Authors : Vijayashree; Jennifer Mary S.; Dr. Girish Kumar D.

Volume/Issue : Volume 11 - 2026, Issue 4 - April

Google Scholar : https://tinyurl.com/3s25mfv3

Scribd : https://tinyurl.com/2xf3vk2c

DOI : https://doi.org/10.38124/ijisrt/26apr1838

Note : A published paper may take 4-5 working days from the publication date to appear in PlumX Metrics, Semantic Scholar, and ResearchGate.

Abstract : This paper introduces CardioAI, a lightweight, full-stack cardiovascular risk prediction system implemented in Python and designed for rapid deployment. The application combines a calibrated logistic-regression model (scikit-learn) persisted with joblib, a Flask-based web backend, and a responsive Jinja2/Bootstrap frontend that visualizes risk trends with Chart.js. Users register as patients or doctors (role-based access managed by Flask-Login); patients can compute personalized risk probabilities from clinically relevant inputs (age, blood pressure, cholesterol, fasting blood sugar, max heart rate, exercise-induced angina, ST depression). Predictions and metadata are stored via SQLAlchemy to a MySQL backend, and an integrated “Contact Doctor” workflow opens a prefilled WhatsApp chat (wa.me) so patients can immediately reach listed clinicians. A lightweight chatbot endpoint supports simple guidance and can optionally call an OpenAI model when an API key is supplied. We train and validate the model on a reproducible synthetic dataset and apply probability calibration to improve reliability of risk scores. The system emphasizes interpretability, usability, and low operational overhead, making it suitable as a prototype teletriage tool; we discuss privacy considerations, limitations of synthetic training data, and pathways to clinical validation and secure deployment.

Keywords : Cardiovascular Risk Prediction, Flask Web Application, Logistic Regression, Calibrated Probabilities, Healthcare Informatics, Patient–Doctor Communication, WhatsApp Integration, Role-Based Access Control, Full-Stack Development, Synthetic Dataset Modeling

References :

1. A. Sharma and R. Kulkarni, “A lightweight machine learning framework for early cardiovascular risk prediction,” International Journal of Health Informatics, vol. 12, no. 3, pp. 145–153, 2023.
2. M. Deshmukh and S. Patil, “Web-based clinical decision support systems using Python and Flask,” Journal of Medical Systems Engineering, vol. 9, no. 2, pp. 88–97, 2022.
3. K. Thomas and V. Rao, “Evaluation of logistic regression models for medical risk classification,” Computational Healthcare Review, vol. 7, no. 1, pp. 34–42, 2021.
4. P. Narayanan et al., “Synthetic dataset generation for healthcare machine learning applications,” IEEE Transactions on Data Engineering in Medicine, vol. 5, no. 4, pp. 210–218, 2022.
5. L. Sen and H. Gupta, “Role-based access and data security in digital health platforms,” Health Informatics Advances, vol. 11, no. 1, pp. 59–68, 2023.
6. J. Verma and S. Dey, “Performance analysis of database-driven medical prediction systems,” IEEE Access, vol. 10, pp. 45110–45120, 2022.
7. F. Banerjee and R. Mehta, “Enhancing patient–doctor communication through integrated messaging tools,” Journal of Telemedicine Applications, vol. 14, no. 2, pp. 77–85, 2023.
8. D. George and A. Kumar, “Visualization techniques for healthcare data monitoring dashboards,” International Journal of Data Analytics in Medicine, vol. 6, no. 3, pp. 120–129, 2022.
9. S. Rao and N. Pillai, “Real-time performance evaluation of interactive web systems for healthcare,” IEEE Journal of Web Computing, vol. 13, no. 1, pp. 48–57, 2023.
10. R. Bose and M. Chatterjee, “User experience improvements in AI-based health applications: A satisfaction study,” Journal of Human-Centered Computing, vol. 8, no. 4, pp. 185–194, 2023.