Prediction of Early Cancer using Novel Improved Elastic NET Regression Model-based Optimization Framework

Improving survival rates as well as enabling timely therapies depend on early cancer prediction. Diagnostic accuracy has increased due to advancements in radiomics, Deep Learning (DL), as well as Machine Learning (ML); yet, significant challenges remain. Reliable forecasts are hampered by high false positives, limited datasets, as well as difficulties with feature selection. Furthermore, prediction is made more difficult by the variety of cancers as well as the evolving biomarkers. Refined prediction methods, advanced feature engineering, as well as real-time data processing are required to address these issues and provide timely, accurate, as well as cost-effective cancer prediction. Hence, this research performs the prediction of early cancer with the help of innovation ML-based optimization methodology. The dataset is first collected from standard benchmark sources like Kaggle. The pre-processing of this gathered data is accomplished using the L2 normalization approach. From these pre-processed data, the features are extracted by the maximum Relevance and Minimum Redundancy (mRMR) technique. The final prediction process of the early cancer is accomplished by the novel Improved Elastic NET (IELNET) regression model, where the parameter tweaking of ELNET is done by the Flower Fertilization Optimization (FFO) algorithm with the intention of error minimization as the fitness function. The results associated with the method can help the clinical diagnosis as well as treatment of cancers and assist doctors in making preliminary evaluations related to the patient's condition. The proposed IELNET-FFO for the early cancer prediction model is 7.46% and 24.45% better than the other methods in terms of prediction accuracy and Mean Square Error (MSE) respectively.