← Back to Publications List

An Advanced Multi-Input LSTM Framework with Attention for Predicting the Risk Level of Cardiovascular Disease

Students & Supervisors

Student Authors
Arnob Aich Anurag
Bachelor of Science in Computer Science & Engineering, FST
Jafir Islam Siam
Bachelor of Science in Computer Science & Engineering, FST
Susanta Roy Emon
Bachelor of Science in Computer Science & Engineering, FST
Nizhum Biswas Akash
Bachelor of Science in Computer Science & Engineering, FST
Supervisors
Dr. Mohammad Saef Ullah Miah
Associate Professor, Faculty, FST

Abstract

Cardiovascular disease (CVD) continues to be the leading cause of mortality globally. There is a need for accurate and clinically interpretable predictive systems for CVD. In this paper, we propose a multi-input Long Short-Term Memory (LSTM) model with an atten- tion mechanism for predicting CVD, enhanced with uncertainty quan- tification via Monte Carlo Dropout and Bayesian-inspired techniques. To bridge predictive modeling with patient care, we further introduce a digital twin simulation for patient trajectory forecasting. The system in- tegrates explainability tools, including attention heatmaps, SHAP, and LIME, alongside calibration analysis through reliability diagrams and Expected Calibration Error (ECE). Experimental results demonstrate strong predictive performance (AUC 0.77–0.81), reliable uncertainty es- timates, and interpretable outputs, supporting its potential for clinical decision support

Keywords

Cardiovascular Disease Prediction Long Short-Term Mem- ory (LSTM) Multi-Input Deep Learning Attention Mechanism Un- certainty Quantification Monte Carlo Dropout Explainable AI SHAP LIME Digital Twin Simulation Clinical Decision Support Risk Level Classification

Publication Details

  • Type of Publication:
  • Conference Name: International Conference on Intelligent Data Analysis and Applications (IDAA 2025)
  • Date of Conference: 12/12/2025 - 12/12/2025
  • Venue: Daffodil International University (Daffodil Smart City), Dhaka, Bangladesh.
  • Organizer: Department of Computer Science and Engineering (CSE), Daffodil International University (DIU).
View on Publisher