Scholarlink

This study presents FlowMI-HybridNet, a comprehensive model for predicting post-myocardial infarction (MI) complications by integrating advanced data handling with a custom neural network. By analyzing data from 246 MI patients, we effectively combined flow cytometry based white blood cell data with clinical information to capture immune interactions. Our preprocessing pipeline included feature extraction through statistical summarization and data balancing techniques, transforming high-dimensional flow cytometry data into a structured format ready for modeling. After handling the data, we built the FlowMI-HybridNet model, a neural network architecture designed to capture nuanced interactions between immune cell populations and clinical outcomes. The model achieved an F1-score of 88% for patients without complications and 57% for those with complications, demonstrating the effectiveness of our approach. These findings highlight FlowMI-HybridNet’s potential to support early intervention strategies based on immune profiling.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

FlowMI-HybridNet: Integrating Data Handling and Neural Network for Enhanced Post-Myocardial Infarction Complication Prediction

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Nada Al-Dausari,
  • Frans Coenen,
  • Anh Nguyen,
  • Eduard Shantsila

摘要

This study presents FlowMI-HybridNet, a comprehensive model for predicting post-myocardial infarction (MI) complications by integrating advanced data handling with a custom neural network. By analyzing data from 246 MI patients, we effectively combined flow cytometry based white blood cell data with clinical information to capture immune interactions. Our preprocessing pipeline included feature extraction through statistical summarization and data balancing techniques, transforming high-dimensional flow cytometry data into a structured format ready for modeling. After handling the data, we built the FlowMI-HybridNet model, a neural network architecture designed to capture nuanced interactions between immune cell populations and clinical outcomes. The model achieved an F1-score of 88% for patients without complications and 57% for those with complications, demonstrating the effectiveness of our approach. These findings highlight FlowMI-HybridNet’s potential to support early intervention strategies based on immune profiling.