Comparison of Prediction Models for Heart Failure Related Data

Haoyang Zhang

2025

Abstract

Using the Heart Failure Clinic Records Dataset from Kaggle, this study assesses how well three machine learning models-Logistic Regression, Random Forest, and K-Nearest Neighbours-predict mortality events associated to heart failure. The dataset includes 299 patients with 12 clinical features such as ejection fraction, serum creatinine, platelet counts, and smoking history. To guarantee reliable model training, data pretreatment addressed outliers, missing values, and scalability concerns. For feature selection, Principal Component Analysis (PCA) was employed to reduce dimensionality while preserving crucial data. The model's performance was assessed using metrics of accuracy, precision, recall, and F1 score; cross-validation was employed to ensure generalizability. According to the results, the Random Forest model outperforms K-Nearest Neighbours (0.786) and Logistic Regression (0.812) by achieving the best accuracy of 0.907. The Random Forest also shows superior precision (0.92) and recall (0.89), effectively balancing false positives and negatives. The promise of machine learning in predictive healthcare is demonstrated by this work, especially in identifying high-risk individuals for early intervention. The results highlight how well ensemble techniques like Random Forest handle complicated clinical data and offer guidance for incorporating machine learning into future studies and clinical practices.

Paper Citation

in Harvard Style

Zhang H. (2025). Comparison of Prediction Models for Heart Failure Related Data. In Proceedings of the 2nd International Conference on Innovations in Applied Mathematics, Physics, and Astronomy - Volume 1: IAMPA; ISBN 978-989-758-774-0, SciTePress, pages 324-330. DOI: 10.5220/0013825200004708

in Bibtex Style

@conference{iampa25,
author={Haoyang Zhang},
title={Comparison of Prediction Models for Heart Failure Related Data},
booktitle={Proceedings of the 2nd International Conference on Innovations in Applied Mathematics, Physics, and Astronomy - Volume 1: IAMPA},
year={2025},
pages={324-330},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013825200004708},
isbn={978-989-758-774-0},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Innovations in Applied Mathematics, Physics, and Astronomy - Volume 1: IAMPA
TI - Comparison of Prediction Models for Heart Failure Related Data
SN - 978-989-758-774-0
AU - Zhang H.
PY - 2025
SP - 324
EP - 330
DO - 10.5220/0013825200004708
PB - SciTePress