Reproducibility as a Stepping Stone to Intelligent Assistants for Data Analysis: Recreating a Study of Physical Activity, Sleep, and Work Shift in Nurses

Abstract

This paper presents our work in reproducing a published study with a framework and implemented system that are intended to support automation of future analyses. This automation will allow us to develop intelligent systems to assist non-experts in getting useful information out of their own data. Our framework takes high-level questions, automatically elaborates them, and run the necessary analyses on the data. Our TGN10+ system implements that framework and was used to reproduce an analysis of behavioral, psychological, and physiological data of healthcare workers to uncover differences in subjects that can help improve their work conditions. TGN10+ can also be used to update the results when new data becomes available, and could be used to analyze new data using the same method. Our ultimate goal is to develop intelligent systems that will empower healthcare workers to analyze their own data, and ultimately anyone who wants to understand their own health and wellbeing.