BROADCASTS.com

The best thing about personalized medicine is that it\u2019s obviously right. The worst thing is we mostly have no idea how to do it. We know that different people respond to different treatments. But outside a few special cases like cancer, we don\u2019t know how to predict which treatment will work for which person. Some psychiatric researchers claim they can do this at a high level;\xa0I think they\u2019re wrong. For most treatments and most conditions, there\u2019s no way to figure out whether a given sometimes-effective treatment will work on a given individual besides trying it and seeing.

This suggests that some chronic conditions might do best with a model centered around a\xa0controlled process\xa0of guess-and-check. When it\u2019s safe and possible, we should be maximizing throughput \u2013 finding out how to test as many medications as we can in the short time before we exhaust our patients\u2019 patience, and how to best assess the effects of each. The process of treating each individual should mirror the process of medicine in general, balancing the need to run controlled trials and gather more evidence with the need to move quickly.

I don\u2019t know how seriously to take this idea, but I would like to try it.