Early detection is very vital for increasing the survival rate of tumor-related deaths. Lung cancer datasets are intrinsically high-dimensional, containing many redundant and irrelevant features. This makes the analysis challenging due to the high computational complexity involved, and it reduces predictive accuracy. We propose a hybrid approach called the ORIGEN-ICT Framework by leveraging the power of ICT in the most efficient feature selection and data mining. This framework combines Information Gain (IG) with Recursive Feature Elimination (RFE) to speed up the process of selecting features. IG ranks the features by measuring their relevance to the target variable; hence, it reduces the dataset to the most informative features. Then, RFE iteratively refines this subset by training predictive models, removing the least important features, and retraining the model at each step. The process results in an optimized feature set that reduces noise and enhances computational efficiency. The validation of ORIGEN-ICT on the UC Irvine Machine Learning Repository lung cancer data set showed a significant enhancement in predictive accuracy to 98% ORIGEN-ICT contributes to overcoming most of the challenges due to its efficient dimensionality reduction in the case of datasets and selecting only relevant features. This lessens computational demands while improving reliability and speed of predictive models. The proposed framework opens ways not only for the faster and earlier finding of lung cancer but also for enhanced clinical decision-making by offering more accurate and interpretable diagnostic tools.