Extracting Information from Awkward Datasets (medical)

Algorithmically evaluating human-factors data necessitates methods able to accommodate an awkwardly diverse set of data types. We'll use Steph's "ML Pipeline" to structure a discussion around how my team and I have integrated algorithms into assessing the development of Rapid Sequence Induction and Intubation (RSII) skills in anesthesia residents. Since this is a current study, I'll be able to detail our decision-making and provide "fresh" code and data for examples. Your comments and questions are welcome!

I will show how I cleaned my data, and methods I've used to interpret and reinforce it. We don't have an automated pipeline yet, but I will bring examples ("sketches" may be more accurate) of what we're thinking about.

Jason is a researcher at the University of Minnesota SimPORTAL, where he develops the human factors data capture platform vision and provides analysis services in a polyglot development environment using functional programming techniques.

https://www.simportal.umn.edu