ADVANCED PREDICTION OF CERVICAL CANCER RISK THROUGH HYBRID SUPPORT VECTOR MACHINES AND DEEP LEARNING TECHNIQUES
International Research Journal of Modernization in Engineering Technology and Science · 2024
Cervical cancer remains a significant health challenge globally, particularly in low-and middle-income countries, where access to preventive care is limited.Early detection and intervention are crucial for reducing mortality rates associated with this disease.This paper proposes an advanced predictive model for cervical cancer risk utilizing a hybrid approach that combines Support Vector Machines (SVM) and deep learning techniques.The primary objective is to enhance prediction accuracy and reliability by using the strengths of both methodologies.The dataset used in this research, sourced from Kaggle, includes a comprehensive array of clinical, demographic, and HPV-related features The hybrid model architecture integrates a neural network for feature extraction with an SVM for classification.The neural network, trained on the dataset, captures complex patterns in the data, providing deep features that serve as inputs for the SVM classifier.The SVM, known for its robustness in high-dimensional spaces, utilizes these features to classify cervical cancer risk with high precision.Experimental results demonstrate the hybrid model's superior performance, achieving an accuracy of 94.01%, which is higher than both traditional SVM and standalone deep learning models.However, precision (0.50) and recall (0.30) indicate room for improvement in correctly identifying positive cases.The ROC curve and confusion matrix further illustrate the model's efficacy and areas needing enhancement.This paper highlights the potential of hybrid models in medical diagnostics, particularly for cervical cancer risk prediction.The integration of deep learning and SVM offers a scalable, accurate, and reliable solution, with significant implications for early detection and preventive healthcare.Future research should focus on improving precision and recall, validating the model with real-world clinical data, and exploring its applicability to other medical condition.