Combination of Binary Particle Swarm Optimization (BPSO) and Multilayer Perceptron (MLP) for Survival Prediction of Heart Failure Patients

Jurnal Infotel Pub Date : 2024-02-05 DOI:10.20895/infotel.v16i1.974

S. Sutikno

引用次数: 0

Abstract

Heart failure is a dangerous condition in which the heart cannot pump blood effectively and can lead to death. To improve this treatment, it needs methods to predict patient survival. This paper proposed combining wrapping features, namely Binary particle swarm optimization (BPSO) and a multilayer perceptron (MLP) classifier called BPSO-MLP. BPSO is used to determine the most relevant feature subset, and MLP is used to calculate its fitness. The experiment used a public dataset containing the medical records of 299 heart failure patients. This dataset comprises 13 features: age, anemia, high blood pressure, creatinine phosphokinase (CPK), diabetes, ejection fraction, platelets, gender, serum creatinine, serum sodium, smoking, time, and death events. The experiment results showed that the proposed method could produce an accuracy of up to 91.11%. The proposed method can increase accuracy by 8.89% compared to MLP (without BPSO). The addition of this wrapping feature has a significant influence on the accuracy results.

查看原文本刊更多论文

二元粒子群优化（BPSO）与多层感知器（MLP）相结合预测心力衰竭患者的存活率

心力衰竭是心脏无法有效泵血的一种危险情况，可导致死亡。为了改善这种治疗方法，需要有预测患者存活率的方法。本文建议将二元粒子群优化（BPSO）和多层感知器（MLP）分类器（BPSO-MLP）结合起来。BPSO 用于确定最相关的特征子集，MLP 用于计算其适应度。实验使用了一个公共数据集，其中包含 299 名心衰患者的医疗记录。该数据集包含 13 个特征：年龄、贫血、高血压、肌酐磷酸激酶（CPK）、糖尿病、射血分数、血小板、性别、血清肌酐、血清钠、吸烟、时间和死亡事件。实验结果表明，拟议方法的准确率高达 91.11%。与 MLP（不含 BPSO）相比，所提出的方法可将准确率提高 8.89%。该包装特征的添加对准确率结果有显著影响。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Jurnal Infotel

自引率

0.00%

发文量

审稿时长

6 weeks