Journal - Digital Health
15 training papers
#1 17.8%
2
Scientific Reports
701 training papers
Top 12% 11.9%
3
Circulation
37 training papers
Top 1% 9.0%
4
npj Digital Medicine
85 training papers
Top 2% 7.8%
5
Nature Communications
483 training papers
Top 18% 6.0%
6
PLOS ONE
1737 training papers
Top 82% 5.1%
7
Journal of the American Heart Association
92 training papers
Top 6% 5.1%
8
Frontiers in Cardiovascular Medicine
33 training papers
Top 5% 4.2%
9
Circulation: Genomic and Precision Medicine
30 training papers
Top 3% 3.9%
10
eLife
262 training papers
Top 16% 2.8%
11
Computers in Biology and Medicine
39 training papers
Top 3% 1.9%
12
Open Heart
18 training papers
Top 4% 1.9%
13
Heart Rhythm
16 training papers
Top 4% 1.6%
14
The American Journal of Cardiology
15 training papers
Top 5% 1.6%
15
PLOS Digital Health
88 training papers
Top 13% 1.3%
16
Hypertension
20 training papers
Top 5% 1.2%
17
BMJ Open
553 training papers
Top 62% 1.0%
18
Frontiers in Physiology
18 training papers
Top 3% 1.0%
19
Communications Medicine
63 training papers
Top 7% 0.7%
20
EBioMedicine
21 training papers
Top 1% 0.7%
21
BMC Medical Informatics and Decision Making
36 training papers
Top 7% 0.7%
22
Nature Medicine
88 training papers
Top 17% 0.7%
23
Nature Genetics
72 training papers
Top 10% 0.7%
24
PLOS Computational Biology
141 training papers
Top 14% 0.7%
25
BMC Medicine
155 training papers
Top 34% 0.7%
26
Atherosclerosis
16 training papers
Top 5% 0.7%
27
The Lancet Digital Health
25 training papers
Top 3% 0.5%