For a simple and accurate prediction of the Remaining Useful Life (RUL) of a component/system, degradation-based algorithms, deployed by data-driven prognostic model, attempt to track a sensed or preprocessed feature called prognostic feature, highly correlated with fault growth. This feature should reflect the fault evolution through the entire component/system life, i.e. having a monotonic trend shape. Extracted features usually show undesirable behaviors such as fluctuation, non-monotonicity and abrupt increase at the end of the component lifecycle which hampers the accurate prediction of the RUL. We must, therefore, be addressed to the identification of new prognostic features having an obvious monotonic trend shape to enhance the prediction of the RUL. In this context, this paper attempts to address this issue by further preprocessing the extracted features in a way that the identified prognostic feature results in a smoothed and trended shape. The qualities of the identified feature are evaluated by a set of established and proposed suitability metrics. Datasets from bearings run-to-failure experiments provided by FEMTO-ST Institute -IEEE PHM 2012 challenge-were used to validate our approach. A mean percentage error of 12.18% was achieved indicating that the model worked accurately and reliably on every tested bearing.