The FDA has challenged developers to create algorithms that identify spikes in adverse event data automatically. The challenge is one of several posted by the regulator to encourage developers to use the data made available through the openFDA initiative and share their work on Reddit.
 |
| FDA Chief Health Informatics Officer Dr. Taha Kass-Hout--Courtesy of the FDA |
Dr. Taha Kass-Hout, the FDA's chief health informatics officer and the creator of openFDA, released details of the challenges in a blog post. The FDA has posted three challenges of varying difficulty for two of the data sets made available through openFDA: adverse events and structured product labeling. The challenge is structured so developers can start out working on relatively simple problems--such as the creation of a word cloud made up of product-label text--and move on to more demanding tasks as they get used to manipulating the data.
The tougher challenges could lead to the development of tools of real interest to researchers and the FDA itself. As well as the aforementioned algorithm for detecting spikes in adverse events, the FDA has challenged people to normalize side effect reports using other publicly available databases, such as those based on medical care claims and emergency room visits. The FDA is also challenging developers to create models that class the text on product labels as mild, moderate or severe. Building on this, the top-tier challenge asks whether the model changes when it is applied to different drug classes.
The FDA wants developers to post links to their projects to an openFDA-focused section of Reddit, a widely read bulletin board-style website. A second, consumer-oriented challenge is also in the works. The FDA expects to post details of the consumer challenge before the end of May.
- read the blog post