On Friday, May 11, Catherine Beauchemin gave a short talk at Nerd Nite Tokyo. She told the public that health science has serious issues with reproducibility of experiment results, and that the research culture incentivizes positive results rather than sound experiments. This is why you get reports in successive months about how eggs are unhealthy, then healthy. Or how a medical practice used for 20 years to reduce post-surgery infections actually makes them more likely. According to Catherine, if doctors and health scientists want to be trusted more, they need to better communicate the uncertainty of results, do fewer (but larger) experiments to produce more robust results, and accept that negative results are just as worthy of publication as positive results.

She also talked about how health science has math problems. Too many people in that field don't understand math/statistics well, and blindly trust equations even when those equations are misapplied and give incorrect results. There is a big opportunity for scientists with a background in math to work with health scientists and improve the quality of research in a field that is important for everyone. She is already starting a project along these lines -- please join her to promote the goals of iTHEMS and health sciences!

And if you want to speak at a future Nerd Nite about your own research, please contact Don Warren (donald.warren@riken.jp) for information.