October 20, 2009, 11:00 am

TB or not TB: Lessons learned from a 20-year quality control problem. By Brian Rohrback, President of Informetrix, Inc. (Chemistry Colloquium Series)

If we are not doing anything else today, let's merge analytical chemistry and multivariate statistics to solve a biology problem with clinical value to sort through an epidemiological threat and its socio-economic impact. A typical Kenyon class assignment? - You have one hour; go.

Twenty years ago, the Centers for Disease Control wanted to exchange a set of biochemical tests for an instrumented method to diagnose tuberculosis. Their choice was to use HPLC to characterize the cell wall mycolic acids and use the pattern of relative concentrations of these acids to differentiate disease states and thereby speed the processing of what had become an unmanageable backlog. When the first pass on the CDC system was completed a dozen years ago, tuberculosis analysis was being done in less than three days and the error rates had been cut dramatically. Looking back after more than a decade of routine use, there are some general conclusions that impact everything from how we organize experiments to how we interpret results. In short, one sees a laboratory system functioning essentially as a process analyzer in a mission-critical application in a world where the operator has limited skill and experience, but still has to succeed 100% of the time executing a complex analysis. The TB application is filled with insight into the tasks in front of many routine quality control analyses, where the issues of deployment and maintenance are both rate-limiting and success-controlling

Location: Tomsich 101

Sponsor: Department of Chemistry

Add to Your Google Calendar