AI and Treatment

Podcast: Download (Duration: 1:03 — 1.5MB)

Subscribe: RSS

Anchor lead: Can an artificial intelligence approach to looking at drugs and cancer find promising treatments? Elizabeth Tracey reports

Can combing through vast databases using tools such as artificial intelligence or machine learning help find new cancer treatments? That’s the hope of one recent study, but William Nelson, director of the Kimmel Cancer Center at Johns Hopkins, says the chances of success are slim.

Nelson: The best of these are the countries that have a better pharmacovigilance database, which typically are the Scandinavian countries, where they actually know who got what drug and they actually know what health conditions they got, in a larger more comprehensive way. Here we have non-interoperable patient medical records. The best and most comprehensive database we have is probably from the Centers for Medicare and Medicaid Services, but the countries with registries have done these kinds of questions and the easy answer is you’re not going to find that some random drug is going to cure everybody’s cancer.  :33

Nelson says at best such studies may find a signal for a successful therapy that will then need further study. At Johns Hopkins, I’m Elizabeth Tracey.