An Interpretable ML Framework for Predicting Structural Heart Disease from ECG Using LLM-Powered Insights
2025
This study developed an interpretable machine learning framework for structural heart disease prediction using 100,000 ECG records. The best-performing model achieved a ROC-AUC of 0.9874, while an LLM-based module translated predictions and feature importance into structured clinical recommendations, improving interpretability and supporting evidence-based decision-making.
Type: Preprint
Year: 2025
Title: An Interpretable ML Framework for Predicting Structural Heart Disease from ECG Using LLM-Powered Insights
Authors: Shakil Ahamed Riaz, Fuad Hasan, Sk. Roushan Khalid, Md. Shohel Arman
Summary: This study developed an interpretable machine learning framework for structural heart disease prediction using 100,000 ECG records. The best-performing model achieved a ROC-AUC of 0.9874, while an LLM-based module translated predictions and feature importance into structured clinical recommendations, improving interpretability and supporting evidence-based decision-making.
Bibtex: @article{Riaz2025, title = {An Interpretable ML Framework for Predicting Structural Heart Disease from ECG Using LLM-Powered Insights}, url = {http://dx.doi.org/10.2139/ssrn.5875405}, DOI = {10.2139/ssrn.5875405}, publisher = {Elsevier BV}, author = {Riaz, Shakil Ahamed and Hasan, Fuad and Khalid, Sk. Roushan and Arman, Md Shohel}, year = {2025} }
Paper Url: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5875405
Doi: https://doi.org/10.2139/ssrn.5875405
Featured: false