A risky clinical trial design, and attacks on machine learning

Charles Piller, an investigative journalist for Science, talks with host Sarah Crespi about a risky trial of vitamin D in asthmatic children that has caused a lot of concern among ethicists. They also discuss how the vitamin D trial connects with a possibly dangerous push to compare new treatments with placebos instead of standard-of-care treatments in clinical trials. *Note of correction: At 9:10 in the Vitamin D trial segment, the host says it is unknown if bone fractures appeared in placebo or treatment groups. In fact, while this information was not disclosed in a peer-reviewed publication, this is known through details submitted to Congress—please see the full story for details.  Next, Birhanu Eshete, professor of computer and information science at the University of Michigan, Dearborn, talks with producer Joel Goldberg about the risks of exposing machine learning algorithms online—risks such as the reverse engineering of training data to access proprietary information or even patient data. This week’s episode was produced with help from Podigy. About the Science Podcast