-
When I teach ML to clinicians, I emphasize that they shouldn't throw their knowledge out the window in favor of variable selection methods. Their knowledge is extremely valuable input to the model; it can be incorporated in many ways.On twitter.com
♻️ 7 Retweets ❤️ 25 Favorites Mood +2 🙂
-
The first is the form of Bayesian Priors - encoding their beliefs as a starting point to analysis. However, the underused way to incorporate knowledge is through feature engineering; knowing how something was measured helps you understand if it's useful.Permalink On twitter.com
-
I try to encourage this kind of collaboration, but oftentimes I work with people who only want an answer. I feel like such collaboration is pretty shallow; if the subject matter expert is engaged, the analysis is much more meaningful.Permalink On twitter.com
-
It's encouraging these sorts of relationships with your collaborators that are part of analyst's soft skills toolkit. I'd love to hear how others are teaching this.Permalink On twitter.com
-
We did this in our Analytics course by pairing clinicians with bioinformaticists for the final project; but also by cultivating an atmosphere of psychological safety.Permalink On twitter.com