Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{196883,
        author = {Mayank Sharma and Vipasha Samal and Ankita Dubey and Anchal Maurya and Ms. Pooja Singh},
        title = {Cardio Scan AI: A Hybrid Machine Learning Approach for Early Prediction of Cardiac Risk with Integrated AI Agent, RAG-Based Doctor Recommendation, and Secure Multi-Role Web Portal},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {11},
        pages = {5369-5375},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=196883},
        abstract = {Heart diseases remain one of the leading causes of mortality worldwide, primarily because many cardiac conditions develop asymptomatically until they reach a critical stage. This paper presents Cardio Scan AI, an enhanced hybrid intelligent framework that integrates ensemble machine learning and deep neural networks for accurate cardiac risk prediction with a full-stack, clinically connected web platform. The system accepts eleven clinical and lifestyle parameters and outputs a probabilistic risk score mapped to one of five risk tiers. Beyond prediction, the framework introduces three principal innovations: (1) a secure multi-role authentication system using JSON Web Tokens (JWT) and Redis-backed session management, supporting distinct patient and doctor portals built in React with HTML/CSS; (2) an AI agent powered by a Large Language Model (LLM) and Retrieval-Augmented Generation (RAG) that matches patients to appropriate doctors—with doctor profiles stored as vector embeddings in Pinecone—and supports voice-based interaction through OpenAI Whisper transcription; and (3) a doctor verification portal where clinicians review AI-generated risk reports, confirm or override assessments, and add clinical notes, with patient data persisted in MongoDB and relational metadata in PostgreSQL. Experimental evaluation on 2,300 samples demonstrates 94.2% accuracy and ROC–AUC of 0.96. The proposed system bridges early cardiac detection, AI-powered specialist referral, and clinical verification within a single accessible platform.},
        keywords = {cardiac risk prediction, machine learning, XGBoost, deep neural network, RAG, LLM, Whisper, Pinecone, vector embeddings, JWT, Redis, MongoDB, PostgreSQL, React, Node.js, Express, AI agent, doctor recommendation, preventive healthcare.},
        month = {April},
        }