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{196362,
        author = {V. Bhavya Phani Sri and Mrs Y. Sirisha and A. Karthik and B. Divya},
        title = {Multi-Modal Parkinson’s Disease Detection Using CNN and XGBoost with Voice Features and Spiral/Wave Drawings},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {11},
        pages = {2846-2854},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=196362},
        abstract = {Parkinson’s disease (PD) is a progressive neurodegenerative disorder that requires early and accurate detection to improve patient outcomes. However, traditional diagnostic methods are often subjective, time- consuming, and dependent on clinical expertise. This paper proposes a novel multi-modal machine learning framework for non-invasive and real-time detection of Parkinson’s disease using voice biomarkers and hand-drawn spiral and wave images.
For voice analysis, 22 acoustic features from the UCI Parkinson’s dataset are standardized and transformed into a 6×4×1 image-like representation, enabling a lightweight Convolutional Neural Network (CNN) to extract deep features. These features are then classified using an optimized XGBoost model, with class imbalance handled through SMOTE. The proposed voice-based pipeline achieves an accuracy of 92.31% and a ROC-AUC score of 0.96, outperforming several traditional approaches.
For image analysis, spiral and wave drawings are processed using a pre-trained MobileNetV2 model to extract high- level feature embeddings, which are further classified using XGBoost. Both modalities are integrated into a user- friendly Streamlit web application that supports CSV upload, manual input, and image-based prediction, enabling accessible and real-time screening.
The proposed system is lightweight, scalable, and deployable on standard hardware, making it suitable for telemedicine and early-stage screening, especially in resource-limited environments.},
        keywords = {Parkinson’s disease, CNN, XGBoost, voice analysis, spiral drawings},
        month = {April},
        }