CNN and Random Forest Convergence: A Cutting-Edge Approach to Early Ovarian Cancer Prediction

Ovarian cancer is asymptomatic in its early stages, early identification greatly improves patient outcomes, but it is still a daunting issue. This paper offers a unique method to improve the prediction accuracy of early ovarian cancer by combining Random Forest (RF) and Convolutional Neural Networks (CNN). A dataset with an 80:20 training-to-testing split, consisting of 5,200 healthy and 5,600 malignant samples, is used to train the model. The RF classifier successfully distinguishes between healthy and malignant cells, while the CNN component efficiently collects high-level information from medical pictures. For the “Healthy” class, the model achieves an accuracy of 93.80% and a recall of 95.20%, yielding an F1-score of 94.49%. The model achieves 95.10% accuracy and 93.60% recall for the “Infected” class, with an F1-score of 94.34%. The 97% total accuracy in both classes highlights how consistently the algorithm can differentiate between the two scenarios. The F1-score's macro, weighted, and micro averages all lie around 94.40%, indicating evenly distributed performance across the sample. The combination of CNN and RF approaches provides a novel approach to early identification of ovarian cancer, which may result in earlier treatments and better patient outcomes. The suggested paradigm shows a great deal of promise for clinical practice integration, giving medical practitioners an effective weapon in the battle against ovarian cancer.