The OMOP Cup featured two separate challenges designed to help predict associations between therapeutic drugs and medical outcomes (or adverse events). Researchers in many fields and from many entities, both public and private, were encouraged to take part.
OMOP provided a data set of hypothetical records for competitors to use in creating their analysis methods. The first challenge rewarded best overall performance, while the second looked at performance as data accumulated. Entries were scored on how accurately they distinguished between ‘true’ drug-event relationships and ‘negative’ controls.
Individuals and teams from around the world — 69 in all — participated in the challenges, with 21 beating OMOP’s own internal benchmarks. The international winners actually had the best methods but were not eligible for the cash prizes.
The US$10,000 prize for Challenge 1 went to David Vogel of Data Mining Solutions. Martijn Schuemie of Erasmus University in the Netherlands developed the method for this challenge and Eric Gottschalk of Activision Blizzard was also a member of the team.
A University of Iowa health informatics team comprising Lian Duan, Mohammad Khoshneshin, Si-Chi Chin and Nick Street took the $5,000 prize for Challenge 2. Vladimir Nikulin of University of Queensland, Australia developed the method for this challenge.
‘The competitors applied an extraordinarily diverse set of technical approaches, and many of their novel ideas may well represent important new directions for methods research in this area,’ said David Madigan, Professor of Statistics at Columbia University and an OMOP investigator.
To maintain momentum for the complex work, OMOP also awarded prizes for early progress. Those whose results were promising will be invited to participate on the OMOP methods development team to implement and test their methods.
OMOP is a public-private partnership created to determine whether existing healthcare data – such as electronic health records or insurance claims –can be used to identify drug risks. The ability to use healthcare data sources and effective statistical tools to analyse them to solve drug safety concerns has been lacking and the OMOP Cup seeks to fill that gap.
OMOP is a two-year project funded through, and managed by, the Foundation for the National Institutes of Health.
