Computational Science and Data Science: Process and Strategy

Meridian Line Systems can tie your scientific research into planned product delivery.

If your digital health system implements new science – specifically where it is implemented in software – then Meridian Line Systems can work with you to ensure that you have a robust chain of evidence required to support the claims of effect that you want to make. Science implemented in software includes signal processing algorithm development, computer vision, and Machine Learning. Progress in each of these areas is likely to progress using a scientific rather than engineering approach.

It is important that a science roadmap – that feeds into the product roadmap – is created and delivered (to run alongside a product roadmap) while ensuring that the scientific method is adhered to.

Meridian Line Systems can ensure that scientific progress follows a planned roadmap within the quality system.

While a classical agile approach is not generally appropriate for scientific activity, sprint-based experiment planning and execution mean that the more open-ended aspect of scientific exploration can be controlled so as to deliver commercial benefit within clear timeframes and within specified budgetary constraints.

Central to the scientific method is the concept of repeatability. The ability to regenerate results should be straightforward in computational and data science experiments but in the rush to achieve results good data management and record keeping is sometimes omitted. This turns what should be a steady scientific advance into a sporadically documented record of past achievements where every attempt to remember what happened when and why feels like an exercise in archaeology. This is especially true of Machine Learning where, in addition to documentation, multiple data sets with complex sequences of data manipulation are needed to recreate experimental results.

Meridian Line Systems can work with you to ensure that the appropriate record keeping and data management processes are in place ensuring easy experiment repeatability.