Phase-based Women's Health Recommendation and Prediction System
To be published in 2025 in the Book Titled: Innovations in Healthcare Technology and Management at Bentham Books
Summary
This research explores the development of personalized health recommendation systems for women based on menstrual cycle phases. Leveraging large language models and research-backed insights, the system delivers tailored diet and exercise regimens to optimize women's health outcomes across different menstrual phases. The work addresses a critical gap in personalized healthcare technology specific to women's unique physiological needs.
Leveraging Deep Learning for Breast Tumor Diagnosis: A Comparison of CNN Architectures
Currently submitted for review
Summary
This comparative study evaluates the efficacy of various Convolutional Neural Network (CNN) architectures including custom CNN, fine-tuned VGG16, and ResNet50 for breast tumor detection using ultrasound images. Working with the BUS-BRA dataset, the research achieved classification accuracies of 85.2%, 91.6%, and 93.4% respectively. The methodology incorporated U-Net for image segmentation and SMOTE for class imbalance mitigation, significantly enhancing model generalizability across a dataset of 1875 annotated ultrasound images.
Research Interests
Applied ML in Healthcare
Developing ML models that address real-world healthcare challenges and improve existing outcomes
Computational Neuroscience
Exploring applications of Machine Learning for neuroscience domain and implementing techniques for enhanced solution
Computer Vision
Advancing image analysis techniques for medical diagnostics and automated detection systems