Scholarlink

This paper enforces an end-to-end AI-based disease prediction dashboard on Streamlit that combines several deep learning models for clinical diagnostics. The system is equipped with three expert-level prediction modules: a CNN-based X-ray image processor for the detection of abnormality in chest radiographs, an LSTM-based ECG signal processor for arrhythmia detection, and a fully-connected network for Electronic Health Record analysis. The platform has an encrypted authentication mechanism, in-depth data visualization features, and clinical decision support capabilities such as risk assessment calculators and automatic recommendation engines. An integrated diagnosis module specifically is unique in that it consolidates predictions from all three modalities with confidence-weighted voting in order to deliver robust patient evaluations. The application also includes model monitoring dashboards for performance tracking and audit logging for compliance purposes. Designed for clinical environments, this tool demonstrates how multiple AI models can be seamlessly integrated into a unified diagnostic workflow while maintaining interpretability through attention maps and feature importance visualizations.

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

CNN-FCN-LSTM Based Multimodal Framework for Disease Diagnosis Using Medical Imaging, EHR, and ECG Signals

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Shatrunjay Kumar,
  • Devshree Babhale,
  • Vedant Shah,
  • Prashant Kharote

摘要

This paper enforces an end-to-end AI-based disease prediction dashboard on Streamlit that combines several deep learning models for clinical diagnostics. The system is equipped with three expert-level prediction modules: a CNN-based X-ray image processor for the detection of abnormality in chest radiographs, an LSTM-based ECG signal processor for arrhythmia detection, and a fully-connected network for Electronic Health Record analysis. The platform has an encrypted authentication mechanism, in-depth data visualization features, and clinical decision support capabilities such as risk assessment calculators and automatic recommendation engines. An integrated diagnosis module specifically is unique in that it consolidates predictions from all three modalities with confidence-weighted voting in order to deliver robust patient evaluations. The application also includes model monitoring dashboards for performance tracking and audit logging for compliance purposes. Designed for clinical environments, this tool demonstrates how multiple AI models can be seamlessly integrated into a unified diagnostic workflow while maintaining interpretability through attention maps and feature importance visualizations.