Development and validation of the Sriya Expert Index Plus (SXI++) framework for heart disease prediction and risk reduction through clinical decision support
Journal of Medical Artificial Intelligence · 2026-02 · 9 · 11-11
DOI: 10.21037/jmai-25-47 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Medical Artificial Intelligence, 2026-02. Verfasst von Reeshabh Kumar und 3 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref).
Abstract (Originalsprache)
Background
Heart disease remains a leading global cause of mortality, creating a critical need for early and accurate diagnosis. Challenges such as limited access to diagnostic tools and healthcare disparities delay interventions and increase socioeconomic burdens. While machine learning (ML) offers potential solutions, many existing models lack adaptability and interpretability. In this study, we propose the Sriya Expert Index Plus (SXI++) framework, a novel model development and validation approach for heart disease prediction that integrates scoring mechanisms with interpretable ML.
Methods
This study utilized two datasets: the Cleveland Heart Disease dataset with 303 records and 14 clinical features, and the Heart Failure Clinical Records dataset with 299 records and 13 features, sourced from the UC Irvine Machine Learning Repository. Missing data were addressed using mean and mode imputation, and datasets were split into the training, validation, and test sets (70/20/10). The SXI++ algorithm was developed using iterative calibration, dynamic feature weighting, and a proprietary deep neural network. Model development included normalization, bivariate correlation analysis, least absolute shrinkage and selection operator (LASSO) regression, and composite weight calculation using algorithms such as XGBoost and principal component analysis. Performance was evaluated using accuracy, precision (positive predictive value), recall (negative predictive value), specificity, and area under the curve (AUC). Decision trees provided interpretable pathways for risk prediction and intervention strategies, while correlation analysis assessed the relationship between SXI scores and heart disease diagnoses.
Results
The SXI++ algorithm demonstrates superior predictive capabilities. On the Cleveland Heart Disease dataset, the algorithm achieved an accuracy of 98.33% [95% confidence interval (CI): 91.2–99.9%], a precision of 99% (95% CI: 94.5–100%), and an AUC of 0.98 (95% CI: 0.95–1.00). On the second Heart Failure dataset, the algorithm achieved an accuracy of 98% (95% CI: 90.1–99.8%), a precision of 97% (95% CI: 89.7–99.7%), and an AUC of 0.989 (95% CI: 0.96–1.00). Strong positive correlations (R2=0.98) between SXI scores and heart disease diagnoses further highlight the reliability of the model as a predictive tool. Decision tree analyses identified critical clinical pathways influencing outcomes and proposed targeted interventions. The SXI++ framework demonstrated exceptional efficacy in heart disease prediction, achieving near 99% accuracy and outstanding precision. This level of performance is enabled by SXI++’s novel integration of statistical formulae, dynamic feature weights derived from 5–7 ML algorithms, and an advanced deep neural network incorporating iterative calibration for reinforced learning.
Conclusions
The SXI++ framework provides a robust, interpretable, and high-performing solution for heart disease prediction. By combining composite ML scoring with a deep learning engine, it offers clinicians evidence-based insights to support early diagnosis and targeted intervention. Our results have been verified on two independent datasets.
Zitieren
APA
Kumar, R., Yadav, P., Kilambi, S., & Banavar, M. (2026). Development and validation of the Sriya Expert Index Plus (SXI++) framework for heart disease prediction and risk reduction through clinical decision support. Journal of Medical Artificial Intelligence, 9, 11-11. https://doi.org/10.21037/jmai-25-47
Vancouver
Kumar R, Yadav P, Kilambi S, Banavar M. Development and validation of the Sriya Expert Index Plus (SXI++) framework for heart disease prediction and risk reduction through clinical decision support. Journal of Medical Artificial Intelligence. 2026;9:11-11. doi:10.21037/jmai-25-47
BibTeX
@article{Kumar2026,
author = {Kumar, Reeshabh and Yadav, Prashant and Kilambi, Srinivas and Banavar, Mahesh},
title = {Development and validation of the Sriya Expert Index Plus (SXI++) framework for heart disease prediction and risk reduction through clinical decision support},
journal = {Journal of Medical Artificial Intelligence},
year = {2026},
volume = {9},
pages = {11--11},
doi = {10.21037/jmai-25-47}
}
EUVIMED