The lay abstract featured today (for What they forgot to tell you about machine learning with an application to pharmaceutical manufacturing by Kjell Johnson and Max Kuhn) is from Pharmaceutical Statistics with the full article now available to read here.

Johnson K, Kuhn M. What they forgot to tell you about machine learning with an application to pharmaceutical manufacturing. Pharmaceutical Statistics. 2024; 1–20. doi:10.1002/pst.2366

Abstract

Machine learning models have been incorporated into most areas of human existence.  These models play an important role in the transportation, E-commerce, marketing, agriculture, and energy industries, to name a few.  The ideal machine learning model will produce an accurate prediction that can be used to decide such as if a vehicle should stop, if a customer would be a likely consumer of a product, or the survival time for a patient with a particular disease.  While there are many resources that describe machine learning models, there are few that explain how to develop a robust model that extracts the highest possible performance from the available data.  This tutorial will describe pitfalls and best practices for developing and validating predictive models with a specific application to monitoring a pharmaceutical manufacturing process.  Best practices include grasping an understanding of the data prior to the model building process, splitting the data, pre-processing the predictors, selecting the types of models to build, and evaluating final models.  The pitfalls and best practices will be highlighted to call attention to specific points that are not generally discussed in other resources and that can be applied when building a machine learning model for any type of application.