Abstract Remaining Useful Life (RUL) prediction is central to maintenance planning in the manufacturing sector. It refers to the study of when some specified equipment or system is likely to fail, based on its current operating characteristics and therefore allows engineers to schedule maintenance, avoid unplanned disruptions and streamline operations, thereby reducing costs and enhancing efficiency. In this paper, we present several data-driven applications for system health monitoring with a focus on oilfield applications, such as artificial lift systems. A holistic and reliably predictive view of RUL requires addressing the effect of various factors on system degradation, such as environment variables and the prior history of operation, both of which can run into thousands of variables with often confounding effects. We describe one such approach for RUL prediction that can potentially highlight causative factors for equipment failure and degradation, that is known as survival analyses. These approaches are paired with workflows that assess the effect of uncertainty on the predictions of equipment health and RUL. We demonstrate the application of survival analyses with a presentation of the prognostic or causative factors driving system degradation. Our applications specifically include field-wide RUL analyses for artificial lift systems, including ESPs. For problems such as these, model-based RUL prediction becomes challenging simply because the formulation of a system model with the complexities in the underlying physics becomes intractable. This is where appropriate feature engineering is critical. Our work presents an argument for increased adoption of RUL prediction methods in the petroleum industry. Failures and unwarranted downtime are commonplace, and the methods presented here have the advantage of being employed by even smaller, regional operators and service companies. To our knowledge, this is the first discussion of Remaining Useful Life prediction methods combined with feature engineering.