A Machine Learning-Based Framework for the Prediction of Cervical Cancer Risk in Women

Abstract

One of the most common types of cancer in women is cervical cancer, a disease which is the most prevalent in poor nations, with one woman dying from it every two minutes. It has a major impact on the cancer burden in all cultures and economies. Clinicians have planned to use improvements in digital imaging and machine learning to enhance cervical cancer screening in recent years. Even while most cervical infections, which generate positive tests, do not result in precancer, women who test negative are at low risk for cervical cancer over the next decade. The problem is determining which women with positive HPV test results are more likely to have precancerous alterations in their cervical cells and, as a result, should have a colposcopy to inspect the cervix and collect samples for biopsy, or who requires urgent treatment. Previous research has suggested techniques to automate the dual-stain assessment, which has significant clinical implications. The authors reviewed previous research and proposed the cancer risk prediction model using deep learning. This model initially imports dataset and libraries for data analysis and posts which data standardization and basic visualization was performed. Finally, the model was designed and trained to predict cervical cancer, and the accuracy and performance were evaluated using the Cervical Cancer dataset.