I just read the article Data Science in Radiology: A Path Forward. It's a reassuring read. It makes my choice to learn programming, statistics and data science seem all the more relevant.

The author argues that AI-methods could radically change clinical radiology. However, radiology researchers, journal editors and peer reviewers often don't know their data science. This unfortunately leads to publications of bad science, which hurts credibility and spreads bad practices among researchers.

The author further argues that it's important to create a culture where we (1) learn more data science, (2) share our data and code openly and (3) support reanalyses that identify invalid findings. This would promote better science and ultimately better outcomes for the patients we serve.