Threshold-Aware Machine Learning for Heart Disease Prediction

DOI

10.2991/978-94-6239-648-7_52How to use a DOI?

Keywords

Heart-disease Prediction; Clinical Risk Stratification; SVM; Random Forest; Model Calibration

Abstract

The decision threshold in clinical heart disease risk assessment is critical for real-world utility but frequently overlooked. This study investigates threshold-aware prediction using a clinical dataset of 919 records (14 features), with heart disease presence (num > 0) as the target. Preprocessing involved median/mode imputation, StandardScaler for numerical features, and one-hot encoding for categorical features. A stratified 80/20 split preserved the original class distribution (55.3% positive). This research compares four models: L2-logistic regression, Radial Basis Function (RBF)-kernel Support Vector Machine (SVM), random forest (300 trees), and a scikit-learn Multilayer Perceptron (MLP) (64-32-16), using class weights for mild imbalance. A multi-dimensional evaluation was conducted, reporting standard metrics (Accuracy, F1, ROC-AUC, Brier score) at the default 0.5 threshold, alongside ROC/PR curves, calibration plots, and threshold sensitivity analyses. Results indicate random forest achieved slightly better ROC-AUC and Brier scores, while threshold-tuned SVM yielded the highest F1. Feature importance analysis identified clinically relevant variables and potential “center” (site) effects. The study concludes that for deployment in screening scenarios, probability calibration quality and deliberate threshold selection are decisive when primary performance metrics are comparable.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Yanzhou Qian
PY  - 2026
DA  - 2026/04/24
TI  - Threshold-Aware Machine Learning for Heart Disease Prediction
BT  - Proceedings of the International Workshop on Advances in Deep Learning for Image Analysis and Computer Vision (IWADIC 2025)
PB  - Atlantis Press
SP  - 470
EP  - 479
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6239-648-7_52
DO  - 10.2991/978-94-6239-648-7_52
ID  - Qian2026
ER  -