Throughout history, science has been driven by observations. However, observations alone have seldom been sufficient to provide useful predictive capabilities. Sophisticated predictive capabilities almost always require an understanding of underlying physical principles.

For example, simple observation of the Sun is sufficient to make a reliable prediction that it will rise tomorrow but a complex combination of observations coupled with a solid understanding of the underlying physics is required to predict when the next Eclipse will occur. Science has evolved by combining observations and theory to develop useful predictive capabilities.

In recent years collection of real time data from in-service situations has become increasingly common due to the availability of sensors and the rise of the Industrial Internet of Things (IIoT). In parallel, the success of Artificial Intelligence and Machine Learning in data rich applications, such as image recognition, internet shopping and speech recognition, has in some cases skewed thinking to the empirical and stretched expectations of predictive analytics. This has increased demand for wider deployment, including to safety critical systems, and this is true of many applications including structural integrity