Hacker News Reader

I'm a machine learning researcher, currently working on robust models (i.e. models that handle outliers properly). More traditional statistics has good techniques for making models robust, but current machine learning techniques do not.

Do you have any thoughts on how improvements in these sorts of machine learning algorithms could make clinical trials more fair? I ask because you specifically mention that HDP uses AI in the article.

View on HN

Actually the AI portion of our product is mostly for making clinical trial recruitment more efficient, although we consciously add demographic tweaks to our algorithms to help identify specific patient populations that many researchers simply don't think about.

What we are also working on to make things more fair is helping open critical clinical trials in areas where there is a higher density of minority populations, as well as using demographic data in identifying where minority patients exist that match up to the latest clinical trials to prove to pharma companies that opening them in those areas is not only a moral improvement, but a financial one as well. (Disclaimer we do all of this de-identified).

Would definitely be great to have a larger discussion on this topic!

Reply on news.ycombinator.com